TW496080B - Method and system for normalizing a plurality of signals having a shared component - Google Patents

Abstract

One aspect of the invention is a method for normalizing a plurality of signals wherein the plurality of signals have a shared component and wherein at least one of the signals has been distorted in a nonlinear way. A distortion function is determined for at least one of the signals which is proportional to the distortion of that signal relative to at least one of the remaining signals. An inverse relative distortion function is generated for the distorted signal responsive to the distortion function that was determined for that signal. The signal is normalized by applying the inverse relative distortion function that was generated for the distorted signal.

Description

496080 A7 B7 Printed by a member of the Intellectual Property Bureau of the Ministry of Economic Affairs and a Consumer Cooperative. V. Description of the Invention (1) Technical Field of the Invention The present invention relates to signal processing, and more particularly relates to the normalization of a plurality of components having a common component. Cross Reference to Related Applications This application and the one proposed by Albert Edger were archived on August 20, 1999, with the agent's document abstract number 021971.0127, and the title was "Automatically Improve Digital Method and device for image quality of the United States "US application serial number_related. This application was proposed by Albert Edger, filed on August 20, 1999, with the agent's document summary number 02 1971 · 0 104, and the title is "Improving Digital Imaging Methods and Systems "US Application Serial Number __ related. BACKGROUND OF THE INVENTION In many signal processing applications, most signals may have a common component. In a non-linear manner, one or more of these related signals may become distorted. In some applications, each signal can be distorted or undistorted in a different way, and the nature of the distortion may be unpredictable. In addition, it may be difficult to decide which related signal becomes distorted. Such distortion is not ideal. For example: in the field of image processing, whether the image is a scanned photo image, satellite image, medical image, or other types of digital (please read the precautions on the back before filling this page).

nnn I a θ ”· flu 1 I n« ϋ an I This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) 2. Description of the Invention (2) Images' may suffer from non-linear distortions. Such distortions may exist when the images are represented in an analogous manner and / or may occur when digital images are generated. For example, in scanning of photographic images During the scanning process, each of the three color channels of a typical digital image may suffer from non-linear distortion, but such non-linear distortion may be different in each color channel. Such non-linear distortion May degrade the image in a way we do n’t want. For photographic images, the images may not look so desirable. For satellite images and medical images, the information provided by the images due to such distortion May not be so useful. DESCRIPTION OF THE INVENTION One aspect of the present invention is a method for normalizing a plurality of signals having a common component. Method, wherein at least one of the signals has been distorted in a non-linear manner. A distortion function is determined for at least one of a plurality of signals, wherein the signal function is relative to at least one of the remaining signals. The distortion of a particular signal is proportional. A relatively inverse distortion function is generated for this distortion signal, where this function is a response to the distortion function determined for this signal. By applying the relative inversion generated for this signal Normalizes the signal in question. The invention has many important technical advantages. Various embodiments of the invention may not have, have one, some, or all of these advantages without departing from this Scope of the invention. The present invention allows signals with a common component to be automatically normalized to each other, where 'one of these signals (please read the precautions on the back before filling out this page). ---- Order ------ · — This paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 mm) -5- 496080 Staff Consumption of Intellectual Property Bureau, Ministry of Economic Affairs Printed by Sakusha A7 B7 V. Invention Description (3) or more has been distorted in a non-linear manner. Although this distortion may not be completely removed, the signal can more accurately represent the desired signal : No. The present invention can be advantageously applied in the scope of image processing. For example, the present invention can be used to correct the distortion in the color channel of a digital image representing a color photographic image. This color channel can be Normalize, so that the overall image is improved, and the color of the image becomes more desirable. For artificial satellite images and medical images, the present invention can be used to increase the accuracy and availability of the information provided by the images. Since the present invention can perform such normalization using a computer and / or scanner, the present invention allows automatic normalization of images. Such automatic normalization can reduce the time and effort required to achieve such normalization, while allowing normalization of a larger number of images. For example, in the case of photographic images, software such as: Photoshop can be used to manually correct non-linear distortion. The present invention allows automatic color correction of any desired image with a little time, a little cost, and the person who corrects the image does not need special knowledge. Brief description of the drawings In order to fully understand the present invention and its advantages, I refer to the following description, together with the drawings, in which: Figure 1 illustrates a block diagram of a general-purpose computer frft fgj · according to the present invention, which can be used; FIG. 2 illustrates an example of a scanner including an embodiment of the present invention; FIG. 3 illustrates a flow chart describing a method of improving a digital image according to the present invention; the paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the notes on the back before filling this page)

496080 A7 B7 V. Description of the invention (4) FIG. 4 illustrates a flowchart describing a method for measuring a grain trace related to pixel intensity (pi X e 1 intens 丨 ty) according to the present invention; FIG. 5 does not show An image divided into sections (segme π t); Figure 6 illustrates the weighting, average intensity, and particle intensity of the red channel of different digital images; Figure 7 illustrates a digital image A scatter diagram of a color channel of FIG .; FIG. 8 illustrates a line of a scatter diagram of a color channel that passes through a digital image using a curve fitting algorithm (curve fitting a 1 g0M thm). Figure 9 illustrates a scatter diagram line of red information passing through a digital image; Figure 10 illustrates a flowchart describing a method for measuring frequency-dependent particle trajectories according to the present invention; Figure 1 1 illustrates a scatter of a color channel of a digital image. Figure 9 2 illustrates a line of a scatter diagram of a color channel that passes through a digital image. The morphine particle trace on a line of pixels from a digital image with a change in pixel intensity drawn in space. Figure 14 illustrates the frequency of the particle trajectory normalized across the intensity. A graph of the frequency in a digital image; Figure 15 illustrates the frequency of the particle trajectory. For the intensity and frequency across the paper, this paper scale applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ( Please read the precautions on the back before filling this page.) -------- Order ------.------ Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7 Printed by a consumer cooperative. 5. Description of the invention (5) A graph of the frequency of day in a digital image normalized; Figure 16 illustrates the removal of high frequencies such as particle trajectories from a digital image. A flowchart of an example of a method of noise; Figure 17 illustrates a flowchart of an example of a method for removing high-frequency noise such as: particle traces from a digital image; Figure 18 illustrates a representative and The two-dimensional Fourier transform associated with the present invention (Fourier t ransform) method; FIG. 19 illustrates a flowchart illustrating the steps of an example method that can be used to normalize a plurality of signals having a common component, in which at least one of the plurality of signals is non-linearly Distort; FIG. 20 illustrates a flowchart illustrating an example of a method that can be used to improve a digital image according to the present invention; and FIG. 21 illustrates an example that can be used to connect an example illustrated in FIG. 19 or FIG. 20 Three scatter diagram examples of the method. Component comparison table 1 0: General-purpose computer 1 2: Processor 1 4: Random access memory 1 6: Read-only memory 1 8: Mouse 2 0: Keyboard 2 2: Disk drive 2 4: Printer ( Please read the precautions on the back before filling in this page) Binding ---- Ordering --------.--- Edge% · This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -8- 496080 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (6) 2 6: Display 2 8: Communication link '3 0: Scanner 3 4: Scanner 3 6: Processor 3 8 : Storage medium 4 0: Scanning hardware 4 2: Image processing software 4 4: Detailed description of the preferred embodiment of the control software By referring to Figures 1 to 15 of the drawings, such as the numbers used and the various drawings Corresponding parts, the preferred embodiments of the present invention and their advantages are most easily understood. Fig. 1 illustrates a block diagram of a general-purpose computer 10 which can be used for image improvement according to the present invention. In particular, the general-purpose computer 10 may include a part of an image improvement system, or may be used to execute an application program including image improvement software. The general-purpose computer 10 can be used to perform any of the well-known ^ 13 — 0〇3,? (: 003, 032, UN IX, MAC-OS, and Windows (Windows) operating system or other operating systems. General-purpose computer 10 includes processor 1 2. Random access memory (RA M) 1 4 , Read-only memory (R a M) 1 6, mouse 18, keyboard 20, and input / output devices such as printer 2 4, disk drive 2 2, display 26 and communication links ((Please Please read the notes on the back before filling this page) This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm> 496080 Α7 Β7 V. Description of the invention (7) (Please read the notes on the back before filling (This page) communication link) 2 8. The present invention includes: programs that can be stored in RAMI 4, ROM16, or disk drive 2 2 and can be executed by the processor 1 2. The communication link 2 8 is connected To a computer network, it can also be connected to an electric 3 tongue line, a day τ spring, a Landau or any other type of communication link. The drive 22 can include a variety of different storage media, such as: floppy disk Drive, hard drive, CD-ROM drive, or tape drive. Although this embodiment With multiple disk drives 22, we can also use a single disk drive 22 without departing from the scope of the present invention. Figure 1 provides only an example of a computer that can be used with the present invention. The invention can be used On general-purpose computers or general-purpose computers without traditional operating systems. The general consumer computer 10 printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs further includes a scanner 3 0, which can be used to scan according to the teachings of the present invention. Improved image. In this embodiment, the image improvement may be performed by software stored in the scanner 30 and executed by the scanner 30, and the result may be stored in a file containing a part of the scanner 30. A storage medium and / or in any storage device in the general-purpose computer 10. Alternatively, the software for improving the image may be stored in any storage medium associated with the general-purpose computer 10 and may be executed by the processor 12 To improve the image scanned by the scanner 30. In addition, 'image improvement can occur inherently in the scanner 30 and the general-purpose computer 10, as described, without departing from the scope of the present invention. The scanner 30 may include a film scanner or a platform scanner in any form without departing from the scope of the present invention. Figure 2 illustrates an example of a scanner 34 constructed in accordance with the present invention. The scanner 34 includes ·· Processor 3 6, Storage Media 3 8 and Scanning Hardware The paper size is applicable to Chinese National Standard (CNS) A4 (210 X 297 mm) ~ \ 〇 496080 Α7 Β7 Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs and Consumer Cooperatives The fifth, invention description (8) 40. The processor 36 controls the operation of scanning the hardware 40 by executing the control software 4 4 stored in the storage medium 38. Although for the sake of simplicity, I exemplify a single storage medium, the storage medium 38 may include multiple storage media or different types of storage media. Therefore, for example, the control software 44 can be stored in ROM memory, RAM memory or a hard disk. Scanning hardware 40 is used to convert an analog image into a digital image using a certain type of optical circuit system. / Any form of optical circuit system can be used in the scanning hardware 40 without departing from the scope of the present invention. After scanning the image by the hardware 40, the image can be improved according to the present invention using the image processing software 42 stored in the storage medium 38. Similarly, the scanned image can be stored in the storage medium 38, and the improved image can also be stored. Alternatively, the scanner 34 may be provided without any image processing software 42. Such software can instead be provided by a general-purpose computer 10 for improving an image received from the scanner 34. Therefore, a scanned image and / or an improved scanned image can be supplied to the general-purpose computer 10 by the scanner 34 through a communication port (not explicitly shown). Although the embodiment of a scanner 34 exemplified by us can be used for image improvement related to the present invention, other scanners can be used without departing from the scope of the present invention. FIG. 3 illustrates a flowchart describing a method used to improve a digital image according to an embodiment of the present invention. The steps described here can be performed using computer software, as can any of the methods described below. This software, as mentioned above, can be installed by the scanner 3 4 or the general-purpose computer 10 (please read the precautions on the back before filling in this page). ---- Order ------.--- ^ · This paper Standards are applicable to China National Standard (CNS) A4 specifications (210 X 297 meals) -11-496080 Printed by A7 B7, Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Implementation of the invention description (9). A digital image received from other than the scanner 30 can also be improved according to the present invention. In step 1 0 02, the glomerular trajectory of a digital image is measured in relation to the pixel intensity. In step 1004, the glomerular locus is measured as a function of frequency. In step 106, a correction array is calculated. The correction array is used to normalize the particle trajectories of the digital image and the intensity in step 10 08, and then, in step 10 10, the particle trajectories of the digital image to frequency are normalized. Into. This normalized particle trajectory data can be used to suppress the particle trajectory in step 1012. In addition, as shown in step 104, the pixel intensity 値 obtained by normalizing the particle trajectory can be used to balance these pixel intensities in all color channels. Some of these steps may be omitted, or other steps may be included without departing from the scope of the invention. Although the present invention can be applied to digital images produced from photographic images, it can also be used to normalize high-frequency noise, such as particle trajectories in other types of images, such as artificial satellites Imaging, medical imaging, and more. The following discussion illustrates the normalization of black and white and color images. For color images, the method described below can be used, or for black and white images, the methods described above can be used for each individual channel in the color image. Alternatively, the black and white method can be used for a subset of the channels, and the color method can be used for the remaining channels. Similar options apply to digital images that are not photographic. Measure the particle trajectory related to the strength. The size of this paper is applicable to China National Standard (CNS) A4 (210 X 297 mm) -12- ----- ^ ---- ^ ---- pack ----- --- Order --- (Please read the precautions on the back before filling out this page) 呓 · 496080 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (i〇) The invention includes measurement related to pixels A method of particle trajectory of intensity. Figure 4 shows the steps involved in this method. In this application, all calculations related to this and other methods may be performed using a general-purpose computer 10, a scanner 34, or a combination of one of these devices with other devices. In step 1 102, the digital image can be divided into several sections. Then, in step 1 104, a transformation such as a Fourier transform is performed, and then in step 1 106, shifting is performed. Divide low and medium frequency terms. Then, in step 1 108, in each color channel, various formulas can be used to calculate the particle intensity (representing the deviation of the magnitude of the pixel intensity due to the particle trajectory) of each segment. In step 1 1 10, 'to help verify the intensity of these particles, a separate formula' can be used to calculate a "weight" for each segment in each color channel. Secondly, in step 11 12, a scatter diagram can be generated by determining the weighted chirp, DC (direct current, direct voltage, which conventionally represents the average chirp of a signal), and the particle intensity of each color channel. In step 1 1 1 4 we can use a curve fitting algorithm to draw a line through the scatter diagram to show the proper particle intensity for each pixel intensity. Segment segmentation In step 1 102, the digital image is divided into data segments, and each segment corresponds to a spatial region of the digital image. Segmentation can be applied to each channel of a digital image, or a single channel in the case of a black and white image. As shown in Fig. 5, a representative of this segmentation displays a plurality of sections, such as: section 1 2 0 2. Although we can use any section (please read the precautions on the back before filling out this page) Packing ---------- Order ---------- · This paper size applies to Chinese National Standards (CNS ) A4 size (210 X 297 mm) -13-496080 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs B7_____ V. Invention description (11) Small, but the area of each section is preferably large enough to include statistics A meaningful number of pixels, such as: between 16 and 10 24 pixels, but not large enough to contain an area of many images of virtually different intensities (between 2 0 0 by 3 0 0 0 pixels An image size). In a single segment of an image that is large enough to provide enough data, but not too large, and therefore has a number of 20,000 pixels multiplied by 3,000 pixels, there are too many different Among the intensity components, a segment containing 2 5 6 (16 X 1 6 segments /) different pixels can provide a good compromise. Using this method, a single 35 mm negative (or slide transparencies) will typically form approximately 24 K non-overlapping segments. Any image size or any segment size can be used without departing from the scope of the invention. Figure 5 shows these sections as squares, however, they can be other shapes without departing from the scope of the invention. The different sections need not be the same shape and size, although making them identical may be useful in processing. In a particular embodiment, the number of segments and their shape and size may be defined by changes in the image associated with frequency, intensity, or frequency content. Also note: sections can be overlapped and panes, such as: in "Archives, US Patent Application No. _" Mixed Image Boxes by Pane "" This article has been incorporated by reference, It seems as if it was published here. Fourier Transform After segmenting this digital image segment, 'we can each pixel' on each color channel (or on a single black-and-white channel) in pixel intensity (please read the notes on the back before filling (This page)

