JP2006246200A - Image processing device and method therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing device and a method by which large changes in density at edges of input image data can be stored, and besides the image data can be obtained by a simple device structure and a simple processing method with small and smoothed amplitudes (high-frequency components) such as textures or the like. <P>SOLUTION: The image processing device includes a frequency separating means for separating input data into a high-frequency component, a middle-frequency component, and a low-frequency component that are different from one another, a setting means for setting based on the separated middle-frequency component contribution degrees of the middle- and high-frequency components high when a middle-frequency component amplitude is large, and the contribution degree of the low-frequency component high when a middle-frequency component amplitude is small; and further setting a gain multiplied by the middle- and high-frequency components, depending on the contribution degree that has been set; a multiplying means for multiplying the gain set by the setting means and the middle- and high-frequency components together; and a first combining means for combining the middle- and high-frequency components that have been subjected to a multiplying process by the gain through the multiplying means and the low-frequency components together. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing apparatus and an image processing method capable of easily obtaining an output image close to an ideal state in which a large edge is stored for an input image having a large edge. The present invention relates to an image processing apparatus and an image processing method in which artifacts such as afterglow do not occur when an output image is obtained by compressing the dynamic range of an image.

  Conventionally, as a method of compressing the dynamic range of an image while preserving the texture, generally, a method of compressing only the low frequency component by dividing the image into a low frequency component and other frequency components is used. Yes. In addition, there is a method of generating a gradation control signal for setting the compression amount of the low frequency component using the low frequency component and compressing the dynamic range of the image based on the gradation control signal. .

  In these image dynamic range compression methods, low frequency components are reduced, while medium frequency components and high frequency components are relatively increased. For this reason, undershoot or overshoot may occur depending on the image.

FIG. 8 is a schematic diagram showing a mechanism of occurrence of undershoot and overshoot due to compression processing of a conventional dynamic range of an image.
As illustrated in FIG. 8, low-frequency components of the image data 100 are extracted from the image data 100 of the original image by a filtering process using a low-pass filter, and blurred image data 102 including the low-frequency components is generated.
Here, the original image has a large edge, that is, a region having a large density change, and there are a pattern or a pattern on the high density side and the low density side of the large edge. For this reason, the density distribution of the image data 100 is step-like, and there are high-frequency components 100a and 100b indicating patterns or patterns on the high density side and the low density side of the image data 100.
It is ideal to obtain the blurred image data 104 having the same step-like density distribution as the image data 100 by the filtering process.

  Next, LUT calculation is performed on the blurred image data 102 by LUT to generate a dynamic range compression signal 106 for compressing the dynamic range of the image. By adding the dynamic range compression signal 106 to the image data 100 of the original image, output image data 110 in which the dynamic range of the image data 100 is compressed is obtained. Similar to the image data 100, the output image data 110 includes high-frequency components 110a and 110b indicating a pattern or a pattern on the high density side and the low density side.

  However, in the output image data 110, overshoot occurs on the high density side and undershoot occurs on the low density side in the edge portion 110c. This is because the blurred image data 102 includes medium and high frequency components, and unlike the output image data 110, the edge portion 102a is inclined. For this reason, when the image data 100 and the dynamic range compression signal 106 are added to generate the output image data 110, undershoot or overshoot occurs due to the middle and high frequency components of the blurred image data 102 (dynamic range compression signal 106). It becomes easy to do.

As described above, when the dynamic range of the input image data is compressed, at least one of undershoot and overshoot occurs, and when the photo print is finished, there is a problem that artifacts such as afterglow or false ring occur in the image. is there.
Therefore, various methods for compressing the dynamic range of an image in which the generation of rear light is suppressed have been proposed (see Patent Document 1, Patent Document 2, and Non-Patent Document 1).

In Patent Document 1, a plurality of blurred image signals representing a blurred image of an original image are created from the image signal of the original image, and a synthesized blurred image signal is created by combining the plurality of blurred image signals. An image processing method is disclosed that performs compression processing of the dynamic range of the original image on the image signal of the original image based on the synthesized blurred image signal.
According to the image processing method of Patent Document 1, even a high-contrast or wide dynamic range image such as backlight or strobe photography, or a relatively flat region such as a low-contrast portion in an original image. The sharpness is not lost, and the false contour generated when the dynamic range compression process is strongly lost can be reduced.

Patent Document 2 discloses a dynamic range compression processing method and compression apparatus for an image intended to perform good dynamic range compression processing without generating artifacts in the vicinity of the edge portion.
In the image dynamic range compression processing method and compression apparatus disclosed in Patent Document 2, a blurred image signal creating unit generates blurred image signals Susk (k = 1 to n) having different frequency response characteristics based on the original image signal Sorg. Then, the band limited image signal creating means creates a plurality of band limited image signals based on the blurred image signal Susk. Next, the absolute value of the band limited image signal whose absolute value is larger than a predetermined threshold value among the plurality of band limited image signals is reduced by the converting unit, and the band limited image signal converted by the integrating unit is integrated, and the integrated signal Create Then, the difference signal obtained by subtracting the integration signal from the original image signal Sorg by the subtracting means is converted and added to the original image signal Sorg. Thereby, it is possible to obtain a good processed image Sproc in which the contrast is compressed and there is no artifact.

  Further, Non-Patent Document 1 discloses a filter that uses a non-linear operation called a bilateral filter for creating a blur signal. When the input image data of an image having a large edge with a rough pattern is input, this vital filter preserves the large edge of the image as it is and also has a small amplitude (high frequency component) such as a rough pattern (texture). ) Can be obtained. As described above, according to the bilateral filter, it is possible to obtain image data having ideal characteristics from which small amplitude (high frequency component) is removed while preserving a large edge (change in density) of input image data.

