JP2006113774A - Image noise calculation method, image processing evaluation method, image processing evaluation device, image processing evaluation program and recording medium with its program recorded - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing evaluation method for evaluating image processing such as halftone processing. <P>SOLUTION: An image processing evaluation part 4 is provided with an image acquiring part 10 for acquiring an input image f and an output image g obtained by operating halftone processing to the input image f, a frequency converting part 11 for frequency-converting the input image f and the output image g, and for acquiring an input spectrum F and an output spectrum G, a VTF correcting part 12 for correcting the input spectrum F and the output spectrum G based on the spatial frequency characteristics VTF of a visual system to acquire the corrected input spectrum F2 and output spectrum G2, an image noise calculating part 14 for calculating an image noise Nz from a differential spectrum DS obtained by calculating a difference between the corrected input spectrum F2 and output spectrum G2 and an evaluated value calculating part 15 for calculating an evaluation value HV for evaluating the halftone processing from the image noise Nz. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image noise calculation method for evaluating image processing such as digital halftoning processing, an image processing evaluation method, an image processing evaluation apparatus, an image processing evaluation program, and a recording medium on which the program is recorded.

Conventionally, as a method for evaluating granular noise generated by digital halftoning processing (hereinafter referred to as “halftone processing”), psychological granularity considering the spatial frequency characteristic VTF (Visual Transfer Function visual transfer function) of the visual system is used. The method to ask is done. For example, in Patent Document 1, a two-dimensional Wiener spectrum of an image to be evaluated is obtained by a two-dimensional Fourier analysis method, and further subjected to a one-dimensional process to calculate an image evaluation value in consideration of the spatial frequency characteristics of the visual system. Thus, a psychological image evaluation method that is not affected by a periodic pattern in a digital image is realized.
Non-Patent Document 1 discloses an evaluation function of granularity considering the spatial frequency characteristic VTF of the visual system, and granularity of the dither method and the error diffusion method, which could hardly be detected by the conventional RMS granularity. It is possible to detect the difference.
Patent Document 2 discloses an image evaluation method using the square of the spatial frequency characteristic VTF of the visual system. This method is characterized in that an evaluation value equivalent to the evaluation function of the graininess in Non-Patent Document 1 is obtained by calculating the graininess in the frequency domain without performing the conversion process to the real space domain.
Patent Document 3 discloses an image evaluation method for evaluating a halftone image by removing a predetermined frequency component from the power spectrum of the halftone image.

JP-A-1-286084 Akito Iwamoto Hiroshi Kodera edited “Digital Hardcopy Technology” Kyoritsu Shuppan, November 15, 2000, p. 95 JP-A-7-177351 JP 2004-64689 A

  In the evaluation of the halftone process, it is important to accurately grasp how much error is generated by the halftone process. However, according to any of the above four conventional techniques, when the gradation value of the pixel of the original image is not uniform and includes a change in gradation value, the change in the gradation value included in the original image is half. Since it remains in the halftone image after the tone processing, it is impossible to evaluate only the error newly generated by the halftone processing (granular noise generated by the halftone processing). Therefore, in the above image evaluation method, the halftone process cannot be accurately evaluated without being affected by the gradation value of the original image. The present invention is an image noise calculation method, an image processing evaluation method, an image processing evaluation apparatus, an image processing evaluation program, and a recording medium on which the program is recorded. The purpose is to provide.

  In order to solve the above problems, in the image noise calculation method of the present invention, an image acquisition step of acquiring an original image having a density level for each pixel and a halftone image that is an image obtained by performing halftone processing on the original image; A difference value acquisition step for acquiring a difference value by performing correction and difference based on the sensitivity characteristic of the visual system for the original image and the halftone image, and an error in density level generated by the halftone process based on the difference value And an image noise calculating step for calculating image noise representing.

  According to this method, since the image noise is calculated from the difference value obtained by performing correction and difference based on the sensitivity characteristic of the human visual system on the original image and the halftone image, the image noise is subjected to halftone processing. This is a value obtained by correcting the error generated by performing it based on the sensitivity characteristic of the visual system. That is, the image noise is not affected by the change in density level of the original image. Therefore, even an original image having a change in the density level of the image, such as a gradation image, represents only an error generated by performing the halftone process without being affected by the change in the density level. Image noise corresponding to the sensitivity characteristic of the visual system can be obtained. Note that the correction and difference processing based on the sensitivity characteristics of the visual system performed on the original image and the processed image are in no particular order, and may be performed after the correction or may be performed after performing the difference. .