i §1 1 flu f f II —ϋ ϋ — nn ϋ — I%, this paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -14-496080 A7 B7 Employees of Intellectual Property Bureau, Ministry of Economic Affairs Printed by the consumer cooperative V. Description of the invention (12) A Fourier transform is performed on the scale, as shown in step 1 104 in FIG. The data forming the digital image contains the intensity of each pixel in the image and is spatially arranged for each color channel. Intensity refers to the brightness of one pixel. For example: a white pixel has greater intensity than a gray or black pixel. The same spatial area of different color channels usually has different pixel intensities. The Fourier transform transforms the pixel intensity of each segment from the spatial domain to the frequency domain. One of the values calculated by the Fourier transform is the average intensity of each segment of each color channel, that is, "D C". Calculated by the Fourier transform The other is the frequency and magnitude of the vertical and horizontal frequency vectors of each frequency element (element) in each section of each color channel. Although this embodiment uses a Fourier transform, other types of transforms can also be used to transform a signal from the spatial domain to the frequency domain, such as a Hadamard transfrom. In addition, the frequency conversion space can be subdivided using the ® pass-through wave canceller. Removing low and medium frequencies Second, in step 1 106 in Figure 4, the low and medium frequency terms of each transformation can be removed from further consideration. Alternatively, these terms may be weakened or this step may be omitted without departing from the scope of the invention. In this embodiment, each segment of each color channel (or a single channel in a black-and-white image) receives this removal within the useful range 1 of a particular embodiment of the present invention. Particle strength with significantly higher frequencies is easier to measure and distinguish from image information. Because of the tendency of substantial particles on photographic film (please read the precautions on the back before filling this page) Loading n · Βϋ I ^ · · ϋ «-a— nn · 1 ΜΗ! -V-α-.% This paper size Applicable to China National Standard (CNS) A4 specification (210 X 297 mm) -15- 496080 Α7 Β7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economy There are approximately equal frequencies, but at higher frequencies, there is less image content. In order to get the best difference between image detail and particle trajectory, it is best to focus on high frequencies. In this embodiment, we use a 6 X 1 6 section 6 times emphasis: other section sizes can be used without departing from the scope of the present invention. In order to filter out the low and mid-range frequencies, we can, for example: filter out the sum of all X-frequency and y-frequency absolute absolute chirps to frequencies less than or equal to 8. Alternatively, if the frequency vector is configured as shown in FIG. 18, a suitable radius around the DC can be selected, for example: 7777 = 8 frequencies within this radius can be removed. Other radii or filtering methods can be used 'without departing from the scope of the invention. Calculating the particle trajectory Secondly, in step 1 108 in Fig. 4, we can use the cross-correlation (c r s s-c o r r e 1 a t i ο η) formula to calculate the particle intensity of each section. In the case of a black-and-white digital image, because there is only one channel, the cross-correlation formula is not used. For color and black and white images, this calculation uses the frequency vector of the digital image from the Fourier transform after removing the low and medium frequency terms. To understand these calculations, an explanation of a digital image will be useful. The total information including a section is composed of information of each pixel of each color channel. For example ... In a typical three-color digital image, one (please read the precautions on the back before filling this page)

n an-OJ0 11 n ϋ ammwm n ϋ I This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) -16-496080 Α7 Β7 V. Description of the invention (14) (Please read the note on the back first Please fill in this page again) The total information in the section) is a combination of information from the red, green and blue channels. The term “color channel” is used for convenience, and the image channel does not necessarily have to be a color channel. The present invention can be applied to an image channel of a digital image, even if such a channel does not represent a color. In this embodiment, this channel represents three color channels of a photographic image. Red information, which represents the red frequency vector, and can be represented by Equation 1. In this patent, bold text represents a complex vector, where 'the vector can be conventionally considered to have a real and imaginary component' or alternatively a size and phase component. R = R s + R II (1) In Equation 1, R represents the signal of each pixel combination recorded in one segment of the entire red channel. R s is the red information part of the segment from the image signal itself. r n is the part from "noise", where the noise can actually be considered as a particle trajectory. Similarly, the green information and blue information are: G = G s + G η (2) B = + (3) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. Assume that only r, G, and B can be measured. R u, G η and B η are calculated and estimated. Similarly, the noise component of a black and white image channel is calculated and estimated. Because the particle pattern is different in each color layer of the original film, 'I can reasonably assume that the particle trajectory in a section of a digital image is actually noise, which is not in the entire color channel. related. In other words, the strength of the noise part of the signal in a section 丨 straight at -17-This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love) Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs Consumer Cooperatives 496080 A7 ____B7__ __ 5. Description of the Invention (15) Each color channel will have a different tendency. On the other hand, the intensity of the image portion of the signal from a segment will tend to be similar in each color channel and is therefore related. Such a correlation can logically result from the principle that the brightness component of a natural image is often stronger than the color component. Therefore, the cross-correlation between colors can be used for the prediction of the ratio mixing of R $ and R η containing R. Weak cross-correlation implies that the signal is very close to zero, so

Rn = R = Rs + Rn (4) On the other hand, strong cross-correlation implies that the R S component is more dominant than R η, so

Rn «R = Rs + Rn (5) Therefore, a reasonable estimate of this noise can be changed to: Estimated R = F (cross-correlation) · R (6) In Equation 6, when the cross-correlation approaches to self Correlation, that is, ideal correlation, F (cross-correlation) approaches 0, and when cross-correlation approaches 0, it approaches 1. Therefore, as long as R is given, the cross-correlation between image channels can be used to arrive at a modified estimate of R ". However, when a strong signal exists, 'the estimate' may be less reliable. Therefore, The second use of cross-correlation is to get an estimate of this reliability 値 'and thus a “weighting” of how much control must be taken into account when averaging with other sectors. Measurement of cross-correlation between information of color channels such as: red and green channels, so it can be expressed algebraically: i? G = (R · + RJ (G · + GJ = R, G · + R〆 ·· + u Three R, G · ⑺ I --- IIJIII-IIII (Please read the precautions on the back before filling this page)

· III This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -18- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economy 496080 A7 ___B7_ _ V. Description of the invention (16) In the formula ' RG indicates cross-correlation between information from a segment of red and green channels. This is after removing the low and mid range frequencies, and for each non-DC frequency term, one frequency at a time, the segment frequency of the red channel is multiplied by the complex frequency product of the segment frequency of the green channel. carry out. For example: in the red channel, the frequency vector at the frequency position 1 on the frequency vector_ in the green channel. In a separate calculation, the frequency vector at frequency position 2 in the red channel is multiplied by the frequency vector at position 2 in the green channel, and so on. It should be noted that there are two vectors in which the vectors are multiplied, that is, the “point product”, so '' represents that these two vectors contain the correlation in the direction of the vector. The products of these individual calculations are summed. Equation 7 also shows that, over a large area of the digital image, there is no correlation between R R n G s, G s R n, and R n G n. Their product is sometimes randomly positive and sometimes randomly negative, so it has a small effect on the final sum. Their average chirp approaches zero. Finally, the formula is shown: one frequency and one frequency, the sum of each segment, the relevant red and green information 槪 is approximately equal to the signal of the relevant red and green images 値. Similarly, B R = BsRs and BG = BiiGs. The cross-correlation between the color channels of each zone, one frequency position and one frequency position, can be calculated using Equation 8-10. BR = RB = Σ R · B Entire Section (8 > BG = GB = Σ G · R Entire Section (9) BG = GB = Σ GB Entire Section (10) Therefore, for a particular section, In other words, the cross-correlation between the two color channels includes the sum of the dot products of the frequency domain vectors at the corresponding frequencies (after removal · · —M ---- 装 -------- Order- -----; --- (Please read the general Italian matter on the back before filling in this page) This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -19- 496080 A7 ______B7___ 5 (17) After the low and medium frequency terms) (Please read the notes on the back before filling this page) Second, the particle intensity can be measured for each segment in each color channel. Particle intensity Is the number of traces left from the particles in the original film. The segment with a large number of particle traces has high particle intensity. The red particle intensity (R gs) of a segment can be calculated by the following formula: ^ = l | RQ | = SVRn * Ra (⑴, that is, the intensity of the red particle in a section is equal to the sum of its absolute noise 値 丨, which in turn equals The sum of the square roots of the color noise and the square root can be estimated based on experience: A · Rn = λΙ [Κ (12) Note that the square root is used to reduce the impact of occasional strong noise to The lowest. This estimate can also be used for other color channels, or a single channel of a black and white image. Because R „= R-R s, the formula for the particle intensity of the red channel can be changed as follows: You -Σλ / λλ: = (13) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Note: For a known section of R, when R s increases and R η becomes small, 値 of “R — R s” approaches 〇 Therefore, R gs approaches 0. Conversely, when R u increases and R s becomes small, the R of “R — R〆 ′” becomes larger, and thus R ss becomes larger. The formula can be written in the same way: The paper size is applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) -20-496080

V. Description of the invention (18)

Gg ^ = lVG * Ga = lVG * (G ~ Gs) (14) Bgs = Σλ / Β · (Β-Β5) (15) Printed by the Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs

A similar formula can be used for a single channel of a black and white image. The image signal of each color channel exists in the image signal of another color channel. For example, some of the red image signals (R s) exist in the green image signals (G s) and the blue image signals (B s). Some of the green image signals (G s) exist in the red image signal (R s) and ^ color image signals (B s), and some of the blue image signals (B s) exist in the red image signal ( R s) and a constant (K) 〇 of 0 or less in the green image signal (G s) are useful for distinguishing the intensity of particles' as described below: KR. Is the number of red image signals in the green image signal. K. R is the number of red video signals in the green video signal ° K R B is the number of red video signals in the blue video signal ° K B is the number of green image signals in the blue image signal ° K B R is the number of blue image signals in the red image signal ° K B It is the number of blue image signals in the green image signal. According to empirical research, based on these constants, the frequency of the input signal and noise data can be used to calculate the estimated image signal of each color. In-implementation, the red, green and blue signals are: Gs- / 2K〇s ^^ y2KCRR (16) (17) "Jing first read the precautions on the back before filling out this page) Μ --- ^ --- ----- ^ ------: ---. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -21-496080 A7 V. Description of the invention (19) Bs ^ y2KBRR, / 2K3CG (18): In the formulas 16—18, R is a set of red frequency vectors of a section, G is a set of green frequency vectors of a section, and B is a blue frequency of a section The set of vectors. These chirps have been obtained from the Fourier transform. As before, the low and medium frequencies have been removed. Solving for the red particle intensity (R gs) ', the formula is: (19) (Please read the back Please fill in this page again for the intensity of green and blue particles (R gs) ~. The group estimate can be used for the constant (K) 値 by using the previously calculated cross-correlation (see the formula can be estimated as follows: 10

Solve. For example: K Pack -------- Order ------.--- Side% ·

RG.RB

K

Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs

Gs

BG

jGB.GR V BR

BG-GB (22)

Solve in the same way, the other K 値 are: KGR KBR G5 Jr (23) GR Ίδ (24) BG: ^ GR (25) This paper size applies the Chinese National Standard (CNS) A4 specification (210 χ 297 mm) 496080 A7 B7 V. Description of the invention (20) KGB: GR z -— BR (26) KBC. BR —__ __ GR (27) We use the obtained 得出 and use to remove the low and medium frequencies (in these formulas R, G, and B), from the frequency vector obtained from the Fourier transform, the intensity of the particles in each color in each segment can be calculated. For example, the present invention estimates the intensity of the red particles through the following series of equations and estimates: _ Rgs = z | R „| = Ia / r ~ r7 (ii) (12) ---- ·· --- J --- -· -Pack—— (Please read the precautions on the back before filling this page)

RgsRn = R-R, (13) ^ = Za / R * (R-rs) (13),

B (16) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs

K

RG K0 BRΊδ GRJr This paper size is applicable to the Chinese National Standard (CNS) A4 specification (21〇 χ 297 S) -23- (19) (22) (23) 496080 A7 B7 V. Description of the invention (21)

/ 2BG / 2 BR

M = SR * B

5G = ZG -B

GR = Σ G · R (28) (8) (10) (9) (Please read the notes on the back before filling this page)

Rn

SjR ·

The formula for IRB (29) to estimate the intensity of green particles and blue particles is

Ggs =

G

ZRB

IR B (30) ^ = Σ · B. Printing of clothing by employees' cooperatives in the Intellectual Property Bureau of the Ministry of Economic Affairs (31) In each case, sum each frequency in the space frequency domain. For black and white images, the estimation of the particle strength of each segment is obtained by adding the square root of the dot product of each frequency vector and itself after removing the low and medium frequencies. After calculating the segment weight and then estimating the particle intensity based on a set of calculations, the second set of formulas can be used to determine a "weight" for each segment of each color channel, as shown in Figure 4. This paper scale applies Chinese national standards (CNS) A4 specification (210 X 297 mm) .24-496080 A7 _____—__ B7 _V. Step 1 1 1 0 in the description of the invention (22). Once again, after removing the low and medium frequencies, a frequency vector derived from the Fourier transform can be used. This weighted 値 provides that the calculated ‘glue intensity (noise) is actually a measure of the particle trajectory rather than the level of trust (c ο n f d e n c e 1 e v e 1) of the star from which the image is fine. To calculate this weighted 値, Equation 8-1 0 can be used. Alternatively, it is also possible to use the formula used to find the total of the frequency vector 値 of each section as follows. This formula (which is a measure of autocorrelation) will add the dot product of the frequency vectors to all frequency positions in themselves in one section. We can find the individual sum of each color channel. (Please read the Zhuyin on the back> Matters before filling out this page)

RR = IR • R GG = EG Transform. Using the results obtained from formulas 8-1 0 and .3 2-3 4, one weighted 値 for each sector can be calculated according to the following formula: 装 -------- Order --- Φ Economy Ministry of Intellectual Property Bureau employee consumer cooperatives print weighting = 1.0

\ RG— RB—GB (35) \ RRvGG Seven formulas 3 5 are useful because the same frequencies of different colors from the image part of a section 値 between color channels are more similar than those from particle trajectories . As explained earlier in this application, 'the particle trajectory recorded from each color channel has a separate pattern in a digital image' and this paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 cm) (%) -25- 496080 A7 B7 printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (23) Therefore, it has different groups of frequencies. On the other hand, images tend to be similar 'in all color channels and therefore have similar frequencies in each color channel. For example, in a section composed entirely of image signals, the cross-correlation combination 'in the numerator of the weighting' equation is approximately equal to the self-correlation combination in the denominator '. The weighted 値 of that sector will approach the square root 値 of 1.00-1, which is finally 0. f However, a segment composed entirely of particle trajectory signals has a weighted 値 approaching 1 · 0-0, which is finally 1 · 0. This is because the cross-correlation combination 値 in the numerator is much smaller than the auto-correlation combination 分 in the denominator. Therefore, the content of a sector is composed of more video signals', the weighting will be closer to zero. The content of a segment is composed of more particle trajectories (noise), and the weighted chirp will approach 1. For black and white images, this weighting is usually set to a fixed frame, such as: After performing these calculations with weighting, DC, and particle intensity mapping, the weighting frame of each segment can be drawn on the entire image in three graphics For each color channel, draw a pattern, as shown in step 1 1 1 2 in Figure 4. To be clear, a graphic is drawn for each color channel of the image, not a graphic per sector. Each segment in this digital image is represented on the graph according to its particle intensity and DC (which is the average intensity of the segment obtained through Fourier transform 値). This paper's dimensions apply the Chinese National Standard (CNS) A4 specification ( 210 X 297 public love) (Please read the notes on the back before filling this page)