JP 2001-298619 A JP-A-10-75364 Fast bilateral filtering for the display of high-dynamic range images.F. Durand and J. Dorsey.To appear in Proceedings of SIGGRAPH 2002, in Computer Graphics Proceedings, Annual Conference Series, July 2002.

  In Patent Document 1, although a plurality of blurred image signals are used, undershoot and overshoot are caused by the one having the smallest degree of blur among the plurality of blurred image signals, that is, the one having the highest frequency component among the blurred image signals. There is a problem that a chute occurs and the afterglow cannot be sufficiently suppressed. Furthermore, it is necessary to create a plurality of blurred image signals, which requires a plurality of low-pass filters, resulting in a complicated apparatus configuration.

  In the image dynamic range compression processing method and compression apparatus disclosed in Patent Document 2, a single blur mask is not created, but conversion is performed for each of a plurality of bands so that an optimum process is performed. There is a problem that the configuration of the signal creating means, the band limited image signal creating means, and the converting means becomes complicated.

  Furthermore, although the bilateral filter of Non-Patent Document 1 can obtain image data having ideal characteristics with respect to input image data, this bilateral filter performs non-linear computation. There is a problem that processing is very complicated.

An object of the present invention is to solve the problems based on the above-described prior art, and with a simple apparatus configuration and a simple processing method, a large density change such as an edge of input image data is stored, and a texture or the like is reduced. An object of the present invention is to provide an image processing apparatus and an image processing method capable of obtaining image data with smoothed amplitude (high frequency component).
Another object of the present invention is to provide an image processing apparatus and an image processing method capable of suppressing the occurrence of artifacts such as rear light when the dynamic range of an image is compressed.

  In order to achieve the above object, a first aspect of the present invention is based on frequency separation means for separating input image data into different high frequency components, medium frequency components and low frequency components, and the separated medium frequency components. When the amplitude of the medium frequency component is large, the contribution of the medium frequency component and the high frequency component is set high, and when the amplitude is small, the contribution of the low frequency component is set high, and the setting Setting means for setting a gain to be multiplied with respect to the medium frequency component and the high frequency component according to the contribution degree, and a gain set by the setting means is multiplied with respect to the medium frequency component and the high frequency component. And a first combining unit that combines the processed medium frequency component and the processed high frequency component multiplied by the gain by the multiplying unit and the low frequency component. There is provided an image processing apparatus characterized.

  In the present invention, a first compression processing unit that compresses the processed image data combined by the first combining unit, and the compressed processed image data and the input image data are combined. It is preferable to have a second synthesis means.

  In the present invention, the gradation control unit further sets a compression amount of the processed image data based on the processed image data combined by the first combining unit, and the gradation control unit. Second compression processing means for compressing the processed image data in accordance with a set compression amount; and third combining means for combining the compressed image data that has been compressed and the input image data. It is preferable to have.

  Further, in the present invention, the frequency separation means has a cutoff frequency of the first filter for separating the low frequency component set lower than a peak of a frequency response at the time of observing an image, and the high frequency component is It is preferable that the cutoff frequency of the second filter to be separated is set to be higher than the peak of the frequency response when observing the image.

  The second aspect of the present invention includes a step of separating input image data into different high-frequency components, medium-frequency components, and low-frequency components, and the amplitude of the medium-frequency components based on the separated medium-frequency components. Is set to a high contribution rate of the medium frequency component and the high frequency component, when the amplitude is small, to set a high contribution of the low frequency component, according to the set contribution, A step of setting a gain for multiplying the medium frequency component and the high frequency component, a step of multiplying the set gain by the medium frequency component and the high frequency component, and a gain multiplied by the multiplication means. The present invention provides an image processing method comprising a step of combining the processed medium frequency component and the processed high frequency component with the low frequency component.

  In the present invention, the processed medium frequency component and the processed high frequency component are further included in a post-process of synthesizing the processed medium frequency component and processed high frequency component multiplied by the gain and the low frequency component. Preferably, the method includes a step of compressing the processed image data in which the low-frequency component is combined, and a step of combining the compressed image data that has been compressed and the input image data.

  Further, in the present invention, the processed medium frequency component and the processed high frequency component and the processed high frequency component and the low frequency component are combined in a subsequent step of the step of synthesizing the processed medium frequency component and the processed low frequency component. A step of setting a compression amount of the processed image data based on the processed image data obtained by combining the high frequency component and the low frequency component, and compressing the processed image data in accordance with the set compression amount And the step of combining the compressed image data that has been subjected to the compression processing and the input image data.

  Furthermore, in the present invention, the cut-off frequency of the first filter that separates the low-frequency component is set lower than the peak of the frequency response at the time of image observation, and the second filter that separates the high-frequency component is used. The cut-off frequency is preferably set higher than the peak of the frequency response at the time of image observation.

  According to the image processing apparatus and the image processing method of the present invention, the input image data is separated into different high frequency components, medium frequency components, and low frequency components, and when the amplitude of the medium frequency component is large, the medium frequency component and the high frequency component are separated. When the contribution of (medium / high frequency component) is high and the amplitude of the medium frequency component is small, the medium frequency component and the high frequency component are multiplied according to the contribution set to increase the contribution of the low frequency component. Set the gain. The set gain is multiplied by the medium frequency component and the high frequency component, and the processed medium frequency component and the processed high frequency component multiplied by the gain are combined with the low frequency component. Thus, by setting the contribution of each frequency component using the amplitude of the medium frequency component, a large edge of the input image data can be detected without being affected by the high frequency component. As a result, it is possible to obtain image data having ideal characteristics from which small amplitudes (high frequency components) are removed while preserving large edges of the input image data.

  Further, according to the image processing apparatus and the image processing method of the present invention, ideal image data having a large edge in the input image data is separated into different high-frequency components, medium-frequency components, and low-frequency components. Multiplying the medium frequency component and high frequency component by a gain corresponding to the contribution based on the medium frequency component and combining it with the low frequency component makes it easy to obtain with simple processing methods without complicated operations such as nonlinear operations. It is done. In addition, since the above-described ideal image data obtained by the image processing apparatus and the image processing method of the present invention can be obtained by a simple processing method, the configuration of the image processing apparatus can be simplified.