  In order to solve the above problems, in the image processing evaluation method of the present invention, an image acquisition step of acquiring an original image having a density level for each pixel and a processed image that is an image obtained by performing predetermined image processing on the original image; A difference value acquisition step of acquiring a difference value by performing correction and difference based on the sensitivity characteristic of the visual system for the original image and the processed image, and a density level generated by performing image processing based on the difference value An image noise calculating step for calculating image noise representing an error, and an evaluation value calculating step for calculating an evaluation value for evaluating image processing based on the image noise.

  According to this method, the evaluation value is calculated from the image noise obtained by correcting the error generated by performing the predetermined image processing based on the sensitivity characteristic of the visual system. Therefore, based on the evaluation value, it is possible to perform an accurate evaluation of the image processing based on the sensitivity characteristic of the visual system without being affected by the change in the density level of the original image. The predetermined image processing includes various image processing that can be performed on the image, such as halftone processing, color conversion processing, image compression processing, and various types of filtering for image processing.

  Here, the difference value acquisition step includes a conversion step for performing frequency conversion for converting the original image and the processed image into the frequency space, and a spatial frequency characteristic representing the sensitivity characteristic of the visual system with respect to the frequency-converted original image and processed image. It is preferable to include a correction step for performing a calculation for correcting by applying the difference and a difference step for acquiring a difference value by subtracting the corrected original image and the processed image.

  In this way, image noise and evaluation values can be calculated in the frequency space.

  Here, the difference value acquisition step expresses the sensitivity characteristic of the visual system with respect to the conversion step for performing frequency conversion for converting the original image and the processed image into the frequency space, and the frequency-converted original image and the frequency-converted processed image. A correction step for performing a correction by applying a spatial frequency characteristic, a reverse conversion step for performing a reverse frequency conversion for converting the corrected original image and processed image into real space, an original image and a processed image subjected to reverse frequency conversion, It is preferable to have a difference step for obtaining a difference value by subtracting.

  In this way, image noise and evaluation values can be calculated in real space.

  Here, it is preferable that the image noise calculation step calculates the image noise by integrating a value obtained by squaring the difference value.

  The inventor of the present invention suitably responds to subjective evaluation by applying a visual spatial frequency on the frequency domain to the image and performing evaluation based on the squared value when evaluating the image. I found that I could perform an objective evaluation. Therefore, since the evaluation value calculated based on the value obtained by squaring the difference value suitably corresponds to human subjective evaluation, it is possible to perform evaluation suitably corresponding to human subjective evaluation.

  Here, the evaluation value calculating step preferably calculates the evaluation value based on a value obtained by averaging the image noise and the density level of the original image.

  In this way, since the calculation is performed based on the average value of the density levels of the original image, it is possible to obtain an evaluation value that matches the average value of the density levels of the original image. The human visual characteristics have different sensitivities with respect to the density level, and the sensitivity is increased for a high density level (that is, dark), and the sensitivity is decreased for a low density level (that is, light). Therefore, by calculating the evaluation value based on the average value of the density levels of the original image, the image processing can be evaluated more accurately.

  Here, the density level may be a luminance level or a brightness level. In this way, it is possible to evaluate the image processing performed on the original image representing the luminance level or the brightness level.

  The present invention can also be an image processing evaluation apparatus. That is, the image processing evaluation apparatus according to the present invention includes an original image having a density level for each pixel, an image acquisition unit that acquires a processed image that is an image obtained by performing predetermined image processing on the original image, the original image, and the processed image. Difference value acquisition means for acquiring a difference value by performing correction and difference based on the sensitivity characteristic of the visual system, and image noise representing an error in density level generated by performing image processing based on the difference value Image noise calculating means for calculating the image processing, and evaluation value calculating means for calculating an evaluation value for evaluating the image processing based on the image noise.