· N «I I— el I n n 口 口 · ϋ_ι I— n an an m n I 496080 A7 B7___ 5. Description of the invention (24) (Please read the precautions on the back before filling this page). In addition, the points drawn for each segment are assigned to the weighted 値 of that segment. The result is a scatter plot, where the X-axis represents the average intensity, and the y-axis represents: particle intensity, and the size of a particular point on the scatter graph contains the weighted 値. For a black and white image, a single scatter diagram can be generated, which is consistent with the modified formula for particle intensity and weighting described above. Fig. 6 illustrates the information of the red channel for a section in a digital image. The y-axis represents particle strength. The X-axis represents the average intensity. The point 1 5 0 2 is the point of the intensity of the red particles and the average intensity of the red in this section. An enlargement of the point 15 0 2 attempts to exemplify the weighted chirp (W V) of the segment. In the same way, one point of this segment is drawn on the green channel on a green scatter diagram, and the blue channel on a blue scatter chart is drawn. After drawing the data of an image segment, the result is three scatter plots, one for each color channel. The figure illustrates a scattergram recorded by a digital image section ′ through a color channel, such as a red channel. A similar scatter plot will show the results for a single channel of green and blue 'or a black and white image. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs to draw a line of estimated particle intensity. Once a scatter plot is generated, a 'curve fitting technique can be used' for each color channel (or a single channel of a black and white image) Draw a line through the fog point, as shown in step 1 1 1 4 in FIG. 4. The weights assigned to each point position may affect the exact position of these lines' larger weights, pulling the lines towards them. Figure 8 illustrates a line drawn through a color channel. For example, a line drawn by a scatter plot of the red channel 17.0 2 ° This curve represents particles. The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ) 〇7-" " ~ 496080 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (25) Drawing of intensity versus intensity. Although we will discuss curve fitting in the following, any curve fitting technique can be used without departing from the scope of the invention. In addition, different techniques can be used for different color channels. These lines can thus be used to determine a useful measure of the estimated particle intensity of a pixel in a digital image of a particular color channel for any pixel intensity in that particular color channel. This metric can be used in various ways to make the digital image look more desirable. The assumption that image information exists in all three channels is true most of the time. However, the bright color areas of a digital image may contain areas with a small correlation between strong images and colors, thus resulting in occasional strong points far from the average range. In order to achieve a curve fit of the estimated particle strength, this curve fitting technique can take occasional strong points into account without letting the curve be distorted by them. For example, it can use a median (med i an) average of the points, or it can find a prototype that most closely matches (prototype). It can also make the most consistent first prototype, suppress the strong point that is twice as high as the prototype line, and then make a second prototype line on the revised data. Such strong points, most likely they represent image details, not noise, so 'suppressing these points may produce a more accurate curve. This most appropriate operation can be performed many times until no point is discarded. In this embodiment, we use a median averaging technique. Figure 9 shows the line 1 8 0 2 drawn for the red information of a digital image. Line 1 8 0 2 can be the same curve or a curve proportional to the curve associated with the scatter diagram, such as: The paper size of the paper in Figure 8 applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) PCT) -28- (Please read the notes on the back before filling this page)

· TaMm i In n 11 ϋ— ϋ One sip, I H ϋ tmmmmm n n ϋ * n I 496080 A7 B7 5. Description of the invention (26) 1 7 0 2. Note: Here the X-axis represents the pixel intensity 値 'and not the average pixel intensity of the entire segment used previously. For a pixel with point 1 0 0: the intensity of the particle is estimated at the point 18 0 6 ° In other words, although the curve 1 70 2 is obtained from a scatter plot based on the average intensity, but if To set the intensity of a pixel, curve 1 0 2 (the curve can be the same as curve 1 7 2) can be used to provide an estimate of the particle intensity of the pixel. Measure the intensity of particles related to frequency After measuring the particle trajectory, you can measure the intensity of particles related to frequency for each color channel (or a single channel of a black and white image), as shown in step 1 in Figure 3. 0 4 shown. Frequency content represents image detail. A section with a higher degree of image detail will have more content at all frequencies. Example: Consider an image of a tree surrounded by the sky and standing on a meadow. The grassy part of the image, due to its many leaves and shades of color, has most of the high energy at all frequencies, that is, 'a large frequency content. Many high energies may have high frequency content. On the other hand, the sky part lacks image details, so it has high frequencies that rarely produce this image. Therefore, any measured high-frequency content can therefore be assumed to be generated primarily from grainy details. To measure the intensity of the particles in relation to frequency, we performed the steps illustrated in Figure 10. In step 2 0 2, the digital image is segmented for the second time by the segment, and in step 2 0 4, the second Fourier is performed. The paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297). (Mm) (Please read the notes on the back before filling out this page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs -29- Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7 ___ V. Description of Invention (27 ) Transform, although, the results of the first segmentation and the first Fourier transform may alternatively be used. For new calculations, different section size, shape, and overlap options can be used (such options have been described previously), and different types of transformations can be used (again, such options have been described previously). In step 2 0 0, in order to identify the estimated particle intensity as a function of frequency, we generate a scatter plot with a fitted curve. Secondly, in step 2008, we calculate the intensity function according to the particle intensity. Among them, the function is obtained by correlating each segment with the frequency 値, and draws the size divided by the estimated particle intensity. In step 2 0 0, draw a line identifying the frequency-dependent estimated particle trajectory. Segment Segmentation When measuring frequency-dependent particle intensities ' In step 2012, the digital image can be segmented into segments using any of the techniques and options of step 1102 in FIG. 4 previously described. Fourier Transform In step 2 0 4 we use any of the techniques and options previously described in connection with step 1 1 4 in Fig. 4 'for each color channel (or a single channel in a black and white image) A second Fourier transform is performed on the pixel intensity of each segment. However, the range of low and medium frequency terms has not been removed, as shown previously in step 1 106 of FIG. Identification of frequency-related estimation questions. Grain track 1 This paper applies the national standard (CNS) A4 specification (210 X 297 public love) -30-(Please read the precautions on the back before filling this page) Pack. 496080 A7 B7 V. Description of the invention (28) (Please read the note on the back? Matters before filling out this page) Using the DC (average intensity) obtained from the second Fourier transform 値, we can identify each color channel (or In a single channel of a black-and-white image, the frequency-dependent particle intensity of each segment is as shown in step 2006. Lines previously drawn through the scatter plot of each color channel (or a single channel of a black and white image) (such as lines 1702 and 1802) can be used for this purpose. For each segment, the DC (average intensity) 値 obtained from the second Fourier transform can be used / for the intensity of the curve 値 to locate a point on the X axis, and therefore, find the estimated on the y axis Particle strength. For example, if a segment has a D C 値 which is equivalent to 値 at point 1 8 0 in FIG. 9, its estimated particle intensity will be 値 at point 1 8 1 0 on line 1 8 2. Therefore, each frequency of a segment is considered to have an estimated particle intensity derived from the average intensity of the segment. Book M / GS vs. Frequency Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs in step 10 of Figure 10, after identifying the intensity of particles related to frequency, in each color channel (or In a single channel of a black and white image), the size of each segment at each frequency can be determined by the estimated particle intensity of each segment in each color wave channel (or a single channel of a black and white image). segmentation. The magnitude of each frequency is determined by the Fourier transform in step 2010. A new scattergram can then be generated, one for each color channel (or one for a black and white image). For the channel in question, the y-axis of the scatter diagram represents the size of each section on each channel, where 'the 値 is the paper size applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm -31-496080 A7 ____ B7 _ V. Description of the invention (29) (Please read the notes on the back before filling this page) It is divided by the estimated particle intensity of that area. For the channel in question, the X-axis of the scatter plot represents the frequency chirp present in the digital image. Therefore, for each frequency in each segment in the digital image, a point on the scatter plot appears during the day. In the example above, 289 frequencies are presented, so the new scatter plot in this example will have 2 889 S drawn points, where S represents the number of segments. In this scatter diagram, there is no weighting added to each point except for the number of points drawn at the same position. Specifically, each point has a weight of 1, but if the second point is drawn at the same position, the points that exist can be assigned a weight of 2. Alternatively, a list of points can be easily maintained and contain two points. FIG. 11 illustrates such a scatter diagram of a color channel. A lower "fog point" represents those segments that have been isolated by the above weights, and a higher probability represents the main particle intensity rather than the image signal. Those points removed from the fog zone have a higher probability of representing the image signal. Book lines to identify the particle trajectories related to frequency. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Second, use any curve-fitting algorithm to draw all points in the lower misty bands in each scatter plot. Line, as shown in step 2 10 in Figure 10. In this embodiment, we use a median average type algorithm. When this line is drawn, the erroneously higher picture points 'outside the foggy band' can be ignored, and thus the results will not be distorted. In order to do this, you can use the following "weighting" formula for each point, where y is a point on the y-axis: This paper size applies the Chinese National Standard (CNS) A4 (210 X 297 mm> -32 -Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7_ V. Description of the invention (30) Any other suitable weighting can be used without departing from the scope of the present invention. Figure 12 shows a color channel The line Έ 2 0 2 drawn by the foggy band is used as an example. Note: When drawing this line, the wrong high points 2 2 0 4 and 2 2 0 6 are ignored. At the same time, only medium and high frequencies are used It may be advantageous to determine this line by plot points. In this way, strong low-frequency content will not adversely affect the estimated DC. Calculate the correction array and then calculate a correction array for each color channel, as shown in Figure 3 Step 1 0 0 6. Find the DC term estimated by the graph. To calculate the correction array, we use the scatter plot drawn in step 2 1 0 (Figure 1 0) to calculate each color. Channel (or single in a black and white image Channel) in an estimated DC (EDC). For example: Figure 12 shows the EDC 値 indicated by the point 2 2 0 8 on the line 2 2 0 2. Then, in this example, these three scatter plots provide An estimated DC 値 for red, green, and blue. After calculating the EDC 値 for the array, use Equation 37 to calculate an array for each color channel (or a single channel in a black and white image), where [x] Cover the range of pixel intensity 値: Array U] = (complex · grain intensity [,]) (3 times) This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)-33: (Please read first Note on the back then fill in this page) · in a— · n I «ϋ ai« i · — J ·· n «ϋ I nn _ Ύ Ύ 4 496080 A7 B7 V. Description of the invention (31) (Please read first Please fill in this page again on the unintended side) Use the estimated DC 所得 obtained from the curve calculated in step 2 1 10 and the intensity of each particle obtained from the curve calculated in step 1 1 4 § ten particle strength, An array is calculated for each channel. The calculation and correction array [X] can be used to calculate each channel (or A black and white channel) is a correction array. We can use Equation 3 8 to calculate this array. Correlation [〇] = 〇 Correlation [小 έΣ (Array Array ⑽ (38) / 7 = »! 2 Particle trajectory that will be related to intensity After the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs prints and calculates the correction array, the particle trajectory of the digital image can be normalized with respect to intensity, as shown in step 10 of FIG. 3. The chirp obtained from the corrected array can be applied to the data of this digital image to achieve normalization. The intensity 値 "X" of each pixel in the digital image can be replaced by 色彩 in the correction [X] for each color channel (or a single channel of a black and white image). This procedure makes the particle trajectory in the digital image have approximately the same visible impact (V 1 s u a 1 1 mp a c t) in the gray level of each channel, which makes the digital image look more desirable. For example, it is more desirable when it is displayed on a typical display with a contrast factor (gamma) of about two and a positive film. Figure 13 is from a normalized paper with pixel dimensions in space. The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love): 34-'Employee Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs Print 496080 A7 B7____ V. Description of the invention (32) Graphical representation of the particle trajectories of the lines of the pixels of the digital image with a change in intensity. It should be noted that the range of pixel intensities over the entire graphic is uniform and does not fluctuate as sharply as the pixel intensities in an unnormalized digital image. In Figure 13 the fluctuations at point 2 3 0 2 are approximately the same as at point 2 3 0 4. Normalizing the particle trajectories related to frequency The particle trajectories of digital images can now be normalized in relation to frequency, as shown in step 1 0 1 0 in FIG. 3. After applying the correction array, we can perform another frequency transformation and transform each channel (or the single channel) of the image that has been normalized in intensity from the time domain to the frequency domain. Alternatively, the correction array can be converted into a function in the frequency domain for use in the original spatial frequency domain representing the image. In this embodiment, a correction array is applied in the time domain and conversion is performed on each channel to perform a frequency normalization procedure to be described. Figure 14 is a graphical 2D plot of the size versus frequency in a digital image after the software has normalized the particle trajectory across the intensity. Here, the average (DC) 値 2402 is 1 · 0. To normalize the particle trajectory across frequencies, this software multiplies the magnitude (M) of each frequency in each segment by a factor (F). This factor refers to the extent to which the line 2 4 0 2 drawn via the scatter diagram deviates from 1.0. Deviation factor (F) = (39) This paper size is applicable to the Chinese National Standard (CMS) A4 specification (210 X 297 mm). 35-(Please read the precautions on the back before filling this page) Installation ----- --- Order ------... 496080 A7 B7 V. Description of Invention (33) (Please read the notes on the back before filling this page) The actual graph of the deviation factor will show a frequency increase curve, The curve is an inversion of the line 2 4 0 2 in FIG. 14. This phase is not affected and is retained. Figure 15 graphically represents a 2D plot of frequency versus frequency in a digital image after the software has multiplied the deviation factor. In the conventional way, a line 2 5 0 2 drawn through this figure shows that the size of the particle trajectory is large and the area of the high and low frequencies is equal. They do not fluctuate as sharply as in an unnormalized digital image. Suppressing particle trajectories After calculating the correction array, the normalized 値 obtained from the correction array can be used to suppress the particle trajectories in the digital image, as shown in step 1 0 1 2 in Figure 3. Other methods of particle suppression may be used without departing from the teachings of the present invention, such as the second method described below. In the digital image normalization according to the teachings above, in each color channel, the intensity of the particles in each segment and in each frequency has been adjusted so that the size component is 1 · 0, as shown in Figure 1 5 As shown. The aspect is not affected by the printing of employee cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. Therefore, the size of a component is a powerful indicator that isolates the particle trajectory (noise) from the image details in the 1.0 range is a number of watermarks. Sizes greater than 1 · 0 are more likely to be generated from an image detail, and sizes less than 1 · 0 are more likely to be generated from a particle trajectory. Therefore, in order to suppress particle trajectories in a digital image, it is best to suppress components that are close in size and smaller than 1.0. On the other hand, components larger than 1.0 can be considered to represent image details and should not be suppressed. -36- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 496080 Ministry of Economic Intelligence Printed by the Consumer Cooperative of the Property Bureau A7 B7 V. Invention Description (34). This is in contrast to unnormalized digital images, where 'the' number of watershed lines between glume trajectory (noise) and the image varies with intensity and frequency, making it almost impossible to accurately match the particle trajectory with Image isolation. As mentioned earlier, images can be normalized 'or data describing the distribution of noise with intensity and frequency can be included in an unnormalized image' to guide particle reduction without departing from the invention Range. In order to suppress particle trajectory, the following method can be applied to the size (M) of each frequency in the normalized digital image segment: F οr each segment in the normalized digital image segment Maintain the original phase information F οr each M I f (Μ < = 0 · 5) in this section (including the DC term) Μ = Ο else suppression factor = 1.0. E-(Μ- 0.5) Μ = Μ * The suppression factor formula suppresses 値 by replacing the normalized frequency of 0 or less with 0. Because some glume trajectories have a normalized size slightly larger than 0.5, this formula can also use a suppression factor to replace their 値 with smaller 値. This suppression factor eases the reduction of frequencies with a higher normalized size, 'because' these tend to correspond to the image details. The above example has a digital image, where the digital image is based on the Chinese paper standard (CNS) A4 (210 X 297 mm) -37------ ί ---: --- -_ 装 i — (Please read the notes on the back before filling out this page). Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 ____B7__ 5. Description of the invention (35) Known particle strength as a function of frequency and intensity Each segment is divided to normalize it, and then the reduction is applied with a watershed fixed at 1.0. A comparable method can also apply the particle reduction of a watershed point with expected noise equal to a function of frequency and intensity directly to the unnormalized components. In order to carry out this method, we can use the above-mentioned glume suppression method, and in the suppression factor, M ′ is used instead of M ′, where the desired particle trajectory of __M advantageously uses the normalized 値 from the calibration array, Other methods of suppressing glomerular trajectory are possible. For example, image-improving software such as PHOTOSHOP can be combined with the normalized combination manually, although such software requires extensive user feedback to find the appropriate level of suppression of particle trajectories. Now, another method for suppressing some or substantially all particles of an image, either automatically or at the user's choice, will now be described. FIG. 16 illustrates a flowchart illustrating a method for removing high-frequency noise such as grain tracks from a digital image. The method shown in Figure 16 can be applied to one or more channels of the digital image or, if the digital image has only one channel, it is applied to a single channel of the digital image. For an image with multiple channels, the method described in Figure 16 can be applied to a subset of the total channels without departing from the scope of the present invention. Likewise, the method described in FIG. 16 can be applied to each of a plurality of channels in a multi-channel digital image without departing from the scope of the present invention. In step 1610, the noise of a particular image channel is normalized. This noise can be normalized using the above method or any other method ---, --- »(* ---- ^ 装 —— (Please read the precautions on the back before filling this page) · This The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love) -38- 496080 A7 B7___ V. Description of the invention (36). Alternatively, although before the rest of the method described in Figure 16 is performed It is better to normalize this noise, but this method can also be applied to an unnormalized digital image. The method described in Figure 16 can be applied to high frequencies that used to have particle traces such as: Noise is normalized to a digital image of size 1 at all frequencies. If the particle has been normalized to a chirp other than 1, then the remaining steps of the method described in Figure 16 are applied Previously, the intensity 値 of the image channel could be scaled to 1. For example: if this particle is normalized to a size of 0 0 at all frequencies, then after performing step 16 2 and before performing other steps, Each frequency can be divided by 100. In a more complex example, if each frequency High-frequency noise, which is normalized to a unique chirp, can be divided by the appropriate chirp before the rest of this procedure begins. This division can also occur after the transformation of step 16 2 as shown above. The method exemplified in 16 can be used for images with a single channel or multiple channels. As long as there are multiple channels, the method described in Figure 16 can be used. Alternatively, the inter-channel An improved method of relationship can be used. Figure 17 related to this method is illustrated as follows. The description of Figure 16 assumes that this method is applied to a single channel, M ', where M' represents a digital image. The intensity of each pixel is arranged in space. For a multi-channel image, there are such multiple channels. In step 16 2, most sections of the image channel are transformed from the spatial domain to the frequency. Domain. Such segment segmentation and transformation can occur according to the above-mentioned methods associated with noise normalization. We can use any transformation that segments and transforms image segments from the spatial domain to the frequency domain. In this embodiment, this Paper size applies to Chinese national standard Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling out this page) Installation -------- Order -----------. Intellectual Property Bureau, Ministry of Economic Affairs Printed by employee consumer cooperatives-39- Printed by employee consumer cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7 _ V. Description of the invention (37) 'I use a size of 16 pixels by 16 pixels. The size of other segments It can be used without departing from the scope of the invention. In addition, as described above, the sections can overlap without departing from the scope of the invention. Similarly, other section shapes and sizes can be used without departing from the scope of the invention. The scope of the invention. For a particular segment, the resulting transformation can be labeled as M'T. As noted above, such transformations generally include a series of two-dimensional vectors, where one component of this vector represents the magnitude of a particular frequency component, and the other component of the vector represents the phase at that particular frequency. In this embodiment, each transformed vector can be conveniently configured in two directions, simplifying the description of the method performed in FIG. 16. Figure 18 illustrates these two directions of the organization. In Figure 18, the vectors are arranged in a 17x17 matrix. The DC frequency term is placed in the center of the matrix and is assigned a label (0, 0). In addition, as shown in Figure 18, this vector is assigned an index 1 4 4. This index can be used when the frequency vector is configured in a one-dimensional array of vectors. In a one-dimensional array, the DC vector is on index 1 44. In a two-dimensional array, the frequency vector is placed on an index, where the index corresponds to a suitable change in frequency and in that position in the x and y directions. In a one-dimensional array, the first frequency vector of the one-dimensional array on the index 0 is the vector appearing in the upper left corner of the two-dimensional array (coordinates 8,-8) illustrated in FIG. 18. The placement of the vectors in the one-dimensional array continues in numerical order, where the reading of the two-dimensional array is from the first column from left to right and the second column from left to right, and so on, until the last column. This configuration is marked by the one-dimensional array enclosed by the brackets in Figure 18, where the marking is a special paper size applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) -40-(Please (Please read the notes on the back before filling this page) Binding ---- Order ------; --- '496080 Α7 Β7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economy , Listed along the two-dimensional array coordinates. In step] L 6 χ 4, the: transformation of each section obtained in step 16 12 can be filtered. Alternatively, 'this step may be omitted' without departing from the scope of the present invention. For the sake of convenience and consistency of the description associated with FIG. 17 ', the transformation obtained at step 1612 may be designated as M'T2. This transformation may be filtered to emphasize signal content and attenuate noise in the image. In this step of the method, it may be advantageous to use a low-pass filter (0 0 pas s f 1 1 t e r). Suitable filters tend to get smooth attenuation and reduction in random noise. Therefore, although any low-pass filter can be used, a suitable filter may be an ideal filter for improving the effective signal present in the transform and suppressing the noise present in the transform. An RMS (root mean square) calculation can be used to further emphasize the effective signal. This filtering operation tends to retain strong changes in frequency and suppress small changes in frequency. In other words, the strong frequency components that are likely to represent a valid signal can be exacerbated. In this embodiment, the filter contains a 3 X 3 matrix, where 値 in the center is 4; 在 in the horizontal left and right or vertical points in the center is 2; and 値 in the 4 corners of the matrix is 1. . Other size filters or other filters may be used without departing from the scope of the invention. If w (a, b) represents the above-mentioned filter, the filtered transformation M'T3 can be calculated according to Formula 40. ,, y]:

Σ Σ Ma, b) (Mr2 (x ^ ayy ^ b)) 1 · -_ 丨 '__I Σ Σ Κα, ό) (40) f Please read the note on the back first and then fill out this page} M ， T3 [ DC] This paper size is in accordance with China National Standard (CNS) A4 (210 X 297 mm) -41-496080 Printed by A7 B7, Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of Invention (39) As noted: In the filtered transform, the DC term is set to zero. In addition, elements outside the edges of this transformation are considered to have zero. In this formula, this treatment is not clearly shown. This filtering operation only affects the size of the frequency vector. The phase of the frequency vector remains unchanged. In Equation 40, we use the transformed two-dimensional representation for calculations. For the purposes of notation, the fact that the size of the transformation is used is indicated by the use of ordinary letters (as opposed to bold letters) in Equation 40 as a matter of transforming M'T2 and M "T3. This notation will be in Illustrate that Figures 16 and 17 are used consistently. When performing a vector operation 'will use bold letters on the particular transformation in question. In steps 1 6 1 6 the filtered transformation of each segment Can be frequency spread. Alternatively, if step 1 6 1 4 is omitted, the original transform obtained in step 1 6 1 2 can be frequency spread in step 1 6 1 6. Step 1 6 1 6 can also be omitted without departing from the scope of the present invention. Because the real world is composed of similar repeating patterns, so the transformation space of a photographic image tends to be any magnification around the DC frequency, and has the same appearance Unless there are duplicate artefacts. For repeated patterns, the frequency characteristics of an image tend to repeat in the second power. In this step, the frequency characteristics at the lower frequency are used to double the frequency in advance. Make door detail Easily photosensitive. In order to perform such frequency expansion, if we use a two-dimensional version of the transformation illustrated in Figure 18, then the center half of this transformation data is enlarged to a full-size block of the transformation data. In other words, the center of the frequency is 9 x Block 9 (Please read the precautions on the back before filling out this page) Loading -------- Order ----- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) ) -42-496080 A7 B7 V. Description of the invention (4〇) expands to every other position in the 1 7 X 1 7 transformation. When a frequency and a half are in the remainder of the transformation, it is not explicitly entered You can use the mean 値 of adjacent frequencies. Again, this operation only affects the frequency. After the frequency expansion of the filtered transform is obtained, it can depend on the frequency expansion transform and the result obtained in step 1 6 1 4 The filtered transform 'calculates a new transform in step 16 1 8. Alternatively, if step 1 6 1 4 is omitted, the new transform depends on the original transform and frequency expansion transform obtained in step 1 6 1 2 Equation 4 1 can be used to calculate this new transformation M T 4. M, T4 [x] = 73 [x] V (0.2 ^ BT [x]) 2 well 4 (Please read the precautions on the back before filling this page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs In Equation 41, the transformation M ′ ET represents the estimated transformation obtained by the frequency expansion. Again, this operation only affects the size of the frequency vector ° In step 1620, 'we can generate a mask (n ^ sk)' such that The high-frequency term of the transform obtained in step 16 18 is attenuated. Since the particle trajectory tends to have the main high-frequency component, this mask can be used to attenuate the particle trajectory and / or other high-frequency noise . This mask can be obtained using a gating function (2at inS f u n c t i ο η) represented by Equation 42.

ϋ ϋ «ϋ n —Bi i < n i n an n« n n I% · 64 (42) Equation 4 2 is a two-dimensional representation of the Fourier transform illustrated in FIG. 18. The X and y terms in Equation 4 2 are configured as shown in Figure 18. The paper size applies the Chinese National Standard (CNS) A4 (210 X 297 mm) -43-496080 A7 B7 V. Description of the invention ( 41) When the two-dimensional matrix is exemplified, it corresponds to the coordinates of the special frequency vector in the transformation. Therefore, this gating function approaches 2 for high frequencies and approaches 0 for low frequencies. This mask can then be calculated using Equation 43, where B (X, y) is included in the mask of X, y, which also refers to the coordinates in the two-dimensional transformation illustrated in Figure 18. B (x, y) = l-Fgatc (x, y) Fgatc < 0.4 (43) B (x, y) = 0.6

Fgacc > 0.4 In this implementation, this mask function is designed to have a minimum 値 of 0.6. Other mask functions may be used without departing from the scope of the invention. Moreover, this step can be omitted without departing from the scope of the invention. In step 162, a noise correction function can be calculated for each segment. In this embodiment, the noise correction function is calculated in response to one of the mask and one of the frequency domain representations of the previously calculated image. For example, the transform M ′ T 4 obtained in step 16 18 can be used to derive the noise correction function. The noise correction function can be obtained from Equations 4 4 and 4 5. (Please read the precautions on the back before filling out this page) One Pack II One r · n · Βϋ I 11 n · Βϋ I iso · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economy l + FKatc (x, y) c > 1 + F_ (x, y) M'TM [x, y] = GR + (1-GR) C (x, y) (45) In Equation 4 4 The noise correction function is limited to 1 plus the gating function at a particular position in the transformation matrix. Again 'Formula 4 4 affects only the magnitude components of the frequency vector. Users of the present invention may want this paper size to comply with the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -44-496080 Α7 Β7 Residual graininess remains on the improved image according to the invention. Therefore, the present invention allows the user to specify the grain residue (g r a 1 n r e s i d u e) G r that he wants to leave after the image is improved. Equation 4 5 takes into account the desired residual particles when calculating the noise correction function M'TM. In step 16 2 4 the noise correction function of each zone is applied to the original transformation of that zone. The result is, for example, the frequency domain representation of the image after the high-frequency noise of the particle trajectory is suppressed. Using Equation 4 6 gives a frequency representation of the suppressed particles. M'TSU] = M'TU] M, TM [x] (46) Equation 46 produces a series of vectors including the product of the vector's M 'T series and the corresponding vectorless unitary unitary in M' TM. In step 16 26, the inverse transform of the noise correction transform M'TS can be calculated for each sector to obtain the spatial domain representation of the improved image after the noise has been suppressed. The method depicted in Figure 17 can be used in digital images with multiple channels. If a digital image has only a single channel, the method described in association with FIG. 16 is more applicable. We can also use the method described in Figure 17 on some of the many channels of the digital image, and apply the method described in association with Figure 16 on other channels. The method described in association with FIG. 17 can be applied to any multi-channel digital image without departing from the scope of the present invention. Examples include digital image representations of color photos (printed photos, negatives, or slides), digital images received from satellites, and medical images. The present invention is very effective in removing uncorrelated high-frequency noises such as those caused by particle trajectories from digital images (please read the precautions on the back before filling this page).