  Furthermore, according to the image processing apparatus and the image processing method of the present invention, ideal image data can be obtained even if the input image data has a large edge. For this reason, when the image processing apparatus and the image processing method of the present invention are applied to compression of the dynamic range of an image, the edge is compressed including the medium frequency component, and the influence of the medium frequency component at the edge can be reduced. Therefore, occurrence of undershoot and overshoot due to compression of the dynamic range of the image can be suppressed. As a result, it is possible to compress the dynamic range of the image while preserving the texture without losing the sharpness of the original image while suppressing the occurrence of artifacts such as back light due to undershoot and overshoot. Therefore, a high-quality image in which texture is preserved without artifacts such as afterglow and without losing the sharpness of the original image can be obtained as, for example, a finished (photo) print.

Hereinafter, an image processing apparatus and an image processing method according to the present invention will be described in detail based on preferred embodiments shown in the accompanying drawings.
FIG. 1 is a block diagram showing a printing system having an image processing apparatus according to the first embodiment of the present invention.
As shown in FIG. 1, the printing system 10 of this embodiment includes an input device 12, a receiving device 14, an image processing device 16, and a printer 18. The input device 12 and the receiving device 14 are connected to the image processing device 16, and the image processing device 16 and the printer 18 are further connected.

For example, the input device 12 photoelectrically reads an image of a subject photographed on a film and acquires input image data as digital image data, and the input image data is output to the image processing device 16.
The input device 12 is, for example, a scanner. This scanner is a device that photoelectrically reads images shot on a film one by one, and separates the light source, variable aperture, and images into the three primary colors R (red), G (green), and B (blue). A color filter plate that has three color filters of R, G, and B to rotate and acts on the optical path by rotating any color filter, and the reading light incident on the film is made uniform in the plane direction of the film A diffusion box, an imaging lens unit, a CCD sensor that is an area sensor that reads an image of one frame of film, an amplifier (amplifier), and R, G, and B output signals are converted into A / D (analog / digital) ) A data processing unit that performs conversion, log conversion, DC offset correction, dark correction, shading correction, and the like.

  In addition, this scanner can be mounted on the main body of the scanner according to the type of film such as a new photographic system (Advanced Photo System) or 135-size negative (or reversal) film, the size of the film, such as strips and slides. A dedicated carrier that can be mounted is prepared. By exchanging the carrier, various types of films and processing can be handled. An image (frame) photographed on a film and used for print creation is conveyed and held at a predetermined reading position by this carrier. In addition, as is well known, a magnetic recording medium is formed on the film of the new photographic system, and a cartridge ID or film type is recorded. Further, at the time of shooting or development, the shooting date and time, Various data such as the presence / absence of strobe light emission, imaging magnification, shooting scene ID, main part position information, and developing machine type can be recorded. The carrier corresponding to the film (cartridge) of the new photographic system is provided with the magnetic information reading means. When the film is conveyed to the reading position, the magnetic information is read, and the various information is transferred to the image processing apparatus. Sent.

  In such a scanner, the reading light emitted from the light source, adjusted in light quantity by the variable aperture, passed through the color filter plate, adjusted in color, and diffused in the diffusion box is held at a predetermined reading position by the carrier. By entering and passing through one frame of the film, projection light carrying an image of this frame photographed on the film is obtained. The projection light of the film is imaged on the light receiving surface of the CCD sensor by the imaging lens unit, is read photoelectrically by the CCD sensor, the output signal is amplified by an amplifier, and sent to the image processing device 16. As the CCD sensor, for example, an area CCD sensor having 1380 × 920 pixels is used.

  In the scanner, such image reading is performed three times by sequentially inserting each color filter of the color filter plate, thereby separating and reading one frame image into the three primary colors of R, G, and B.

  The scanner uses an area CCD sensor and adjusts the reading light with a color filter plate to separate the original image (film projection light) into three primary colors and read the image. As a scanner for supplying image data to the apparatus, three line CCD sensors corresponding to reading of the three primary colors R, G, and B are used, and slit-like reading light (projection light) is scanned and conveyed by a carrier. The image reading apparatus may read an image by so-called slit scanning.

The reception device 14 is installed on, for example, a counter such as a lab store or a convenience store store, and an operator inputs order information data such as order contents and order items to receive and register in order to receive customer orders. In general, a personal computer (PC) having a monitor display screen is used.
The accepting device 14 accepts image data of an image to be printed, for example, at the same time as a customer order. Note that the image data is not limited to the data recorded on one side of the recording medium, and the image data may be formed on both sides of the recording medium.

  In the accepting machine 14, when the order information data is input and the order is confirmed, for example, an acceptance number is given to the order information data and the acceptance date and time is recorded. In addition, the order information data is collated, and if the customer is a previously input customer, a customer ID assigned in advance is assigned. On the other hand, if the customer has not been input before, a new customer ID is assigned. In this way, the accepting device 14 assigns the order number, the reception date and time, and the customer ID to the order information data, and outputs them to the image processing device 16 together with the image data for which the print order has been placed.

  The accepting machine 14 may be used by a customer himself to accept and register a simple order, and is usually composed of a personal computer (PC) having a monitor display screen. In addition to the monitor (display), the PC constituting the accepting machine 14 includes a memory or a storage device (HDD) that temporarily stores input devices such as a keyboard and a mouse, order information, software, and the like. Of course, a printer for printing a voucher issued when an order is confirmed is provided. Furthermore, a media drive for reading image data from a medium such as a memory card, CD, MO, or FD may be provided.