  The present invention may be an image processing evaluation program for causing a computer to perform processing or a recording medium storing an image processing evaluation program. That is, an image processing evaluation program according to the present invention includes an image acquisition unit that acquires a processed image, which is an original image having a density level for each pixel, and an image obtained by performing predetermined image processing on the original image; Difference value acquisition means for acquiring a difference value by performing correction and difference based on the sensitivity characteristic of the visual system, and a density level error generated by performing image processing based on the difference value. The image noise calculating means for calculating the image noise to be expressed and the evaluation value calculating means for calculating an evaluation value for evaluating the image processing based on the image noise are characterized in that they are functioned. As a recording medium on which the program is recorded, various computer-readable media such as a flexible disk, a CD-ROM, an IC card, and a punch card can be used.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram showing a configuration of an image processing evaluation apparatus 1 according to the present invention. The image processing evaluation apparatus 1 is roughly divided into an input unit 2 that acquires an input image (original image) f, and a halftone processing unit 3 that acquires an output image (processed image) g by performing halftone processing on the input image f. And an image processing evaluation unit 4 that calculates an evaluation value for evaluating image noise including granular noise generated by performing halftone processing. As described above, the image processing evaluation apparatus 1 evaluates the performance of the halftone processing itself by evaluating image noise such as granular noise generated by the halftone processing applied to the input image f. . For example, in the improvement of an algorithm for halftone processing, the algorithm can be suitably improved by performing a quantitative performance evaluation using the evaluation value.

  The image processing evaluation apparatus 1 of the present embodiment is configured by a general-purpose personal computer 5 having a CPU, RAM, ROM, and hard disk (not shown), and the CPU of the personal computer 5 is stored in the hard disk. The image processing evaluation program is read into the RAM and executed, thereby functioning as the input unit 2, the halftone processing unit 3, and the image processing evaluation unit 4 described above. This image processing evaluation program may be stored in advance in a hard disk or a ROM, or supplied from the outside by a computer-readable recording medium such as a CD-ROM 6 and personalized via the CD-ROM drive 7. It may be stored by being stored in a hard disk provided in the computer. Of course, it may be stored by accessing a server or the like that supplies a program via a network means such as the Internet and downloading data.

The input unit 2 acquires an image input to the image processing evaluation apparatus 1 as an input image f. As a form of acquiring the input image f, the input image f read by the CD-ROM drive 7 or the scanner provided in the image processing evaluation apparatus 1 may be acquired, or the image recorded on the hard disk is read and input. It may be acquired as an image f. Actually, the CPU acquires the input image f by storing the input image f in a predetermined area of the hard disk.
In the present embodiment, the input image f is expressed by having a gradation value of 256 gradations (8 bits) for each pixel, and the density of the image decreases as the gradation value increases. That is, the gradation value represents the density of black and white for each pixel. When the gradation value is high, the image becomes light and the density level indicating density is low, and when the gradation value is low, the image is dark and the density level is high. . For example, the gradation value “0” is black and “255” is white. As described above, the gradation value corresponds to the density level. Further, since the input image f has a gradation value for each pixel, it can be said that the input image f is an image expressed by the gradation value and the coordinates of the pixel in the real space. In the following description, processing performed on a monochrome image composed of black and white will be described for convenience of explanation. However, the input image f is an image that is displayed in color by providing gradation values for each color, such as RGB data or CMYK data. Alternatively, an image having a number of gradations different from 256 gradations may be used. For example, a color image composed of RGB data can be evaluated by separately performing the processing described below for the gradation values of each RGB color.

  The halftone processing unit 3 uses a well-known image processing method such as a systematic dither method or an error diffusion method for the input image f acquired by the input unit 2 to input an input image having 256 gradation values. A halftone process is performed on f. By performing the halftone process, an output image g which is a halftone image expressing the input image f with binary gradation values (“0” or “255”) is acquired. Actually, the halftone process is performed by the CPU performing an operation for performing the halftone process on the input image f stored in the RAM from the hard disk. The acquired output image g is stored in the hard disk. Similar to the input image f, the output image g having gradation for each pixel is an image expressed in real space.

The image processing evaluation unit 4 uses the input image f acquired by the input unit 2 and the output image g acquired by the halftone processing unit 3 to evaluate halftone processing performed on the input image f.
FIG. 2 is a diagram showing the configuration of the image processing evaluation unit 4. As shown in FIG. 2, the image processing evaluation unit 4 includes an image acquisition unit 10, a frequency conversion unit 11, a VTF (Visual Transfer Function visual transfer function) correction unit 12, an image noise calculation unit 14, and an evaluation value calculation. Part 15.