l # 11 I bite, · ϋ in I in · Ί · — 11 I%-This paper size applies to China National Standard (CNS) A4 (210 X 297 public love) 45 496080 A7 B7 ____ V. Description of the invention (43) it works. In step 1710, the noise of each image channel is normalized. Noise may be normalized using the above or any other method. Alternatively, although the noise is better normalized before implementing the rest of the method described in FIG. 17, this method can also be applied to an unnormalized digital image. The method described in Figure 17 can be applied to high-frequency noise with particle trajectories, which are normalized to a digital image of size 1 at all frequencies. If this particle has been normalized to a non-1 chirp, the intensity 値 of this image channel can be scaled to 1 before the remaining steps of the method described in Figure 17 are applied. For example ... If this particle has been normalized to 値 of 1 0 at all frequencies, then after performing step 1 7 1 2 and before performing other steps, each frequency can be divided by 1 0 0 . In a more complex example, if each frequency has been normalized to a unique chirp for particles or other high-frequency noise, then it can be divided by a suitable chirp before proceeding to the remaining steps. This division is also performed in the transformation that occurs in step 1712. The method exemplified by Fig. 17 makes use of the relationship between multiple channels of a digital image. For convenience, the method illustrated by FIG. 17 will be described below in association with a digital image representing a color photographic image. Such digital images usually have three channels-a red channel, a green channel, and a blue channel. To extend the method described in Figure 17 to images with more channels, the cross-correlation functions calculated below may be extended to take into account all useful cross-correlation. Similarly 'If Figure 17 is applied to a dual-channel digital image, the cross-correlation described below can be adjusted. This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -46- (Please read the precautions on the back before filling out this page) Installation% Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7___ V. Description of Invention (44) To consider only these two channels. Figure 17 applies this method to three channels, R''G 'and B'. (Please read the notes on the back before filling in this page) R ′ represents the intensity of each pixel in the red channel of a digital image. G 'represents the intensity of spatial arrangement of each pixel of the green channel of a digital image. B ′ represents the spatially arranged intensity of each pixel of the blue channel of a digital image. The notation used here can avoid confusion with the previous formula. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. In step 17 12, most sections of each image channel are transformed from the spatial domain to the frequency domain. If high-frequency noise is removed from only a subset of the plurality of image channels, only those image channels can be transformed or all image channels can be transformed. In other words, if the noise is only removed from a single channel, the information of other channels may or may not be used to determine the noise content of the channel from which the noise is to be removed. The segmentation and transformation of each channel can occur according to the above-mentioned methods associated with noise normalization. Any method of segmenting the image channels or any method of transforming from the spatial domain to the frequency domain can be used. In this embodiment, we choose a segment size of 16 pixels X 16 pixels. Other segment sizes can be used 'without departing from the scope of the invention. Further, as described above, the sections may overlap without departing from the scope of the present invention. Likewise, other segment shapes and sizes can be used 'and + deviates from the scope of the invention. Variable shape and size segments can also be used. For a particular segment, used as a photographic image of the present invention, the resulting transformation can be labeled R 'T, G' τ, and B. 'As noted above, such a transformation includes a series of two-dimensional vectors.' The paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 717 ^ 496080 A7 B7 Staff Consumption of Intellectual Property Bureau, Ministry of Economic Affairs Cooperative printed 5. Invention Description (45), one component of this vector represents the magnitude of a specific frequency component, and the second component of the vector represents the phase at that specific frequency. In this embodiment, each transformed vector can be conveniently configured in two directions, and the description of the method performed in FIG. 1 is simplified. Figure 18 illustrates these two directions of the organization. Figure 18 was previously described in association with the description of Figure 16. In step 1714, the cross-correlation between the corresponding sections of the complex image channel can be calculated for the transformation that has been calculated in step 1712 for each section. Alternatively, a subset of the segments may be used without departing from the scope of the invention. Because a plurality of channels of a digital image tend to display the same image pattern, the corresponding sections of cross-correlation of different channels will tend to be image-details, rather than random noise such as particles. Inadequately correlated frequencies will tend to represent noise, such as particles. Therefore, the present invention makes use of the correlation between the channels of the image details. Various embodiments of the invention may utilize this relationship in a variety of different ways. The following mathematical description provides an example of how correlation can be used to remove noise. However, by performing calculations that are proportional to the cross-correlation between channels of a digital image, we can use any mathematical method to reduce the noise in the digital image, including the reduction of particles, without departing from the present invention. Range. Equations 4 7-4 9 provide a method of measuring the cross-correlation between the channels of the digital image. RT [x] -GT [x] RO, X [x] = 1 + | RT [x] || GT [x | where DC = 0 This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) (%) (Please read the unintentional matter on the back before filling this page) «tt tm§ ammmmw I n« ϋ — n a e, ttn I n < (47) -48- 496080 A7 B7 V. Description of the invention ( 46) RT [x] -BT [x] R, B, X [x]: (48) where DC = 0 (49) (Please read the notes on the back before filling this page) 个 呵 一 —P, TW , B, TW 1 10 丨 G, T_B, TW | Where DC = 0 These formulas printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs can be applied to each frequency vector in each section of the digital image. The calculation result of Equation 4 7-4 9 is the matrix of the number of cross-correlation, in which each unit is proportional to the cross-correlation between two frequency vectors in different channels, where the frequency vector represents the two channels in this channel. Space corresponds to the same frequency in the segment. For convenience, Equations 47-49 use a vector representation of a one-dimensional matrix of the spatial frequency domain transformation of the image section discussed in connection with FIG. Therefore, the numerator of each formula is the dot product of two vectors, and the denominator of each formula takes into account each size of these vectors. The matrix generated by Equation 4 7 is proportional to the cross-correlation between a particular segment at a particular frequency, red and green channels. Equation 4 8 produces a result that is proportional to the cross-correlation between a particular segment of the digital image at a particular frequency, the red and blue channels. Equation 4 9 produces a matrix that is proportional to the cross-correlation of a particular segment at different frequencies, between green and blue channels. We can use these three formulas to calculate a matrix for each segment of the digital image. The cross-correlation DC term is set to zero. I use the R, G ', R, B' and G 'B' notation to avoid confusion with the particle normalization formula described above. To avoid confusion, I use R ′ T, B ′ T and G ′ T notation to represent the transformation obtained in step 17 12. -49- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 496080 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (47) For large in-phase frequency vectors, The results of Equations 4 7 to 4 9 tend to approach 値 1, and for large out-of-phase frequency vectors, 'approaching to 値 1 °, small vectors tend to get results close to 0, while in-phase and approximately A vector of size > of 1 produces a result of approximately 0 · 5. In step 1 7 1 6, the cross-correlation high-frequency components related to the red channel may be attenuated. Alternatively, this step can be omitted without departing from the scope of the invention. Generally, the color film used to produce a photographic image has a green layer, in which light passes through the layer, then the blue layer, and then the red layer. Because the red layer receives light after it passes through the green and blue layers, the image in the red layer is slightly blurred compared to the green and blue layers. The red layer tends to lack the high-frequency information that makes up the image information. Instead, most of the high-frequency information from the red channel tends to be noisy and can be discarded. Using different types of color film to the right 'may attenuate different channels of this digital image. In addition, if other images, such as satellite images or medical images have channels that tend to lack high-frequency image content, then the methods described below or similar methods can be used to attenuate the high-frequency components of those image channels. Equation 5 0 illustrates a red frequency attenuation that can be used in the present invention.

RFA (0,0) = 0 This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -50- (Please read the precautions on the back before filling this page)

· · Ϋ ϋ m I n I ϋ 1 ° J0 i — nm —a 1 n I · 496080 A7 B7 V. Description of the invention (48) (Please read the precautions on the back before filling this page) In formula 5 0, X corresponds to the X coordinate exemplified in the two-dimensional array of frequency vectors shown in FIG. 18 for a particular point, and y corresponds to its y coordinate. D C 値 at coordinates (0 '0) is set to zero. Although we can easily calculate the red frequency attenuation 値 using the coordinates corresponding to the position of each frequency vector in the two-dimensional configuration of the frequency transformation illustrated in Figure 18, the calculation result of Equation 50 can be stored in In a matrix of multiple chirps, each chirp in the matrix includes: the result of formula 50 and the position of the chirp in the matrix, where the position corresponds to a one-dimensional array represented by a vector of frequency transformation, corresponding to The proper position of the frequency. For example: using the example transformation illustrated in Figure 18, the 値 of R F A [2 8 8] should be obtained by applying formula 50 to coordinates (8, 8), and RFA [144] is the 値 of RFA (0, 0). After calculating the red frequency attenuation 値 using Equation 5 0, these 値 can be applied to the cross-correlation between the red and green channels using Equation 5 1 and between the red and blue channels using Equation 5 2 Cross-correlation. R, G, X2 [X] = RFA [X] R, G, X [X] (51) Printed by R, B, X2 [X] = RFA [X] R, B , X [X] (5 2) These formulas include: a majority of a double number of chirps per frequency. Although I have exemplified a method of attenuating high-frequency red information 'but' another method can be used without departing from the scope of the present invention 1 ° In step 1 7 1 8 the total cross-correlation between channels can be calculated Measurement. This calculation is performed for each segment of this digital image 'and this paper size applies Chinese National Standard (CNS) A4 (210 X 297 mm) -51-496080 A7 B7 Employees of the Intellectual Property Bureau, Ministry of Economic Affairs Printed by the Consumer Cooperative V. Invention Description (49) is a representation of the total cross-correlation between the various image channels in that sector. Although we can use any measurement that is proportional to the total cross-correlation between channels, Equations 5 3-5 6 provide a measure of the total cross-correlation of a particular segment. 288 CCC1 = X (RPA [X] (RT [X] GTPC])) + (BT [X] GT [X]) (53) χ-0 but the limitation is: cccm 288 CCCN = J] RFAPC] χ- 0 CCC2 = CCC1 /-/1.2 VCCCN CCC3 = --0.2-_5CCC2 1 + CCC2 (0.25) In this embodiment, Equation 5 3 is applied to consider only between the red and green channels and between the green and blue channels. A cross-correlation measurement. Because the image passes through the other layers of the photographic film to reach the red and blue layers, the cross-correlation between the red and blue layers usually does not add any valuable information about similar image details. Therefore, this term has been omitted from Equation 53. However, such an item may be used with or without frequency attenuation without departing from the scope of the invention. In addition, for the above reasons associated with the red frequency attenuator, the correlation between the red and green channels is attenuated in Equation 53. In this case, Equation 5 3 uses the 2 8 9 vector example used in the example in Fig. 18 to calculate the total cross-correlation in the entire frequency domain. Equation 5 4 reflects the measurement of the cumulative plus 棹 added by the red frequency attenuation. Finally, equations 5 5 and 5 6 can be applied. The resulting cross-correlation measurement range is 0-1 and when the image details are not present, about (54) (55) (56) This paper scale applies Chinese national standards (CNS) A4 size (210 X 297 mm) -52-(Please read the precautions on the back before filling this page)

I n n n ^ eJI βϋ n I n n ψί i 496080 A7 B7 5. The description of the invention (50) is 0.2. In step 17 2 0, a weighted value for each section of the channel can be calculated based on the original transformation calculated in step 1 7 1 2 and the cross-correlation between each channel and other channels. Transform. We can use any weighted transformation in response to any measurement related to one or more channels of this digital image without departing from the scope of the invention. In this embodiment, equations 5 7-5 9 can be used to calculate the weighted transformation. RT2 [X] = | RT: [Xj + 〇 · 9 R'GX2 [X] + OJRBXZfX], but the restriction is R'T2 [XP0 (57) DC = 0 GT2 [X] = | GT [X]. .9G'B'X [X] + OJR'GpfX] but the restriction is G'T2 [X] > 〇 (58) DC = 0 BT2PC] = | BT [X] | + O ^ GIB ^ PC] + OJRO ^ pC] But the restriction is B, T2 [XP〇 (59) DC = 0 In each case, the weighted transformed DC term is set to 0. In addition, each component of the weighted transformation is also limited to a value 値 greater than or equal to zero. Note ·· Equation 5 7-5 9 calculates only the size part of the frequency transform, and generates a modified size for each transform. In step 17 2 2, a weighted post-transformation of each section of each channel obtained in step 17 20 may be filtered. Alternatively, this step may be omitted without departing from the scope of the present invention. In addition, 'If step 1 7 2 0 is omitted, the transformation obtained in step 1 7 1 2 can be filtered for each channel. This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (please Please read the notes on the back before filling this page.) -------- Order ------, --- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs -53- 496080 A7 ____ Β7 ____ Description of the invention (51). Each transform may be filtered to emphasize signal content and attenuate noise in the image. In this step of the method, it may be advantageous to use a low-pass filter. Suitable filters tend to get smooth attenuation and reduction in random noise. Therefore, although any low-pass filter can be used, a suitable filter may be an ideal filter for improving the effective signal existing in the transformation and suppressing the noise existing in the transformation. An R M S calculation can be used to further emphasize the effective signal. This filtering operation tends to keep strong changes in frequency and suppress small changes in frequency. In other words, the aggravable strong frequency components that are likely to represent an effective signal. In this embodiment, the filter contains a 3 X 3 matrix, where 値 in the center is 4; 在 in the horizontal left and right or vertical points in the center is 2; and 値 in the 4 corners of the matrix is 1. . Other size filters or other filters may be used without departing from the scope of the invention. If w (a, b) represents the above-mentioned filter, the filtered transforms R'T3, G'T3, and B'T3 can be calculated according to the formula 5 8-6 2 '. -------- γ ---- Installation -------- Order --------- (Please read the precautions on the back before filling this page) RT3 [x, y ] = 亡 死 W (a, b) (RT2 [X + a, y + bj) 2 •-丨 (60) Printed by the Consumer Cooperative of Intellectual Property Bureau, Ministry of Economic Affairs

1 I RT3 [DC] = 0 〇T3 [x, y]: I; XW (a, b) GT2 [xH.a, y ^ b] 2 a · * ·! B * —1 a * · —I b »— GT3PC]: (61) This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) -54-496080 Α7 Β7 Printed by the Consumers’ Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (52) ZZW (a, b) BT2 [x + a, y + b] 2 B'T3 [x, y] = Bu 1 (v- (62) | ZZW (^ b) I a —lb—l BT3 [DC] = 0 As noted: In each filtered transformation, the 'D c term is set to 0. In addition, components outside the edges of this transformation are considered as having 値 0 °. In this formula, this The treatment is not clearly shown. This filtering operation only affects the size of the frequency vector. The phase of each transformed frequency vector remains unchanged. In Equations 60-62, we use the transformed two-dimensional representation to calculate. To For the purpose of the notation, the fact that the size of the transformation is used is derived from the use of ordinary letters (as opposed to bold letters) as the transformations r, T2, R, T3, G, T2, G 'D3, B in Equations 60-62. 'T2 and B' T 3. This is indicated by the fact that in step 1 7 2 4 The filtered transformation of each section of the channel may be expanded by frequency. Alternatively, if step 17 2 2 is omitted, then the transformation obtained in step 17 20 or the transformation obtained in step 17 12 The transformation can be frequency expanded in steps 1 7 2 4. Steps 1 2 and 4 can also be omitted without departing from the scope of the present invention. Because the real world is composed of similar repeated patterns 'so' a photo The transformation space of an image tends to have any appearance around the DC frequency ', with the same appearance, unless there are duplicate artifacts. For repeated patterns, the frequency characteristics of an image tend to be repeated to the second power. In this step In the lower frequency, the frequency characteristics are used to double the frequency, to make the threshold details easy to be sensitive. In order to perform such frequency expansion, if I use the figure (please read the precautions on the back before filling this page) Binding ----- ^ ---%-This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 public love) -55- 496080 A7 B7 V. The transformation described in (53) 1 8 Two-dimensional version, then the center of this transformation data ~ It is half-expanded into a full-size block of transform data. In other words, the center of the frequency 9 X Ό block is expanded to every other position in the 1 7 X 1 7 transform. When a frequency is in the remainder of the transform, When not explicitly entered, M [J can use the average chirp of adjacent frequencies. Again, this operation only affects the frequency $ size. Existing phase information that is not even at the center half of the transformation, whose frequency has been changed, is retained. After the frequency expansion of the filtered transform is obtained, a new transform can be calculated for each channel in step 1726. This new transformation may depend on the frequency expansion transformation of the channel and the filtered transformation of the channel obtained in step 172. Alternatively, if step 17 2 is omitted, the new transformation depends on the frequency expansion transformation and the transformation obtained in step 1 7 1 2. Equations 6 3-6 5 can be used to calculate this new transformation R 'T4, G, T4 and B' Ding 4. RT4 [X] = V〇.8 RT3 [X] 2 + 0.2R ^ [X] 2 (63) GTT4 [X] = V〇.8GT3 [X] 2 + 0.2G'ET [X] 2 (64) B , T4 [X] = VoJBTSpq2 + 0 · 2ΒΈΤ [Χ] 2 (65) In Equation 63-65, transform R 'ET, G, ET, and B' ET to represent the frequency expansion obtained in step 1 7 2 4 The estimated transformation. Again, the operations performed by Equations 6 3-6 5 only affect the size of the frequency vector. Phase information is retained. In step 1728, we can generate a mask 'to reduce the high-frequency terms of the transformation obtained in step 1726. Because the particle trajectory tends to have the main high-frequency component, this mask can be used to make the particle trajectory and the paper size apply the Chinese National Standard (CNS) A4 specification (210 X 297 mm) ----, ---- 1 ---- Equipment— (Please read the phonetic on the back? Matters before filling out this page) Order: Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs-56- 496080 A7 B7 V. Description of Invention / Or other high frequency noise reduction. This mask can be obtained using a gating function represented by Equation 42. Then use Equation 4 3 to get the mask. Equations 4 2 and 4 3 are explained above in connection with FIG. 16. Other mask functions and / or gating functions may be used without departing from the scope of the invention. In addition, this step can be omitted without departing from the scope of the present invention. In step 1730, a noise correction function can be calculated for each section of each channel. In this embodiment, the noise correction function is calculated in response to one of the mask and one of the frequency domain representations of the previously calculated image. For example, the transform obtained in step 1 7 2 6 can be used to derive this noise correction function. The noise correction function can be obtained from Equation 6 6-7 1. B (X, y) in these formulas is derived from Formula 43. CR (x, y): —M〇._f C3) B (x, y) c “1 + '㈣ (Please read the notes on the back before filling this page) (66) CR (x, y) = l + F.atc (x, y) CR > 1 + Fpost (x, y) CG (x, y) = 2Ifei3㈣c〇, 1 +, (x, y) 2 (67) C〇 (x, y) = l + Fgale (x, y) cg > 1 + Fgatc (x, y) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economy CB (X, y) = (1: 0-4CCC3) Β (χ〇〇d (x, y) CB (x, y) = l + F. atc (x, y) 2 cb > l + Wx, y) RTMIXyl ^ G ^ + (1-GR) CR 〇c, y) GTM [ x, y] = GR + (l-GR) C〇 (x, y) BTM [x, y] = GR + (KR) C "x, y) (68) (69) (70) (71)- 57- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 496080 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (55) In formulas 6 6 — 6 8 'The noise correction function is limited to 1 plus the gating function at a particular position in the transformation matrix. Again, the formulas; 6 6 — 6 8 and 6 9 — 1 only affect the magnitude components of the frequency vector. The original phase information can be retained. The user of the present invention may want to leave residual graininess on the improved image according to the present invention. Therefore, the present invention allows the user to specify the grain residual GR that he wants to leave after the image is improved. Equation 6 9-7 1 In calculating the noise correction functions R 'TM, G' TNI, and B 'TM, the desired residual particles are taken into account. In step 172, the noise correction function of each section of each channel is applied to the original transformation of the appropriate section of the appropriate channel. The result is the frequency domain representation of the image after the high-frequency noise of the particle trajectory is suppressed. Using equations 7 2-7 4 gives a frequency representation of the suppressed particles. R, TS [X] = R'T [X] R, TM [X] (72) G, TS [X]: G'T [X] G'TM [X] (73) B'TS [X] = B, T [X] B, TM [X] (74) Again, the operations performed by Equations 7 2-7 4 only affect the size of the frequency vector, while the phase information remains unchanged. In step 1 7 3 4 the image details can be added to the frequency domain representation of each segment of the red channel of the image. As previously noted, light usually reaches the red layer of the film after it passes through the green and blue layers of the color film. Therefore, the details in the red part of the image tend to become blurred (please read the precautions on the back before filling out this page). Binding — Binding ------; — This paper size applies the Chinese National Standard (CNS) A4 specification 210 X 297 mm) 58 496080 A7 __B7 ____ 5. Description of the invention (56) (Please read the notes on the back before filling this page) Yes. In step 1 7 3 4 we can use the information of the green and blue channels to restore such details to the red channel of the image. Note: Steps 1 7 3 4 can be performed on the red channel of any digital image, not just the red channel after applying the steps of Figure 17. Furthermore, steps 1 7 3 4 can be omitted without departing from the scope of the invention. Furthermore, the individual steps 1 7 3 4 may include an embodiment of the invention. Because the green channel of an image tends to have the most image detail, it can be used to add detail to other channels. The technique described in step 1 7 3 4 can also be applied to: digital images obtained by adding detail to a different channel of a different type or a color film that is structurally different from a typical color film. Other methods of improving the details of the red image can be used without departing from the scope of the invention. In this embodiment, Equation 75 can be used to calculate, for each section of the red channel, the modified red channel transform for that section. (RTSPC] RFAPC > RTSPC] = 1 1-RFA [X] + RFA [X] (75), (1-RFA [X]) (0.7GTS [X] + 〇.4B'TS [X] )) The Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs printed the formula 7 5 using the above-mentioned red frequency attenuator associated with step 1 7 1 6. In step 1 7 3 6, the inverse transform of the noise correction transform is R 'TS, G' TS, and B 'TS, where these transforms can be obtained for each segment to obtain the Improved spatial domain representation of images. If 'Image detail has been added' in step 1 7 3 4 then the red frequency transform obtained using Equation 7 5 can be applied. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) _ 59-496080 A7 B7 V. Description of the invention (57) Balanced color channel (please read the precautions on the back before filling this page) After the particle trajectories are normalized, the normalized chirp obtained through the correction array can be used to balance the color content of the digital image, as shown in step 1 0 1 4 of FIG. 3. Alternatively, the colors can be balanced without using the normalized chirp. The original frame of a digital image often contains distortion from the scanner's attempt to improve the digital image and distortion of the original image on the film due to the film being left too long. These distortions can consist of random and evenly distributed color patterns that no longer reflect particles in the color layer of the original film. By using the normalized 颖 of the glomerular trajectory finally obtained in steps 1 0 8 and 1 0 1 0 in FIG. 3, (or by other methods) the present invention can also generate an estimated re-display of the original color 値(Rec ο nstructi ο η). These reappeared puppets can be said to be balanced in the sense that they are more evenly distributed through the normalized digital image than the original digital image. In other words, balancing color content can further improve digital images. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs In order to balance the color content, we find each channel, the darkest pixel intensity in the normalized frame of digital images, that is, the lowest. Also find the brightest pixel intensity of each channel in the normalized frame of a digital image. The degree is the highest frame. The technical improvement is to use the lowest and highest thresholds from a low-pass version of the digital image. A low-pass filter found to be advantageous for this method is a median filter with a radius of usually more than 3 pixels. Alternatively, the radius may be proportional to the square root of the maximum linear dimension of the digital image in pixels. This software then redistributes the pixel intensity 値 to match the special bit depth (Ιμ t depth) (intensity -60- this paper size applies to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 496080 A7