  Further, a film that has been shot may be received by a counter or the like provided with the receiving device 14. The photographed film may be developed, the developed film may be read by a scanner, image data may be created, and output to the image processing device 16.

  In addition, the installation place of the receiving machine 14 is not specifically limited. Further, the accepting machine 14 may be accessible via a network such as the Internet. Moreover, the form of the voucher is not particularly limited, and it is only necessary to record the date and time when it can be received.

  In the present embodiment, the order information data input to the accepting machine 14 is customer information such as a customer's name, a customer's telephone number, and a customer's address. In addition, the order information data includes print processing contents (order contents) such as the type, size and number of prints, the finished image to be printed, and the presence or absence of decorative prints.

  In this embodiment, a scanner that photoelectrically reads an image photographed on a film such as a negative or a reversal is used as the input device 12 and is used as a supply source of input image data of the image processing device 16. The image data supply source that supplies the image data to is not limited to this. For example, a media drive provided in the receiving machine 14 can also be used as an input machine.

  The image processing device 16 performs image processing on the input image data acquired by the input device 12 or the receiving device 14. The image processing apparatus 16 also has an operation system having a keyboard and a mouse for inputting various conditions input (setting), process selection, instructions, or instructions such as color / density correction. A display for displaying an input image, various operation instructions, and various condition setting / registration screens is connected. The image processing device 16 will be described in detail later.

The printer 18 produces a photographic print based on the output image data. For example, the printer 18 exposes a photosensitive material with a light beam modulated in accordance with output image data output from the image processing device 16, develops it, and outputs it as a (finished) print.
Further, the printer 18 is not limited to the silver halide photographic type, and for example, a printer that records an image on a recording medium such as paper such as an ink jet printer, an electrophotographic type, and a thermal melting type printer is used. be able to. In addition, the printer 18 may be of a type according to the required image quality or print form, and may record an image on one side or both sides of a recording medium.

  The image processing apparatus 16 of this embodiment includes a basic image processing unit 20, a condition setting unit (gradation control unit) 22, a matrix (hereinafter referred to as MTX) 24, an edge storage signal processing unit 26, and a lookup table. (Hereinafter referred to as LUT) 28, an adder (combining means) 30, and a data converter 32.

  In the image processing apparatus 16 of the present embodiment, the basic image processing unit 20 is connected to the condition setting unit 22, the MTX 24, and the adder 30. The condition setting unit 22 is connected to an LUT (compression processing means) 28. The MTX 24 is connected to the edge preservation signal processing unit 26. The edge preservation signal processing unit 26 is connected to the LUT 28. The LUT 28 is connected to the adder 30. The adder 30 is connected to the data conversion unit 32. The data conversion unit 32 is connected to the printer 18.

The basic image processing unit 20 has an appropriate color / density (tone reproduction, color reproduction), image structure (sharpness, graininess), etc., in the input image data of the image input from the input device 12 or the reception device 14. Basic image processing for outputting a simple image is performed.
By applying this basic image processing to the input image data, the image is completed in terms of image quality. Basic image processing includes, for example, image enlargement or reduction (electronic scaling processing), tone correction, color / density correction, saturation correction, and sharpness processing.

  The condition setting unit 22 has a function of changing the setting of the compression table provided in the LUT 28 as will be described later. The condition setting unit 22 sets and changes the compression table based on, for example, the density range and density distribution of the input image data obtained by the basic image processing unit 20.

The MTX 24 generates light / dark image data from input image data (image data of each color of R, G, and B) input from the input device 12 or the reception device 14. An image based on light / dark image data is referred to as a light / dark image.
As a method for generating the bright and dark image data, a method of taking one third of the average value of the image data of each color of R, G, and B (Y = (R + G + B) / 3), or color image data using the YIQ rule The method etc. which are converted into bright and dark image data are illustrated. As a method of obtaining bright and dark image data (hereinafter also referred to as Y component data) using the YIQ standard, for example, only Y component (luminance) defined by the YIQ standard is obtained by R, G and Y = 0.3R + 0.59G + 0.11B. A method of calculating from the image data of each color of B is exemplified.

  The edge storage signal processing unit 26 detects edges of input image data, stores edge density distributions, and generates edge storage image data in which a small amplitude (high frequency component) such as texture is smoothed. For example, when the input image data of an image having a large edge with a rough pattern is input, the edge storage signal processing unit 26 stores the large edge of the input image as it is and also has a rough pattern (texture). Edge-saved image data in which a small amplitude (high-frequency component) is smoothed is generated.

  The LUT 28 performs image processing on the edge-preserving image data obtained by the edge preserving signal processing unit 26, and obtains dynamic range compressed data that appropriately compresses the dynamic range and the like of main input image data. In the LUT 28, a compression table for performing dynamic range compression processing on edge-stored image data is set.

The compression amount of the dynamic range in the LUT 28 is basically set and changed by the condition setting unit 22. The compression amount of the dynamic range in the LUT 28 is optimized for each scene, and the compression amount of the dynamic range in the LUT 28 is different for each scene.
Although the compression table in the LUT 28 is set and changed by the condition setting unit 22, a predetermined input unit may be provided and set and changed by manual operation.

  In the present embodiment, the condition setting unit 22 and the edge preservation signal processing unit 26 may be connected. In this case, based on the edge storage image data generated by the edge storage signal processing unit 26, the signal generation unit that generates the gradation control signal for changing the setting of the compression table in the LUT 28 is the condition setting unit 22 or the edge storage. The signal processor 26 is provided. Thus, a gradation control signal is generated by the signal generation unit based on the edge-stored image data, and the setting of the compression table in the LUT 28 is changed by the condition setting unit 22 based on the gradation control signal, so that the dynamic range is compressed. The amount can be set. In this way, dynamic range compressed data may be obtained.