  The image acquisition unit 10 acquires the input image f. Further, the output image g obtained by performing the halftone process on the input image f by the halftone processing unit 3 is acquired (image acquisition process).

  The frequency conversion unit 11 performs a two-dimensional discrete Fourier transform (hereinafter referred to as DFT (Discrete Fourier Transfer)) on the acquired input image f and output image g, thereby obtaining the input image f and the output image g. An input spectrum F and an output spectrum G represented on the frequency domain are acquired (conversion processing).

  The VTF correction unit 12 corrects the input spectrum F and the output spectrum G by using the spatial frequency characteristic VTF of the visual system, which is a spectrum corresponding to the sensitivity characteristic of human vision, to thereby correct the input spectrum. F2 and the corrected output spectrum G2 are acquired (correction processing).

  The difference unit 13 calculates a difference spectrum DS from the corrected input spectrum F2 and the corrected output spectrum G2 (difference processing). The conversion process, the correction process, and the difference process described above are performed as a difference value acquisition process for acquiring a difference spectrum DS as a difference value from the input image f and the output image g.

  The image noise calculation unit 14 calculates the image noise Nz from the difference value DV (image noise calculation process). The image noise Nz is a value representing an error generated by performing the halftone process on the input image f on the basis of the amount perceived by human vision.

  The evaluation value calculation unit 15 calculates an evaluation value HV from the image noise Nz (evaluation value calculation process).

The image processing evaluation apparatus 1 has been described above. Since the feature of the present invention resides in the processing performed by the image processing evaluation unit 4, the flow of processing performed by the image processing evaluation unit 4 will be described next.
FIG. 3 is a flowchart showing the flow of processing of the image processing evaluation program performed by the image processing evaluation unit 4. Hereinafter, it demonstrates in detail according to a flowchart.

  When the process starts, first, in step S100, the image acquisition unit 10 acquires the input image f and the output image g from the input unit 2 and the halftone processing unit 3 (image acquisition process). The image acquisition process is performed by the CPU reading the input image f and the output image g stored in the hard disk and storing them in the RAM.

  Next, in step S110, the frequency conversion unit 11 performs a two-dimensional DFT on the input image f and the output image g to obtain an input spectrum F and an output spectrum G (conversion processing). Here, if the gradation value of each pixel of the input image f is represented by f (x, y) and the gradation value of each pixel of the output image g is represented by g (x, y), the input spectrum F obtained by DFT The output spectrum G is represented by the following generally known equation.

ここで、Ｗ₁＝ｅｘｐ（−ｊ２Π／Ｍ）、Ｗ₂＝ｅｘｐ（−ｊ２Π／Ｎ）である。Ｍ，Ｎはそれぞれ、入力画像ｆおよび出力画像ｇのｘ方向、ｙ方向の画素の数、ｕ，ｖはそれぞれ、ｘ方向、ｙ方向の空間周波数である。
ステップＳ１１０では、式（１）および式（２）を演算することにより周波数空間上の入力スペクトルＦおよび出力スペクトルＧを得る。なお、実際には、周波数変換部１１は、ＤＦＴの高速化アルゴリズムであるＦＦＴ（Ｆａｓｔ Ｆｏｕｒｉｅｒ Ｔｒａｎｓｆｅｒ）を用いることにより２次元ＤＦＴに要する演算量を少なくしており、ＣＰＵが、周知のＦＦＴアルゴリズムに従って、ＲＡＭに記憶された入力画像ｆおよび出力画像ｇに演算を行うことにより行われる。得られた入力スペクトルＦおよび出力スペクトルＧはＲＡＭに格納される。

Here, W ₁ = exp (−j2∥ / M) and W ₂ = exp (−j2Π / N). M and N are the numbers of pixels in the x and y directions of the input image f and the output image g, respectively, and u and v are the spatial frequencies in the x and y directions, respectively.
In step S110, the input spectrum F and the output spectrum G on the frequency space are obtained by calculating the expressions (1) and (2). In practice, the frequency conversion unit 11 reduces the amount of calculation required for the two-dimensional DFT by using FFT (Fast Fourier Transfer) which is a DFT speed-up algorithm, and the CPU follows the well-known FFT algorithm. The calculation is performed on the input image f and the output image g stored in the RAM. The obtained input spectrum F and output spectrum G are stored in the RAM.