5. The scope of the invention description (58)), wherein the depth is suitable for the hardware of the special embodiment of the present invention. For example: The following table represents the range of normalized pixel intensities of an estimated reappearance from a digital image: R 3 · 5 ____2 0 · 3 42.6 _ 2 7.910.1 H

GR 4 In order to improve the visible display of these chirps, the following equations can be used for each channel using non-low-pass data: balanced 4 値 maximum = pixel pixels). The balanced example for the entire example is then: R 〇__0 __ 〇

GR .2 5 5 .2 5 5 2 5 5 (Please read the precautions on the back before filling out this page) The consumer cooperation of the Intellectual Property Bureau of the Ministry of Economic Affairs can be used to print or use color improvement or correction between related signals. Other methods of non-linear distortion, which may or may not have particle normalization and / or attenuation. Any such method can be used without departing from the scope of the invention. Figure 19 illustrates a flowchart illustrating the steps of a method that can be used to normalize a plurality of signals having a common component, wherein at least one of the plurality of signals is distorted in a non-linear manner. Although this aspect of the invention can be used to improve the color of a digital image including a scanned photographic image, the method described in Figure 19 can be used in many additional applications. For example, it can be used to correct non-linear distortion in a plurality of signals having a common component. Therefore, it can also be used in applications such as: audio applications, medical imaging applications, satellite imaging applications, and so on. The method described in Figure 19—Order ------ Λ ---. This paper size applies to China National Standard (CNS) A4 (210 X 297 public love) -61-496080 A7 B7 _ V. Description of the invention (59) Can be used for any of these applications' without departing from the scope of the invention. (Please read the precautions on the back before filling this page) In step 1910, the distortion of one of the multiple signals is measured relative to the other signals. A distortion function can generate 'where' the signal is proportional to the distortion of at least one of the signal relative to the remaining signal and the plurality of signals. Although only one measurement can be distorted relative to the remaining correlated signals, it may be advantageous to measure the distortion relative to multiple or all remaining signals. As described above, for a digital image containing three color channels, the distortion of one channel can be measured relative to the other two color channels. In step 1912, an inverse distortion function can be generated based on the distortion measured in step 1910. Such an inverse distortion function can be generated for one, some, or all of a plurality of signals having a common component. In step 1914, a signal can be normalized using the inverse distortion function calculated in step 1912. Again, one, some, or all of the plurality of signals can be normalized according to the inverse distortion function calculated in step 1912. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs In some embodiments of the present invention, it may be necessary to filter one or more of the plurality of signals before determining the distortion function in step 1910. Such filtering is needed when noise and / or other effects may interfere with the calculation of a correct distortion function. For example, a median filter may be useful when removing distortion. May need to be used to correct non-linear distortion in multiple signals, where each signal or multiple signals have been distorted in a non-linear manner, but where the determination of one, some, or all of the signals relative to each other -62- This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7___ 5. Description of the invention (60) Distortion is difficult. For such a correlated signal, the distortion can be corrected by making the small steps of the plurality of signals relatively change by the same amount. For signals that are a function of a single variable, the distortion function calculated in step 19 10 may depend on the slope of one or more of the signals at some point in the image . In this case, the anti-distortion function generated in step 1912 will attempt to equalize the slopes of the plurality of signals in small increments in the signal. For signals whose functions are multiple variables, the distortion function may depend on the rate of change of one, some, or all of the signals in multiple directions at multiple points in the signal. For example, the distortion function calculated in step 19 10 may depend on the gradient at one, some, or all points of the plurality of signals. Among them, a gradient measurement is used To determine the distortion of a signal relative to other signals, a scattergram can be used to generate this distortion function. Each point on the scatter diagram can be obtained by calculating a scatter diagram indicator in response to the magnitude of the gradient of some of the signals on the common point in the signal. A weight can be calculated in response to the angle between these gradients (for example: the cosine of the angle between the gradients). In addition, such weighting can be proportional to the magnitude of the gradient. A curve fitting algorithm can be used to fit a curve to the points on the scatter plot and use this curve to calculate an inverse distortion function. For example: This inverse distortion function may be proportional to the area under the curve. FIG. 20 illustrates a flowchart illustrating a method that can be used to improve a digital image according to the present invention. This aspect of the invention can be used to improve the color of a digital image. Digital image, regardless of whether it is a scanned photograph. The paper size applies the Chinese National Standard (CNS) A4 (210 X 297 mm) -63- —1 ---- • Installation -------- Order ------, — (Please read the notes on the back before filling this page) 496080 A7 B7 V. Description of the invention (61) (Please read the notes on the back before filling this page) Images, satellite images, Medical or other types of digital images may be distorted in a non-linear manner due to distortions in analog images and / or due to distortions that occur when digital images are generated. This distortion may vary with respect to intensity and / or frequency, and may vary differently for each channel. Such non-linear distortion may degrade the image in ways we don't want. For photographic images, the images may look less desirable. For satellite images and medical images, the information provided by the images may be less useful due to such distortions. Printed by a member of the Intellectual Property Bureau of the Ministry of Economic Affairs and a Consumer Cooperative. As mentioned above, many digital images include multiple channels. For example, digital images based on a color photo generally include a red channel, a green channel, and a blue channel. Among them, an image is composed of multiple channels, and the image information contained in each channel generates a common component among the three image channels. Therefore, over a small area in this digital image, the intensity tends to change across the small area in this image by an equal amount. Although the magnitude of the intensity may be quite different, the rate of change of such magnitude tends to be similar for an undistorted image. Therefore, the present invention attempts to determine the distortion of an image according to the intensity change in multiple directions for each channel (or a subset of channels) of an image. An inverse distortion function can be generated for one, some, or all channels in an image, so that when such an inverse distortion function is applied, the intensity of each channel tends to change by the same amount, especially when compared with other When the channels are compared over a small area of the image. In some image areas, when the image is not distorted, compared with other channels, the intensity of one channel tends to change linearly. However, in other images or in the same image area, when the image is not distorted , Compared with other channels -64- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 ____B7 _ V. Description of the invention (62) The intensity of a channel tends to maintain a constant ratio between magnitudes over a small area. In order to automate the improvement of color digital images, the present invention uses the average of these two variations, where the change corresponds to the way a color image tends to be stored digitally. In step 1916, an image i is scanned to produce a digital image. Such scanning can occur using the scanner 30 or the scanner 34 described above. Alternatively, steps 1 9 1 6 can be omitted and the invention is applied to an image that has been previously scanned. In addition, the present invention can be applied to digital images having a plurality of channels, wherein each channel has a common signal component that distorts one or more channels in a non-linear manner. The term “common components” broadly refers to signals that are in some way interrelated. Alternatively, before performing the remaining steps in the method described in FIG. 20, the particle normalization aspect of the present invention discussed above may be applied to a digital image. Alternatively, before performing the remaining steps in the method described in FIG. 20, the one or more particle reduction techniques described above may also be applied to the image. Alternatively, the method described in FIG. 20 may be applied to the image without any particle normalization and / or particle reduction, or different techniques may be used after normalization or reduction. In step 119, each channel of the image (or a subset of the channels of the image) can be filtered. Therefore, various effects, such as noise and image resolution effects, will not be distorted with nonlinearity. The removal of the interference. In a small number of pixels, noise caused by particles such as ... may interfere with the removal of non-linear distortion. In addition, because the light used to produce a color photographic image passes through most layers of the film, 'the paper size of a digital image applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -65 · " (Please read the notes on the back before filling this page)

n n n n One: ° 4 ·· ϋ n an ft n n I 496080 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (63) or multiple channels may be distorted. As mentioned above, the red layer of a color film tends to receive light after it passes through the green and blue layers of the film. So the image in the red layer of the film tends to be slightly more distorted than in the green and blue layers. For example: the image may be less clear on the red channel. Such optical distortions may be related to the removal of non-linear distortions in the colors of the image. Therefore, in step 1918, each channel of the image can be blurred in advance by filtering the image (p r e-b 1 u r). In this embodiment, a median filter with a radius of 10 pixels can be used for a film scanned at 2000 dpi per color channel. Different radii 'can be used without departing from the scope of the invention. In addition, different scanning resolutions can be used without departing from the scope of the invention. In addition to applying a median filter, other methods can be used to pre-blur the image. Basically every channel of the image can be changed, thereby reducing or eliminating the noise and resolution effects as described above. In step 1920, 'at least two channels of the logarithmic image', we can generate a vector based on the intensity change rates (slopes) in multiple directions at most points of the image. In this embodiment, although such a vector is generated for each of the plurality of channels, it may be generated for a subset of the plurality of channels without departing from the scope of the present invention. In this embodiment, the slope can be calculated on all points of the image or on a subset of the points of the image to generate a vector without departing from the scope of the present invention. In this embodiment, the generated vector represents a particular image channel, a gradient in intensity at the point in question. In other words, this vector includes: (Please read the precautions on the back before filling out this page) This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) -66-496080 A7 B7 Employees of Intellectual Property Bureau, Ministry of Economic Affairs Printed by the Consumer Cooperative. 5. Description of the Invention (64) At a particular point in the image channel, the first component representing the intensity slope in the X-direction and the second component representing the intensity slope in the y-direction. Vectors that are proportional to the intensity slopes at other points in the image can also be selected without departing from the scope of the invention. The slopes in the X and y directions are chosen for convenience only. We can use the formulas 7 6-7 8 to calculate the vectors of the red, green and blue channels at a particular point (X, y), which are: R ”2 (X, y), G” 2 (X , Y) and B ″ 2 (x, y), where i and j represent unit vectors in the X and y directions, respectively. R, 2〇c, y) = 〇r (x + 1, y) -IT〇 cl, y)) i + (B / (x, y + l) -F / (# x, yl)) j (76) G, 2〇c, y) = (G, (x +1, y) -G \ x -1, y)) i + (G, (x, y +1)-G, 〇c, y -1)) j (77) B ^ 2 (x, y) =: (B \ x +1, y)-Bn (x -1, y)) i + (B \ xy y +1)-B ^ (x, y -1)) j (78) In these formulas, R ”·, G ”and B” represent the filtered intensity 値 of the three image channels. Therefore, 'a vector can be calculated for each point in each channel of a digital image. For images with more or fewer channels, a similar formula can be used for each channel. We can also calculate the vector only for the subset of points in each channel without departing from the scope of the present invention. Equations 7 6-7 8 do not calculate the slope calculated at the edges of the image channel. On such edges, different slope measurements can be used or these points can be ignored without departing from the scope of the invention. Likewise, different slope calculation methods can be used without departing from the scope of the present invention. This embodiment uses the adjacent left and right points of the pixel in question 'instead of the pixel itself' to calculate the approximate slope of the $ -direction at that point. A similar method can be used in the y-direction. Alternatively, you can use more in the χ one or y direction. Nn H n ϋ n «l— n ϋ · nnnnnn — 1 one 01 I n ·· ϋ · 注意 nnnn I (Please read the notes on the back before filling in 冩(This page) This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -67- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 496080 A7 B7____ 5. The pixel of invention description (65) X-direction or y-direction slope without departing from the scope of the present invention. In addition, a certain average of the approximate slope can be used for a particular pixel without departing from the scope of the present invention. In step 192, we can compare vectors from a plurality of channels to generate scatter plot data, where the data produces a distortion function that compares the distortion of one channel representing the image with the other channels of the image. it works. Again, such scatter plot data can be generated for one, some, or all channels without departing from the scope of the present invention. In addition, the comparison of vectors is done to generate other channels of scatter plot data that may involve some or all of the imagery. In this embodiment, the scatter plot data depends on the vector calculated in step 1920. Because undistorted images tend to have gradient vectors with similar directions, a weight can be calculated for each point on the scatter plot, where the point between this addition and the vector calculated in step 1920 The product is proportional. In this embodiment, such a point product is calculated for a particular channel, and for each of the other channels, and each such point product may be one point in the day. For example, for the red scatter diagram, we draw a point with a weight, where the weight is proportional to the point product of the vector calculated for the red channel and the green channel in step 1920. In addition, a weight is calculated for each point where the red channel is to be placed on the scatter diagram, where the weight is equal to the point of the vector at that point calculated for the red channel and the blue channel in step 1920. The product is proportional. Therefore, the weighting of a particular point on the scatter plot can be proportional to the similarity between the intensity slopes of the two channels at the particular point. Because the channels are distorted relative to each other, this embodiment of the present invention determines a scatter plot point for a particular channel, where this point applies the Chinese paper standard of the channel in question to the Chinese National Standard (CNS) A4 specification (210 X 297 public love) 68 (Please read the notes on the back before filling out this page) ------- I ^ ----------% · 496080 A7 B7 V. Description of the invention (66) The slope at a particular point is compared to the slope of the intensity at the corresponding point of each additional channel. Alternatively, a subset of the channels may be compared without departing from the scope of the invention. The present invention utilizes that the relative magnitude of the slope vector at a particular point will also be generated for a particular channel in step 1920, and at a corresponding point with another channel of the image in step 1920. The magnitudes of the calculated slope vectors are compared to determine an index of the scatter plot. Another indicator of the scatter plot includes the intensity of a point in the image channel. The points of the scatter plot can be generated as follows. Our image will be distorted and therefore the correction will be in the red wave, our image. Scatter chart The axis index produces a scatter chart for each channel of the image. In this example, a scattered X-axis index is generated for the channel, the green channel, and the blue channel for the pixel intensity in question. The 値 between y 1 and -1 of the scatter plot, where the 値 is obtained by comparing the relative magnitude of the slope vector at a point corresponding to a particular channel with the size of a corresponding vector at another channel of the image. A description assigned to a particular point. Therefore, you can use formulas 7 9-9 3 to locate special points on the red, scatter plot. In the case of Equations 8 5-9 0, the weighted chirp can be set to a predetermined chirp like zero. The slope of the points is weighted as above. Negative results of green or blue (please read the precautions on the back before filling this page)