Next, a method for creating a compression table in the LUT 28 of this embodiment will be described.
First, an overall (dynamic range) compression rate α is calculated, and a compression function f (α) using this is set.
For example, a function as shown in FIG. 2 is set in the condition setting unit 22, and the compression rate α is calculated from the dynamic range (DR) of the image signal using this function. In this function, the compression ratio α is 0 when the dynamic range is smaller than the threshold value DRth, and the dynamic range compression processing is not performed for an image with a small dynamic range. This is because when the compression process is performed on an image having a small dynamic range, the contrast of the image is reduced, and conversely, the image quality is lowered.
In addition, it is known that a spot-like brightest image, such as an electric lamp present in the image, can be obtained by skipping to the lowest density on the print rather than producing gradation by dynamic range compression processing. ing. For this reason, in the function shown in FIG. 2, even if the dynamic range becomes larger than the threshold value DRmax, the compression rate α is set so as not to become smaller than the lower limit value αmax.

Using this compression rate α, an overall compression function f (α) is created.
As shown in FIG. 3A, the compression function f (α) is a simple decreasing function whose slope is a compression rate α with a certain signal value as an intersection with the reference value Y 0, that is, the horizontal axis (output 0). . This reference value Y0 is a reference density and may be set as appropriate according to the density at the center of the image of the main subject or the like. For example, when the person is the main subject, the print density, which is substantially the same density as the skin color, is exemplified between 0.5 and 0.7, and preferably about 0.6.

Further, the condition setting unit 22 sets the bright portion (dynamic range) compression rate α _light and the dark portion (dynamic range) compression rate α _dark , the light portion compression function f _light (α _light ), and the dark portion The compression function f _dark (α _dark ) is generated.
As shown in FIG. 3B, the compression function f _light (α _light ) of the bright part is a decreasing function that is lower (minus side) than the horizontal axis (output 0) on the bright part side from the reference value Y0. The function of the slope of the straight line portion is the light portion compression rate α _light, and the output on the dark side of the reference value Y0 is zero. This compression rate α _light is set so that the image signal in the bright portion becomes the image signal in the image reproduction area of the print according to the image feature quantity such as the density histogram or highlight.
On the other hand, the compression function f _dark (α _dark ) of the _dark part is a decreasing function that is above the horizontal axis on the dark part side with respect to the reference value Y0, as shown in FIG. a function as a compression ratio alpha _dark, the output of the bright portion than the reference value Y0 is 0. Similarly, the compression rate α _{dark is} also set so that the image signal in the dark portion becomes the image signal in the image reproduction area of the print according to the image feature quantity such as the density histogram or shadow.

Thus, after calculating the overall compression function f (α), the _light compression function f _light (α _light ), and the dark compression function f _dark (α _dark ), as shown in the following equation: It adds them to create a compression function f _total (α), using the compression function f _total (α), to create a compressed table of LUT 28.
f _total (α) = f (α) + f _light (α _light ) + f _dark (α _dark )

  According to the method of creating the compression table of the LUT 28 of this embodiment, the dynamic range compression is changed to the gradation of the intermediate density portion by fixing the reference value Y0 and independently setting the compression ratios of the bright part and the dark part. The dynamic range compression can be performed by adjusting only the bright part and the dark part.

When the compression function f _light (α _light ) and the compression function f _dark (α _dark ) become functions as shown in FIGS. _3D and 3E, the points P and Q are used. Since artifacts due to the discontinuity of γ are generated, as shown in FIGS. 3B and 3C, the artifacts are prevented from being generated as a function with a smooth differential coefficient. Is preferred. This point is described in detail in JP-A-3-222577.

  The adder 30 adds the input image data and the dynamic range compressed data obtained by the LUT calculation by the LUT 28. As a result, processing for compressing the dynamic range is performed on the input image data, and output image data is obtained. Output image data is output from the adder 30 to the data image converter 32.

  The data converter 32 converts the output image data output from the adder 30 into print image data that can be output by the printer 18. For example, a three-dimensional lookup table is used for conversion from output image data to print image data.

Next, the edge preserving signal processing unit 26 of the image processing apparatus 16 of the present embodiment will be described in detail.
4A is a block diagram showing the configuration of the edge preserving signal processing unit of the image processing apparatus according to the present embodiment, and FIG. 4B shows the weight W on the vertical axis and the medium frequency component M on the horizontal axis. It is a graph which takes a value and shows the conversion table of LUT_W of the edge preservation | save signal processing part.

As shown in FIG. 4A, the edge preserving signal processing unit 26 includes a first LPF 40, a subtractor 42, a second LPF 44, an LUT_W (setting unit) 46, a multiplier (multiplying unit) 48, and an adder. (Synthesizing means) 50.
The first LPF 40 and the subtracter 42 are connected to the MTX 24. The subtractor 42 is connected to the second LPF 44 and the multiplier 48. The second LPF 44 is connected to the LUT_W 46. The LUT_W 46 is connected to the multiplier 48. The first LPF 40 is connected to the adder 50. The adder 50 is connected to the LUT 28.

The first LPF 40 separates the low frequency component L from the light / dark image data (input image data) output from the MTX 24. By this first LPF 40, a rough image gradient of the input image data is obtained. This input image data has a high frequency component H, a medium frequency component M, and a low frequency component L.
As the first LPF 40, for example, an FIR (Finite Impulse Response) type low-pass filter that is usually used for generating blurred image data can be used. In addition, it is preferable to use an IIR (Infinite Impulse Response) type low-pass filter for the first LPF 40 in terms of being able to generate blurred image data in which the image is largely blurred by a small circuit.

Further, a median filter (hereinafter referred to as MF) may be used for the first LPF 40. The use of MF is preferable in that blurred image data in which the edge of a bright and dark image is preserved and noise (high-frequency components) in a flat portion is cut is obtained. In addition, it is particularly preferable to use MF and LPF together and weight-add the images obtained by using both of MF and LPF in that blur image data with a largely blurred image can be generated taking advantage of MF. .
Furthermore, a wavelet transform type filter may be used for the first LPF 40, or a non-linear filter such as a morphology filter and an ε filter may be used.