Next, in step S120, the VTF correction unit 12 applies the VTF to the input spectrum F and the output spectrum G, and acquires the corrected input spectrum F2 and the corrected output spectrum G2 (correction processing).
VTF is a function that approximates the spatial frequency characteristic of the visual system. In the present embodiment, the function shown in the following equation when the image observation distance is 350 mm is used. Here, freq is a spatial frequency, and its unit is [cycle / mm].
VTF (freq) = 5.05exp (−0.843 × freq)
{1-exp (−0.611 × freq)} (freq> = 0.79)
= 1.0 (freq <0.79) (3)
The VTF correction unit 12 obtains F (u, v) and G (u, v) by performing an operation that applies VTF to the input spectrum F and the output spectrum G. Here, the input spectrum F (u, v) and the output spectrum represented by the following equations are applied to the input spectrum F (u, v) and the output spectrum G (u, v) represented by the spatial frequencies u and v in the horizontal and vertical directions. By multiplying G (u, v) by VTF, a corrected input spectrum F2 and a corrected output spectrum G2 expressed by the following equations are obtained.
F2 (u, v) = F (u, v) × VTF ((u ² + v ² ) ^1/2 ) (4)
G2 (u, v) = G (u, v) × VTF ((u ² + v ² ) ^1/2 ) (5)
The correction process is performed by the CPU performing calculations shown in Equation (4) and Equation (5), and storing the obtained corrected input spectrum F2 and the corrected output spectrum G2 in the RAM.

Next, in step S130, the difference unit 13 calculates a difference spectrum DS from the corrected input spectrum F2 and the corrected output spectrum G2 according to the following equation (difference processing).
DS (u, v) = F2 (u, v) −G2 (u, v) (6)
The difference process is performed by the CPU reading the corrected input spectrum F2 and the corrected output spectrum G2 stored in the RAM, and then performing the calculation shown in Expression (6). The obtained difference spectrum DS is stored in the RAM.

Next, in step S140, the image noise calculation unit 14 calculates the image noise Nz from the difference spectrum DS according to the following equation (image noise calculation process). That is, the image noise Nz can be obtained by accumulating a value obtained by squaring the difference spectrum DS for each frequency with respect to the frequency.
Nz = Σ _u Σ _v (DS (u, v)) ² (7)
The image noise calculation process is performed by the CPU reading the difference spectrum DS stored in the RAM and executing the calculation shown in Expression (7). The obtained image noise Nz is stored in the RAM.

Next, in step S150, the evaluation value calculation unit 15 calculates an evaluation value HV from the image noise Nz according to the following expressions (8) and (9) (evaluation value calculation processing).
HV = exp (−1.8 × D _AVE ) × Nz (8)
Here, D _AVE is a value obtained by averaging the gradation values of all the pixels of the input image f, and can be expressed by the following equation. In general, since the human visual system changes in sensitivity to image noise with respect to density levels, correction is performed based on the gradation value of the input image f.
D _AVE = Σ _x Σ _y F (x, y) / MN (9)
The evaluation value calculation process is performed by the CPU reading out the image noise Nz stored in the RAM and executing the calculations shown in Expression (8) and Expression (9). The obtained evaluation value HV is stored in a hard disk or the like.
When step S150 ends, the process ends.

  The image processing evaluation apparatus 1 can obtain the evaluation value HV for evaluating halftone processing as described above. Here, the evaluation value HV changes in accordance with the magnitude of the error between the input image f and the output image g, and is a value that matches the sensitivity characteristic of human vision. Therefore, by evaluating the magnitude of the evaluation value HV, it is possible to suitably evaluate the halftone process. In general, the output image g obtained by performing the halftone process on the input image f has an increased granularity indicating the degree to which the gradation value of the image looks grainy compared to the input image f. It can be said that the image noise Nz generated by the halftone process is a value indicating increased graininess. Therefore, the image processing evaluation apparatus 1 can be an image processing evaluation apparatus that evaluates the graininess generated by the halftone process.

  Further, in the image processing evaluation unit 4 of the present invention, in FIG. 2, the image acquisition unit 10, the frequency conversion unit 11, the VTF correction unit 12, and the image noise calculation unit 14 constitute an image noise evaluation device. By evaluating the image noise Nz existing between the input image f and the output image g, even if the input image f has a change in the gradation value (density level) of the image, such as a gradation image, half Only errors generated by performing tone processing can be accurately evaluated.