· «II II in · ϋ n I 一 -eJ ·-« ϋ nn I--I Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 2 k-2R-2 (xy) G * 2 (xy) R " 2 ) + | G " 2 | j · 2 | R ^ 2 | 2 | G ^ 2 |, (79) (80) (81) This paper size applies to China National Standard (CNS) A4 (210 x 297 mm) .69-496080 Α7 Β7 V. Description of the invention (67) (1 2 丨 G, 2 | 1 Y2Gi (x.y >-L \ | G # 2 | + / 1 2 | B, 2 |) Y 丨 B. 2 (xy)-1 \ | R, 2 Bu | B2 |〉 ft 2 | B " 2 |) V — 1 2B ^ 2 (xy) 1 | B, 2 + | G, 2 | \ 111 \ J IT2G. 2 ^ lR * 2 (xy) 秦 〃2 || G〃2 | W_______ R, 2B'2 (82) (83) (84) (85) (85) (Please read the notes on the back before filling this page )

WlG-2 (Xty) W- 2G * 2 (x * y) GH2VF2p2 | B, 2.G, 2V | G # 2 || B ^ 2 | BH2V 「B ， 2 " R, 2 丨 B " 2 (87 ) (88) (89) 2B-2 (x, y) > taste 2 | ^ Rix, y) '= R'x, y) (91) χ (3 ( χ · χ) = G " (x, y) (92) ^ B (xy) = B '(x, y) (93) Formula 9 1 indicates: by determining red at a particular point: color wave Channel ^ g to determine the X-axis index of the special point scatter plot. Draw two points for the special point of the red channel. Each point will have the same χ-one index on the red scatter plot. For the two In terms of points, one point corresponds to this special point. The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -70-496080 A7 B7. 5. Description of the invention (68) The slope of the red and green channels Comparison of vectors, and another point corresponds to the comparison between the red and blue vectors at this particular point. For points representing the comparison of red and green vectors, Equation 7 9 can be used to determine the y-index of the scatter plot. For red Compared with the blue vector at special points, you can use the formula 8 0 to determine the scatter plot. The y-index of this point. Comparing the weight of the red channel to the points of the green channel on the scatter plot can be calculated using Equation 8 5 and the weight of the points representing the comparison between the red and blue channels can be calculated using Equation 8. 6 of decision 0 For any point on the green channel, two points of the green scatter diagram can also be generated. In each case, the X-axis index of the green scatter diagram is determined using Equation 9 2. For the corresponding The first point of the comparison between the vectors of the green channel and the red channel, Equation 8 1 is used to determine the y-axis index of the point on the green scatter plot, while Equation 8 7 represents the point at which the point is drawn. For the point of comparison between the vectors corresponding to the green and blue channels at this particular point, Equation 8 2 is used to determine the y-axis index 'and Equation 8 8 is used to determine the special point. For any point on the blue channel, two points of the blue scatter diagram can also be generated. In each case, the X-axis index of the blue scatter diagram is determined using Equation 9 3 The first point of the comparison between the vectors corresponding to the blue and red channels is 8 3 is used to determine the y-axis index of the point on the blue scatter chart, and formula 8 9 represents the weighting by day out at this point. The corresponding blue and green channels at this particular point The point of comparison between the vectors, formula 8 4 is used to determine the y-axis index 'and formula 9 0 is used to determine the weighting of this particular point. This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) (Please read the precautions on the back before filling out this page) -Installation -------- Order ------ J ---, Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs- 71-496080 A7 B7 I 丨 丨 · * 丨 丨 " — V. Description of the invention (69) f Please read the note on the back first and then fill out this page) In step 1 9 2 4 we according to step 1 9 The scatter plot data produced in 2 produces a scatter plot for each of the plurality of channels. These steps can be combined without departing from the scope of the invention. In addition, if only one or some of the plurality of channels will be corrected for distortion, only a scattergram will be generated for those channels to be corrected. Again, as described above, the scatter plot generated in this embodiment of the present invention includes points that are directly proportional to the distortion of the channel, where the scatter plot is generated for each of the other channels. However, distortions relative to a subset of the extra channels can be used without departing from the scope of the invention. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Figure 21 illustrates six points that can be drawn corresponding to a single spatial location in the image. For the red scatter plot, the pixel intensity in question determines the X-axis index of each of the two points, 1946. The y-axis index of the point 1934 is determined by Equation 79. The weight assigned to point 1 9 3 4 is determined by Formula 85. Here, Formula 79 yields a y-axis index of 0.5. Since the sinus of formula 81 is the negative of formula 79, the y-index corresponding to the point 1 9 3 6 on the green scatter chart is a y-index of 0.5. The X-point at 1 9 3 6 is determined using Equation 9 2 and represents the intensity of the pixel in question in the green channel. The weighting of points 1 9 3 6 generated by Equation 8 7 will be the negative weighting of points 1 9 3 4 generated by Equation 8 5. Therefore, various implementations of the present invention may use only one set of formulas, such as formulas 8, 5, 8 and 88, to calculate the weighting without departing from the scope of the invention. Again, point 1 9 3 8 has an X-index on the pixel in question, determined by Equation 91. The y-index is determined by the formula 80. 72 The paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 496080

Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of invention (π) and its weight is determined by formula 86. This point provides a measure of the distortion of the red signal when compared to the blue signal. On the blue scatter chart, y of point 1 · 9 4 0-the index is y of point 1 9 3 8-the negative number of the index is positive, as Equation 8 3 is only a negative number of Equation 80. The X-index 1 950 of the point 1 9 0 0 corresponds to the intensity 値 of the pixel in question in the blue channel, and is determined using Equation 9 3. Similarly, points 1 9 4 2 on the green scatter diagram and points 1 9 4 4 on the monitor scatter diagram have y-indexes (negative additions) that are negative to each other, as Equation 84 is Equation 82 negative number. After generating the scatter diagram in step 19 2 4, a curve fitting algorithm can be used to generate a curve through the scatter diagram, as illustrated in step 1 9 2 6. In this embodiment, we use an average type of curve fitting algorithm. Any curve-fitting algorithm can be used without departing from the scope of the invention. A curve can be generated for each scatter plot. If distortion correction is applied to the channels produced by the scattergram, it is assumed that only a scattergram will be generated in step 192 2. If no distortion correction is generated for a particular channel, both the generation of the scatter diagram in step 19 24 and the generation of the curve in step 19 26 can be omitted for this channel. In step 1928, an inverse distortion function can be generated based on the curve generated in step 1926. This inverse distortion function can be any function that is proportional to the curve produced in step 1 9 2 6. The curve produced in step 1 9 2 6 is proportional to the relative distortion of a particular channel to one or more additional signals having a common component. In this embodiment, the inverse distortion function of each channel can be generated using equations 9 4-9 6. This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 mm) -73-(Please read the precautions on the back before filling this page) Loading -------- Order ----- -* --- · 496080 A7 ________B7___ 5. Description of the invention (71) ΓΚ [€] = Σήί ^ ψ (94) «· -〇ya — -Kj (Please read the notes on the back before filling this page) fg, Meal ... Gang (⑹ Formula 9 4 can be used in the red channel 'Formula 9 5 is used in the green channel, while Formula 9 6 is used in the blue channel i: ° Because the digital image is involved, so 'A separate number of intensity 可以 can be used. For example: this embodiment uses an 8-bit intensity 値' which can produce 2 5 6 different possible strengths 因此. Therefore 'this anti-distortion function includes 2 5 6 値A look-up table with a range of 0-2 5 5. Therefore, in formulas 94 to 96, "C" has a maximum value of 255, of which digital images have 256 possible intensities. Other numbers of bits Can be used 'without departing from the scope of the present invention. "Y" in Equations 9 4-9 6 represents the number of channels that make up this digital image. Equation 9 4 R 1 represents on the red channel, on s, the X-indicator is “1” 'y' printed in step 1 9 2 6 by the consumer property cooperative of the Intellectual Property Bureau of the Ministry of Economics to print the production curve. In other words, in the step In 1 9 2 6, the curve generated by the scatter plot is the distortion drawn in response to the intensity of the pixel. Equation 9 4 calculates an inverse distortion function for each pixel intensity in question, where the function is the same as the intensity in question. Previously, the area under the curve generated in step 1 9 2 6 is directly proportional. Equations 9 5 and 9 6 are similar to equation 9 4. Use y / y — 1 square root in equation 9 4-9 6 because it is not used This may lead to over-correction. The correction of each channel is based on its distortion relative to other channels. If a correction is made without considering this factor, it is assumed that the paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -74-'496080 A7 B7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Invention Description (72) As long as one channel is corrected, relative correction can be made for each channel. In this example Because every channel has to be corrected Therefore, if the proper square root is not taken, it will cause a net overcorrection. In step 1930, the inverse distortion function determined in step 1928 can be used to normalize each image channel. For normalization, the intensity of each pixel in the image is processed. For a particular spatial location, the intensity of the red channel at that pixel is used as an indicator, and the intensity of red is changed to the formula 9 The resulting inverse distortion function. This red intensity 値 is replaced by the intensity supplied by the inverse distortion function. Similarly, the green intensity 値 of a particular pixel is replaced by the green intensity 产生 generated by the inverse distortion function generated for this green channel in step 192, where the inverse distortion function of this green channel is The indicators include the current pixel intensity of this green channel. Finally, using the blue inverse distortion function generated in step 192, the pixel's blue channel intensity 値 is similarly corrected. Although the invention described herein deals with calculations in the spatial and frequency domains, it is also possible to use calculations in the opposite domain fairly without departing from the scope of the invention. Although I have described the present invention in detail, it is understood that various changes, substitutions and substitutions can be made hereto without departing from the spirit and scope of the present invention as defined by the appended claims. In order to help the patent office and any readers who have published patents on this application explain the scope of the patent application attached to this point, patent applicants wish to note that they do not intend that any of the attached patent application scopes solicit US patent law U · S. C ·) § 1 1 2 3 3 6 6 is archived (Please read the precautions on the back before filling this page) This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) -75- 496080 A7 _B7_ V. Description of the invention (73) Date exists, unless the word "means for" ("means for ") or" step of "" stepf 〇r ") Used in special patent applications. (Please read the notes on the back before filling out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs

-76- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)

Claims (1)