  The subtractor 42 subtracts the low frequency component L obtained by the first LPF 40 from the light and dark image data generated by the MTX 24 to obtain a medium high frequency component MH. The medium high frequency component MH includes a texture component such as a picture and a noise component.

The second LPF 44 separates a high frequency component H including a texture component indicating a picture and the like from light / dark image data (input image data). The second LPF 44 can obtain the medium frequency component M of the input image data (bright and dark image data). The medium frequency component M indicates a large edge in the image without being influenced by a texture component (high frequency component) such as a pattern and a pattern. For this reason, in the medium frequency component M, it can be determined that a portion having a large amplitude is an edge.
The first LPF 40, the subtractor 42, and the second LPF 44 constitute frequency separation means for separating input image data (bright and dark image data) into different high frequency components, medium frequency components, and low frequency components.

  The LUT_W 46 sets the contribution degree of the image (input image data) based on the value (amplitude) of the medium frequency component M, and sets a gain according to the set contribution degree. In this LUT_W 46, when the value (amplitude) of the medium frequency component M is large, the contribution of the medium frequency component and the high frequency component (hereinafter also referred to as the medium high frequency component) of the input image data (bright and dark image data) is increased. When the value (amplitude) of the medium frequency component M is small, the contribution of the low frequency component of the input image data (bright and dark image data) is increased. The LUT_W 46 is set with a conversion table in which the contributions of the medium-high frequency component MH and the low-frequency component L are determined as weights based on the value (amplitude) of the medium-frequency component M.

  In this embodiment, the texture such as the pattern and the pattern is a high-frequency component, and if the pattern is large, it is erroneously detected as an edge. In order to prevent erroneous detection of an edge due to the high frequency component, the medium frequency component M is used for edge determination. When the amplitude of the medium frequency component M is very large, it can be determined as an edge. Further, when the contribution degree of the medium / high frequency component MH is high, the resulting edge-preserving image data is an original image. For this reason, it can be said that the high contribution of the medium and high frequency components is high in the contribution of the original image.

In the conversion table of the LUT_W 46 of this embodiment, for example, as shown in FIG. 4B, the weight W is 1.0 in a predetermined range where the absolute value of the value (amplitude) of the medium frequency component M is large. That is, the contribution degree of the medium and high frequency components is set high.
On the other hand, in the conversion table of LUT_W46, the weight W monotonously decreases in the range where the absolute value of the medium frequency component M is small, and the weight W is set to 0 when the absolute value of the medium frequency component M is equal to or less than a predetermined value. Thus, the contribution of the low frequency component of the input image data (bright and dark image data) becomes relatively large. That is, the contribution of the low frequency component is set high. With this LUT_W 46, the weight W corresponding to the value of the medium frequency component M output from the second LPF 42 is calculated, and the gain of the multiplier 48 is set.

The multiplier 48 multiplies the medium high frequency component MH obtained by the subtractor 42 by the gain set by the LUT_W 46 to obtain a processed medium high frequency component MH _W. The processed medium and high frequency component MH _W is output to the adder 50. The gain of the multiplier 48 is determined according to the weight W output from the LUT_W 46.

The adder 50 includes a low-frequency component L of the input image, in which the treated in the high-frequency component MH _W synthesized (added). As a result, the density distribution of the input image data is preserved, and the edge preserved image data E is generated from the image data obtained by smoothing a small amplitude such as a texture. For example, a large edge of the input image data is stored, and the texture and the like are smoothed.
As described above, in the edge preserving signal processing unit 26, the input image data is separated into each frequency component, and among these, the weight W (contribution) is determined based on the medium frequency component, and the gain processing is performed. The edge-preserving image data E is obtained by recombining the finished medium and high frequency components MH _W and the low frequency components of the input image.

In the edge preserving signal processing unit 26 of the image processing apparatus 16 of the present embodiment, the weight W (contribution) is determined based on the medium frequency component M of the input image data, and the medium high frequency component at a ratio according to the weight W. The gain of MH is set, and the processed medium high frequency component MH _{W obtained} by multiplying the medium high frequency component MH by this gain and the low frequency component L of the input image data are synthesized. As a result, only the large edge of the original image is detected without being affected by the high-frequency component, the large edge of the original image (input image data) is preserved, and the high-frequency component such as texture is smoothed. Saved image data E can be obtained.

  Further, the edge storage image data E is output to the LUT 28. The edge-preserving image data E has a dynamic range compressed by the LUT 28 to obtain dynamic range compressed data. The dynamic range compressed data and the input image data are combined (added) by the adder 30. Thereby, output image data in which the dynamic range of the input image data is compressed is obtained.

  In addition, since the image processing apparatus 16 according to the present embodiment can detect a large edge of the original image, when compressing the dynamic range of the image, the edge portion of the original image in which artifacts such as backlight are likely to occur. Because it is compressed to include mid-high frequency components, the relative amount of low-frequency components and mid-high frequency components does not easily change, and undershoot and overshoot are less likely to occur. Artifacts such as afterglow are less likely to occur during printing. In addition, when compressing the dynamic range of an image, the low frequency component is compressed except for the edge portion, so that the texture of the image is preserved. As described above, in the image processing apparatus 16 according to the present embodiment, since the occurrence of artifacts such as back light can be suppressed, the original image (input image data) can be input to a print reproduction area such as photographic paper without losing the sharpness. Output image data obtained by compressing image data can be obtained. For this reason, it is possible to obtain a high-quality finished (photo) print with the print 18.

Next, the operation of the image processing apparatus 16 according to the present embodiment will be described with reference to an example in which an input image is acquired by reading a captured image from a film on which an image is recorded with a scanner.
FIG. 5 is a schematic diagram showing the image processing method by the image processing apparatus according to the first embodiment of the present invention in the order of steps.

  First, a film on which an image is recorded is read by a scanner, and input image data 60 (R, G, and B color image data) is obtained. The image read by the scanner has a large edge, and there are a pattern or a pattern on the high density side and the low density side of the large edge. For this reason, the density distribution of the input image data 60 is stepped, and there are high-frequency components 60a and 60b indicating patterns or patterns on the high density side and the low density side. Further, the edge portion 60c of the input image data 60 is substantially vertical.

Next, the input image data 60 is output to the image processing device 16. The image is output to the edge preserving signal processing unit 26 via the basic image processing unit 20 and the MTX 24. In the edge preserving signal processing unit 26, a gain for multiplying the medium-high frequency component MH is set based on the medium-frequency component M of the input image data 60, and the medium-high-frequency component MH is processed. a high-frequency component MH _W, and the low-frequency component L of the input image data is synthesized, edge-preserving image data 62 is created. The edge storage image data 62 has a density distribution obtained by removing high frequency components from the input image data 60, and the edge portion 62a corresponding to the edge portion 60c of the input image data 60 is stored without changing its characteristics. ing. That is, the edge portion 62 a of the edge-preserving image data 62 is substantially vertical as is the edge portion 60 c of the input image data 60.

  Next, the edge storage image data 62 is output to the LUT 28 of the image processing apparatus 16, and the edge storage image data 62 is compressed (LUT calculation) by the compression table of the LUT 28 to create dynamic range compression data 64. The dynamic range compressed data 64 is output to the adder 30.

  Next, the input image data 60 and the dynamic range compressed data 64 are combined by the adder 30 to finally generate output image data 66 in which the dynamic range of the input image data 60 is compressed.

  In this embodiment, the output image data 66 is obtained by compressing the dynamic range into a print reproduction density region 68 such as photographic paper in a state where the high frequency components 60a and 60b of the input image data 60 are stored. In this case, since the edge storage image data 62 reproduces the characteristics of the input image data 60 as substantially ideal, the edge portion of the dynamic range compressed data 64 is maintained as the density distribution of the input image data 60. . For this reason, in the edge portion 66c, the medium frequency component is also compressed, the influence of the medium frequency component can be reduced, overshoot is unlikely to occur on the high concentration side, and undershoot is unlikely to occur on the low concentration side. As described above, since it is possible to suppress the occurrence of overshoot and undershoot at the edge where the density change of the image is large, it is possible to suppress the occurrence of artifacts such as rear light in the finished print.

  As described above, in the image processing apparatus 16 of the present embodiment, when the dynamic range of the image is compressed, overshoot is hardly generated on the high density side, and undershoot is not easily generated on the low density side. The generation of artifacts such as back light in printing can be suppressed, and the texture of the original image can be preserved. Therefore, the original image can be compressed to a print reproduction area such as photographic paper without losing the sharpness of the original image, and a finished print with high quality image quality can be obtained.

Next, a second embodiment of the present invention will be described. In this embodiment, the same components as those in the image processing apparatus according to the first embodiment of the present invention shown in FIGS. 1 to 5 are denoted by the same reference numerals, and detailed description thereof is omitted.
FIG. 6 is a block diagram showing an edge preservation signal processing unit of the image processing apparatus according to the second embodiment of the present invention.

  As shown in FIG. 6, the edge preserving signal processing unit 26a of the present embodiment is provided with a second LPF 44 as compared with the edge preserving signal processing unit 26 (see FIG. 4A) of the first embodiment. Are different from each other in that the subtracter 52 is provided, and the other configuration is the same as that of the edge preserving signal processing unit 26 of the first embodiment, and thus detailed description thereof is omitted.

  In the present embodiment, the second LPF 44 and the subtractor 52 are connected, and the first LPF 40 is connected to the subtractor 52. The subtractor 52 is connected to the LUT_W46. The second LPF 44 is connected to the MTX 24, and input image data (bright and dark image data) is input.

In the edge preserving signal processing unit 26a of this embodiment, the separation of the low-frequency component L by the first LPF 40 and the separation of the medium-high frequency component MH by the second LPF 44 are performed in parallel.
In this embodiment, the subtracter 52 subtracts the low frequency component L from the first LPF 40 from the medium low frequency component LM from the second LPF 44 to obtain the medium frequency component M. The intermediate frequency component M is input to the LUT_W 46 and a gain is set based on the intermediate frequency component M. Hereinafter, since the method for creating the edge-preserving image data E is the same as that of the edge preserving signal processing unit 26 of the first embodiment, detailed description thereof is omitted.

  Also in the present embodiment, as in the first embodiment, when there is a large edge in the original image, this large edge can be detected, and the original image (intermediate high frequency component) can be detected by the medium frequency component M representing this edge. ), And ideal edge-preserving image data close to the image characteristics of the input image data can be obtained. Therefore, overshoot and undershoot are less likely to occur at large edges, and artifacts such as rear light are less likely to occur. Thereby, even when the dynamic range is compressed, output image data from which a high-quality image can be obtained can be obtained.