wyear I ^% t to find out if there are Mm Weng: / i GT mentioned whether the $ is allowed to be amended. Printed by the Consumers' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs? Xr -f " Sigh No Supplement A, Patent Application Scope Annex I A: No. 89 1079 No. 19 Patent Application Chinese Application Patent Scope Amendment January 1991 Amendment 1 · A method for normalizing a plurality of signals 'wherein each of the plurality of signals has a common component' and wherein at least one of the plurality of signals has been distorted in a non-linear manner Including: For at least one of the plurality of signals, determining a distortion function, wherein a distortion function of a particular one of the plurality of signals is related to at least one of the remaining signals in the plurality of signals. The distortion of the signal is proportional to the distortion of the signal. 'Generates a relatively reverse distortion function for at least one of the plurality of signals, where the relative reverse distortion function generated by a particular signal in the plurality of signals is The response is a distortion function determined by the particular signal; the relative anti-distortion generated by applying one of the plurality of signals by applying Number and at least one normalized in the plurality of signals. 2. The method of claim 1 further comprising: filtering at least one of the plurality of signals before determining the distortion function for at least one of the plurality of signals. 3. The method according to item 2 of the scope of patent application, wherein we use a median filter to filter at least one of the plurality of signals. This paper uses the Chinese National Standard (CNS) A4 size (210 X 297 cm) n! |-........ Nil— nn ..... n I-I ϋ (Please read first Note on the back then fill out this page) Order 496080 A8 B8 C8 D8 Γ, patent application scope 4 · The method of the first patent application scope, wherein each of the plurality of signals is a function of a single variable, and wherein the distortion function of at least one of the plurality of signals depends on The slope of at least one of the plurality of signals at a plurality of points. 5. The method of claim 1, wherein each of the plurality of signals is a function of a plurality of variables, and wherein the distortion function of at least one of the plurality of signals depends on a plurality of points The rate of change of at least one of the plurality of signals in multiple directions. 6. The method of claim 1, wherein the distortion function of at least one of the plurality of signals depends on the gradient of at least one of the plurality of signals at a plurality of points. 7. The method according to item 1 of the scope of patent application, wherein each of the plurality of signals is a discrete function; wherein the distortion function of at least one of the plurality of signals is represented by a band diagram (Hi s togi * am), and a first strip map point associated with a first point in the plurality of signals is determined as follows: Calculate the number of signals including at least one of the plurality of signals A first gradient that is one of the gradients at the first point; calculate a second gradient that includes at least one other of the plurality of signals at the first point; a second gradient that is a response to the first gradient and a second gradient Strip chart indicators; Calculate the cosine of the angle between the first and second gradients and the first and second (please read the precautions on the back before filling out this page) The paper size of the paper is applicable to China National Standard (CNS) A4 (210X297 mm) -2- 496080 A8 B8 C8 D8 VI. One of the two gradients of the patent application scope. The first weighting; (Please read the notes on the back before filling this page) Decide the first strip chart point in response to the first strip chart index, and A second strip chart index where at least one of the plurality of signals is proportional to the size of the first point, and wherein the weight of the first strip chart point is proportional to the first plus. 8. The method according to item 7 of the scope of patent application, further comprising: using a curve fitting algorithm to curve fit the points on the strip chart. 0. The method according to item 8 of the scope of patent application, wherein the The inverse distortion function of at least one of the plurality of signals depends on the area under the curve. The method according to item 1 of the patent application range, wherein the distortion function of at least one of the plurality of signals depends on A strip chart containing a series of points, wherein each point of the strip chart depends on the gradient of at least one of the plurality of signals at a particular point and the gradient of another one of the plurality of signals at the particular point. 1 1. The method of item 10 of the scope of patent application, further comprising: Printing by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, using a curve fitting algorithm to curve fit the points on the strip chart. For example, the method of claim 11 in the scope of patent application further includes: wherein the inverse distortion function corresponds to the area _ under the curve. 13. The method according to item 12 of the scope of patent application, further comprising: 'using a median filter' to filter the at least one of the plurality of signals before determining the distortion function of the at least one of the plurality of signals. One paper size is applicable to Chinese National Standard (CNS) A4 specification (210X297 mm) 496080 A8 B8 C8 D8 91 · L 22. The year and month of the paint are not covered, and the scope of patent application signals. 1 4. If the method of the first item of patent application scope, further includes: (Please read the notes on the back before filling this page) Determine a distortion function for each of the plurality of signals, among which the plurality of signals The distortion function of a particular signal in is proportional to the distortion of a particular signal related to at least one of the remaining signals in the plurality of signals; and a relative inverse distortion function is generated for each of the plurality of signals, where The relative inverse distortion function of a particular signal in the plurality of signals is a distortion function determined in response to the particular signal. 15. A method for improving a digital image, comprising: determining a distortion function for at least one of a plurality of channels of the digital image, wherein a distortion function of a particular channel in the plurality of channels and Is proportional to the distortion of a particular channel in relation to at least one of the remaining channels in the plurality of channels (or in the case where the digital image contains two channels, related to the remaining other channel) ; And the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs prints at least one of the plurality of channels of the digital image to generate a relative inverse distortion function, wherein a phase of a particular one of the plurality of channels is in phase. The anti-distortion function is a response to the distortion function determined for the particular channel:: ',, V The relative anti-distortion function i applied to the first of the plurality of channels is applied to the plurality of channels. The at least one in the channel is normalized. 16. The method according to item 15 of the scope of patent application, further comprising: before determining a distortion function of the at least one of the plurality of channels, filtering the at least one channel of the plurality of channels. -4- This paper size applies to Chinese National Standard (CNS) A4 specification (210X297 mm) 496080 A8 B8 C8 D8 Supplementary VI. Patent application scope 17 • The method of item 16 of the patent application scope ‘where’ uses a median filter to filter at least one of the plurality of channels. (Read the precautions on the back before you fill out this page) 1 8 · If you are applying for the method of item 15 of the patent scope, where the distortion function of the at least one of the plurality of channels depends on the complex number at multiple points The rate of change of the at least one of the channels in multiple directions. 19 · The method of claim 15 in the scope of the patent application, wherein the distortion function of at least one of the plurality of channels depends on the gradient of at least one of the plurality of channels at a plurality of points. 2 0. The method according to item 19 of the patent application range, wherein / at least one of the plurality of channels includes a plurality of pixels on a two-degree spatial plane, and wherein a distortion function of a first pixel is The gradient depends on the slope of the approximation calculated by using another pixel in the vicinity of the first pixel. 2 1 · The method according to item 19 of the scope of patent application, wherein the distortion function of at least one of the plurality of channels depends more on the gradient of at least one of the plurality of channels at a plurality of points. 2 2 · According to the method of claim 15 in the scope of patent application, which is printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, each of the plurality of channels is a discrete function; wherein, the number of At least one distortion function is obtained from a band diagram, and a first band diagram point related to a plurality of first points of a plurality of channels is determined as follows: The calculation includes the plurality of One of the at least one gradient of the channel at the first point is a first gradient; the paper size including the gradient of the at least one of the plurality of channels at the first point is calculated according to the Chinese National Standard (CNS) A4 specification (210X297) (Centi) 91, 496080 A8 B8 C8 D8 Correct X |, Ή VI. Patent application scope-second gradient; (Please read the notes on the back before filling this page) Calculate one of the first and second gradients of the response Index of the first strip chart; Calculate the first weighting of the cosine of the angle between the first and second gradients and the magnitude of the first and second gradients; determine the first strip chart point to respond to the first FIG index strips, and the number of multiplexed channels in the at least one indicator in a second proportional band diagram of a first point size, and wherein the first weight to the first weight bands is proportional to the point of FIG. _ 2 3. If the method of item 22 of the scope of patent application, further includes: using a curve fitting algorithm to curve fit the points on the strip chart 0 2 4. As of item 23 of the scope of patent application A method, wherein an inverse distortion function of at least one of the plurality of channels depends on an area under the curve. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 2 5. The method of claim 15 in the scope of patent application, wherein the distortion function of at least one of the plurality of channels depends on a band diagram containing a series of points, where Each point of the strip chart depends on the gradient of at least one of the plurality of channels at a particular point and the gradient of another one or more of the plurality of channels at the particular point. 2 6. The method according to item 25 of the scope of patent application, further comprising: using a curve fitting algorithm to curve fit the points on the strip chart. 02 7. The method according to item 26 of the scope of patent application Among them, the size of this copy is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) I " 496080 6. Scope of patent application The inverse distortion function of at least one of several channels depends on the area under the curve. (Please read the notes on the back before filling out this page) 2 8 · If the method of the scope of patent application No. 27, further includes: Before determining the distortion function of at least one of the plurality of channels, use a median A filter that filters the at least one channel of the plurality of channels. 29. The method according to item 15 of the scope of patent application, further comprising: determining a distortion function for each of the plurality of channels, wherein a distortion function of a particular one of the plurality of channels and a The distortion of at least one associated special channel of the remaining channels in the plurality of channels is directly proportional; and a relative inverse distortion function is generated for each of the plurality of channels, where the The relative inverse distortion function of a particular channel is a distortion function determined in response to that particular channel. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 3. The method of item 29 in the scope of patent application, wherein the distortion function of each of the plurality of channels is a special channel, depending on the A strip chart of points, wherein each point of the strip chart depends on the gradient of at least one of the plurality of channels at a particular point and the gradient of another one or more of the plurality of channels at the particular point. 3 1 · The method of item 30 in the scope of patent application, further comprising: using a curve fitting algorithm to fit a curve on the points on the strip chart. 03 2 · The method of item 31 in the scope of patent application Among them, the inverse distortion function of each of the plurality of channels is a special channel, taking -7- This paper size applies to China National Standard (CNS) A4 specification (210X 297 mm) A8 B8 C8 D8 496080 6. The scope of patent application depends on the area under the curve associated with the particular channel. 3 3 · If the method in the scope of patent application No. 32, further includes: (Please read the notes on the back before filling this page) Before determining the distortion function of each of the plurality of channels, use a median A filter that filters each of the plurality of channels 0 3 4 · — a digital image scanning system, including: scanning hardware operable to scan a photographic image and convert the photographic image into a digital image; And computer software, associated with the scanning hardware, and operable to "determine a distortion function for at least one of the plurality of channels of the digital image, wherein a particular one of the plurality of channels The distortion function is related to at least one of the remaining channels in the plurality of channels (or to the remaining channel if the digital image contains two channels). Distortion is directly proportional; at least one of the plurality of channels of the digital image printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs produces a relative anti-distortion function, wherein, the The relative inverse distortion function of a particular channel among several channels is in response to the distortion function determined for that particular channel ϋ 'V' y ,: A is generated by applying the first of the plurality of channels Relative inverse and distortion ^ " number, normalize the at least one of the plurality of channels. 3 5 · For example, the digital image scanning system of item 34 of the scope of patent application, wherein the computer software is more The operation is used to: filter the at least one channel in the plurality of channels before determining the distortion function of the at least one of the plurality of channels. This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) ) -8- 496080 A8 B8 C8 D8 WL j rc j.: Sj mm m --- · ι-fine, ii J 6. Application scope of patent (please read the precautions on the back before filling this page) 3 6 · If The digital image scanning system of claim 34, wherein the distortion function of the at least one of the plurality of channels depends on the at least one of the plurality of channels at a plurality of points in a plurality of directions. Rate of change. 3 7 · 如The digital image scanning system according to claim 34, wherein each of the plurality of channels is a discrete function; wherein the distortion function of the at least one of the plurality of channels is obtained from a band diagram Obtained by the computer software, and wherein a first strip pattern point associated with a first point of a plurality of channels is determined as follows: the calculation includes the at least one of the plurality of channels A first gradient of the gradient at the first point; calculating a second gradient that includes the gradient of the other of the plurality of channels at the first point; calculating a first response to the magnitude of the first gradient and the second gradient Strip chart indicators; printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs to calculate the first weighting of the cosine of the angle between the first and second gradients and the magnitude of the first and second gradients; determine the points of the first strip chart To respond to the first strip chart indicator, a second strip chart indicator that is proportional to the size of the at least one of the plurality of channels at the first point, and wherein the weight of the first strip chart point is weighted A first proportional plus Quan. 3 8. If the digital image scanning system in item 3 of the scope of patent application is applied, the paper size is applicable to China National Standard (CNS) A4 specification (210X297 mm) -9- 496〇8〇 A8 B8 C8 D8 6. The scope of patent application, among which, the computer software is further operable to: use a curve fitting algorithm to draw a curve on the points on the strip chart. △ 口 3 9. Digital system, wherein the inverse of the at least one of the plurality of channels depends on the area under the curve. Including: the image scanning system takes a distortion function of 40. — a digital image processing system a computer-readable storage medium; computer software is stored in the computer-readable storage medium and is operable to: Each channel of a plurality of channels of a digital image determines a distortion function, wherein a distortion function of a particular channel in the plurality of channels is related to each of the remaining channels in the plurality of channels. The distortion of a particular channel is directly proportional to the anti-distortion function of the plurality of channels of the digital image, wherein each of the plurality of channels pairs the anti-distortion function and each wave of distortion determined by the particular channel. The phase function of a particular channel in a channel is directly proportional (please read the precautions on the back before filling out this page). The employee consumer cooperative of the Intellectual Property Bureau of the Ministry of Economy The relative inverse distortion function produced by that one is normalized to each of the complexes. Among the plurality of channels, 41. For example, the digital image processing system of item 40 in the patent application range, wherein the computer software is further operable to: determine the distortion of the at least one of the plurality of channels Prior to the function, this paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) -10- 496080 A8 B8 C8 D8 9L Llid The at least one wave in the road. 4 2. The digital image processing system according to item 40 of the patent application, wherein the distortion function of the at least one of the plurality of channels depends on the at least one of the plurality of channels at a plurality of points. Rate of change in multiple directions. 43. The digital image processing system according to item 40 of the scope of patent application: wherein each of the plurality of channels is a discrete function; and wherein the distortion function of the at least one of the plurality of channels is It is obtained from the computer software from a band diagram, and a first band diagram point related to a plurality of first points in a plurality of channels is determined as follows: The calculation includes the plurality of A first gradient of the at least one gradient at the first point in the channel; calculating a second gradient including the gradient of the other of the plurality of channels at the first point; calculating a response to the first gradient and the second gradient Of the first strip chart index; calculate the first weighting of the cosine of the angle between the first and second gradients and one of the magnitudes of the first and second gradients; and determine the first strip chart point 'in response A first strip chart index, a second strip chart index that is proportional to the size of the at least one of the plurality of channels at a first point, and wherein the weighting of the first strip chart point is equal to the first weighting Directly proportional. (Please read the precautions on the back before filling out this page) The paper size applies to China National Standard (CNS) A4 (210X297 mm) -11- 496080 々, patent application range 4 4 · If the digital image processing system of item 43 of the patent application range, in which the computer software is more operable to: use a curve fitting algorithm to point on the strip chart Make a curve fit. 4 5. The digital image processing system according to item 44 of the scope of patent application, wherein the inverse distortion function of the at least one of the plurality of channels depends on the area under the curve. 4 6. A method for improving a digital image, the digital image including at least two color channels, the method comprising: generating a first color channel transform function to be applied to a digital image; The first color channel, the first color channel transform function is selected so that when it is applied to the first color channel, the first color channel and a second color channel of the digital image will be in the The digital image is changed across a small distance by approximately equal amounts; and the first color channel is modified to respond to the first color channel transform function. 4 7 · If the method of patent application range 46 items, it also includes too thick first color channel. 48. The method according to item 47 of the scope of patent application, wherein the first color channel transform function depends on the gradient of the first color channel at different points after the first color channel is filtered. 49. The method of claim 46, wherein the digital image includes a first color channel, a second color channel, and a third color channel. The method includes: generating a first color channel; A two-color channel transform function to be applied to the second color channel of the digital image. The second color channel transform function is selected so that this paper scale applies the Chinese National Standard (CNS) A4 specification (210X29? Cm). (Please read the notes on the back before filling this page)-Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs-12- 496080 A8 B8 C8 D8 91. 1. 22 (Please note this page, please fill in this page) When it is applied to the second color channel, the first color channel, the second color channel and the third color channel of the digital image will be in the digital image. Change over a small distance by approximately equal amounts; generate a third color channel transform function to apply to the third color channel of the digital image, the third color channel transform function is selected so that it be utilized When the third color channel is reached, the first color channel, the second color channel, and the third color channel of the digital image will change across the small distance by approximately equal amounts in the digital image; modify The second color channel is modified in response to the second color channel transformation function to modify the third color channel to respond to the third color channel transformation function; and wherein the first color channel transformation function is selected so that it is When applied to the first color channel, the first color channel, the second color channel, and the third color channel of the digital image will change across the small distance by approximately equal amounts in the digital image; 50. The method according to item 49 of the scope of patent application, further comprising: printing by the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs before generating a color channel transformation function associated with each channel, filtering the first color wave Channel, second color channel, and third color channel. 51. The method of claim 50 in the scope of patent application, wherein each of the first, second, and third color channel transformation functions depends on the first, second, and third color channels at a plurality of points. The rate of change in most directions is 0 5 2. The method of item 50 of the patent application scope further includes: Considering the first color channel, this paper size is applicable to China National Standard (CNS) A4 (210X297) 496080 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs A8 B8 C8 D8 Each color channel of the transformation function includes the following: The computer-readable image system is obtained from a number and the digital shadow function is selected so that the first color image of the image is approximately equal to the first color image (please first (Please read the notes on the back and fill in this page) ^, filter the second color channel; and filter the third color channel; where the first, second, and third color channels depend on most points The rate of change in the first, second and third color directions. 5 3. — A changed digital image system, a computer-readable storage medium; a changed digital image, stored on the storage medium, where the changed digital image is a digital image ' By: generating a first color channel transform function to correspond to the first color channel, the first color channel is transformed when the digital color channel and a digital color channel are applied to the first color channel; The second color channel will change after the amount of the digital image spans a small distance; and the digital image is modified to reflect the first color channel transformation function 5 4. After the change in item 53 of the scope of patent application Digital image system, wherein the changed digital image is further generated by filtering the first color channel. 5 5 · According to the digital image system after the change in the scope of the patent application No. 54, wherein the first color channel transformation function depends on the different points at different points after the first color channel has been filtered. The gradient of the first color channel. This paper size applies to Chinese National Standard (CNS) A4 (210X297 mm) -14- 496080 9L A8 B8 C8 D8 L (Please read the notes on the back before filling this page) image system, where the digital image includes a first color channel, a second color channel, and a third color channel, and after the change The digital image is produced by: generating a second color channel transform function to apply to the second color channel of the digital image, the second color channel transform function is selected so that When applied to a second color channel, the first color channel, the second color channel, and the third color channel of the digital image will change across a small distance in the digital image by approximately equal amounts. ; Generating a third color channel transform function to apply to the third color channel of the digital image, the third color channel transform function is selected so that when it is applied to the third color channel, The digital The first color channel, the second color channel, and the third color channel of the image will change across a small distance by approximately equal amounts in the digital image; modify the second color channel to respond to the second color Channel transform function; Modify the third color channel to respond to the third color channel transform function; and printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, the first color channel transform function is selected so that it is When applied to the first color channel, the first color channel, the second color channel, and the third color channel of the digital image will change across the small distance by approximately equal amounts in the digital image. -15- This paper size is applicable to Chinese National Standard (CNS) A4 (210 X 297 mm)