In the edge preserving signal processing units 26 and 26a of the first and second embodiments described above, the cutoff frequency in the first LPF 40 that extracts the low-frequency component of the input image data is the frequency response at the time of image observation. It is preferable to set it lower than the peak. In addition, the cutoff frequency in the second LPF 44 that removes high-frequency components of the input image data is preferably set to be higher than the peak of the frequency response at the time of image observation.
That is, as shown in FIG. 7, the frequency response characteristic at the time of observing an image is indicated by a curve 70. In this case, when the peak 70a of the curve 70 is divided into _two regions S ₁ and S ₂ , the cut-off frequency in the first LPF 40 is set in the region S ₁ and the cut-off frequency in the second LPF 44 is set in the region S _2. It is preferable to set within.
Here, FIG. 7 is a graph showing frequency response characteristics during image observation, with the relative contrast sensitivity on the vertical axis and the spatial frequency on the horizontal axis.
A curve 70 shown in FIG. 7 is expressed by the following equation when the relative contrast sensitivity is MTF and the spatial frequency is u.
MTF (u) = 5.05exp (−0.138u) · (1-exp (−0.1u))

In the present invention, by setting the cut-off frequencies of the first LPF 40 and the second LPF 44 in accordance with the visual characteristics, the texture recorded in the original image can be effectively stored when the dynamic range is compressed. As a result, the texture can be recognized more clearly when reproduced in a finished (photo) print or the like.
In the present invention, the configuration of the frequency separation means is not particularly limited, and a filter is provided that separates input image data into three different components: a high frequency component, a medium frequency component, and a low frequency component. But you can.

  The image processing apparatus and the image processing method of the present invention have been described in detail above. However, the present invention is not limited to the above-described embodiments, and various improvements and modifications can be made without departing from the gist of the present invention. Of course it is good.

1 is a block diagram illustrating a printing system having an image processing apparatus according to a first embodiment of the present invention. 3 is a graph showing a function of the overall compression rate of dynamic range compression processing in the image processing apparatus of the first embodiment shown in FIG. 1. (A), (b), (c), (d), and (e) are each a figure which shows an example of the compression function used with the image processing apparatus shown in FIG. (A) is a block diagram showing a configuration of an edge preserving signal processing unit of the image processing apparatus of the present embodiment, and (b) shows a weight W on the vertical axis and the value of the medium frequency component M on the horizontal axis. 2 is a graph showing a conversion table of LUT_W. It is a schematic diagram which shows the image processing method by the image processing apparatus of 1st Example of this invention in order of a process. It is a block diagram which shows the structure of the edge preservation | save signal processing part of the image processing apparatus which concerns on 2nd Example of this invention. It is a graph which shows the frequency response characteristic at the time of observing an image by taking a relative contrast sensitivity on a vertical axis | shaft and taking a spatial frequency on a horizontal axis. It is a schematic diagram which shows the mechanism of generation | occurrence | production of the undershoot and overshoot by the compression process of the dynamic range of the conventional image.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Print system 12 Input machine 14 Reception machine 16 Image processing apparatus 18 Printer 20 Basic image processing part 22 Condition setting part 24 Matrix (MTX)
26 Edge Preservation Signal Processing Unit 28 LUT
30, 50 Adder 32 Data conversion unit 36 Compression amount determination unit 40 First LPF (low pass filter)
42, 52 Subtractor 44 Second LPF (low-pass filter)
46 LUT_W
48 Multiplier 60 Input image data 62, E Edge storage image data 64 Dynamic range compressed signal 66 Output image data 70 Curve 70a Peak

Claims (8)

Frequency separation means for separating input image data into different high-frequency components, medium-frequency components, and low-frequency components;
Based on the separated medium frequency component, when the amplitude of the medium frequency component is large, the contribution of the medium frequency component and the high frequency component is set high, and when the amplitude is small, the low frequency component Setting means for setting a high contribution and setting a gain for multiplying the medium frequency component and the high frequency component according to the set contribution;
Multiplication means for multiplying the medium frequency component and the high frequency component by the gain set by the setting means;
A processed medium frequency component and a processed high frequency component multiplied by a gain by the multiplication means;
An image processing apparatus comprising: a first synthesizing unit that synthesizes the low-frequency component. A first compression processing means for compressing the processed image data synthesized by the first synthesis means;
The image processing apparatus according to claim 1, further comprising: a second synthesizing unit that synthesizes the compressed processed image data and the input image data. A gradation control unit configured to set a compression amount of the processed image data based on the processed image data combined by the first combining unit;
Second compression processing means for compressing the processed image data in accordance with a compression amount set by the gradation control unit;
The image processing apparatus according to claim 1, further comprising a third combining unit that combines the compressed image data that has been subjected to the compression process and the input image data.   The frequency separation means has a cutoff frequency of a first filter that separates the low frequency component set lower than a peak of a frequency response at the time of observing an image, and a second filter that separates the high frequency component. The image processing apparatus according to claim 2 or 3, wherein a cutoff frequency is set to be higher than a peak of a frequency response at the time of image observation. Separating the input image data into different high frequency components, medium frequency components and low frequency components;
Based on the separated medium frequency component, when the amplitude of the medium frequency component is large, the contribution of the medium frequency component and the high frequency component is set high, and when the amplitude is small, the low frequency component Setting a high contribution, and setting a gain for multiplying the medium frequency component and the high frequency component according to the set contribution;
Multiplying the set gain by the medium frequency component and the high frequency component;
An image processing method comprising: combining the processed medium frequency component and processed high frequency component multiplied by the gain by the multiplication means and the low frequency component. Further, in a subsequent step of combining the processed medium frequency component and processed high frequency component multiplied by the gain and the low frequency component,
Compressing processed image data in which the processed medium frequency component and the processed high frequency component and the low frequency component are combined;
The image processing method according to claim 5, further comprising a step of synthesizing the compressed image data that has been compressed and the input image data. Further, in a subsequent step of combining the processed medium frequency component and processed high frequency component multiplied by the gain and the low frequency component,
Setting a compression amount of the processed image data based on the processed image data obtained by combining the processed medium frequency component and the processed high frequency component and the low frequency component;
Compressing the processed image data according to the set compression amount;
The image processing method according to claim 5, further comprising a step of synthesizing the compressed image data that has been compressed and the input image data. The cutoff frequency of the first filter that separates the low frequency components is set lower than the peak of the frequency response at the time of image observation, and the cutoff frequency of the second filter that separates the high frequency components is the image observation. The image processing method according to claim 6, wherein the image processing method is set to be higher than a peak of frequency response of time.

Images