Hereinafter, effects of the present embodiment will be described.
(1) By evaluating the image noise Nz existing between the input image f and the output image g subjected to the halftone process, the input image f has a gradation value (density level) like a gradation image. Even when there is a change in the image noise Nz, it is possible to obtain image noise Nz that represents only errors generated by performing the halftone process. Therefore, by evaluating image noise such as graininess caused by the halftone process based on the evaluation value HV calculated from the image noise Nz, an accurate evaluation of the halftone process can be performed.

  (2) By evaluating the image noise Nz corresponding to the human visual characteristic (visual transfer function), it is possible to preferably evaluate the halftone process on the basis of the human visual system.

  (3) Based on the value obtained by squaring the difference between the input spectrum F and the output spectrum G, the image noise Nz was calculated. The inventor of the present invention suitably responds to subjective evaluation by applying a visual spatial frequency on the frequency domain to the image and performing evaluation based on the squared value when evaluating the image. I found that I could perform an objective evaluation. Therefore, it is possible to obtain the image noise Nz and the evaluation value HV that suitably correspond to the characteristics of the human visual system.

(4) In the above equation (7), the evaluation value HV is a function of the average value D _AVE of the input image f. That is, in Expression (7), the evaluation value HV decreases when the gradation value of the input image f is high, and the evaluation value HV increases when the gradation value of the input image f is low. In general, the human visual system has low sensitivity when the gradation value of the image is high (that is, the image is whitish and the density level is low). On the other hand, the sensitivity is high when the gradation value of the image is low (that is, dark and the density level is high). By using the evaluation value HV as a function of the average value D _AVE of the input image f as shown in Expression (7), an evaluation value HV corresponding to a change in sensitivity with respect to the density level of the human visual system can be obtained. . Therefore, it is possible to obtain an evaluation value HV that more appropriately corresponds to the characteristics of the human visual system.

(Second Embodiment)
Hereinafter, a second embodiment of the present invention will be described with reference to the drawings.
In the first embodiment, the image noise Nz and the evaluation value HV are calculated from the corrected input spectrum F2 and the corrected output spectrum G2. That is, the calculation is performed on the frequency space. In the second embodiment, an evaluation value is calculated in real space by performing inverse discrete Fourier transform on the corrected input spectrum F2 and the corrected output spectrum G2.

  FIG. 4 is a diagram illustrating a configuration of the image processing evaluation unit 4 according to the second embodiment. As shown in FIG. 4, the image processing evaluation unit 4 includes an image acquisition unit 10, a frequency conversion unit 11, a VTF correction unit 12, an inverse frequency conversion unit 20, a difference unit 13, and an image noise calculation unit 14. And an evaluation value calculation unit 15. About the image acquisition part 10, the frequency conversion part 11, the VTF correction | amendment part 12, and the evaluation value calculation part 15, the process similar to the structure which exists in the said embodiment is performed.

  The inverse frequency transform unit 20 performs a two-dimensional inverse discrete Fourier transform on the corrected input spectrum F2 and the corrected output spectrum G2, and in the real space, the corrected input image f2 and the corrected output. An image g2 is acquired. At this time, processing is performed according to the following equation using the corrected input spectrum F2 and the corrected output spectrum G2 shown in equations (4) and (5).

The difference unit 13 calculates a difference value DV by the following equation.
DV (x, y) = f2 (x, y) -g2 (x, y) (12)

The image noise calculation unit 14 calculates the noise Nz by the following equation. That is, the image noise Nz can be obtained by integrating the values obtained by further squaring the difference value DV for all the pixels.
Nz = Σx _{, y} (DV (x, y)) ² (13)
The evaluation value calculation unit 15 calculates the evaluation value HV according to Expression (8) and Expression (9) using the image noise Nz calculated according to Expression (13). In the second embodiment, the evaluation value HV calculated in the frequency domain in the first embodiment is calculated in the real space, but mathematically equivalent calculation is performed. Therefore, a value similar to the evaluation value HV calculated in the embodiment is obtained. Therefore, the same effect as that of the first embodiment can be obtained.

  The present invention is not limited to the above-described embodiments, and can be implemented in various forms. Hereinafter, modifications of the present invention will be described.

(Modification 1) In the above embodiment, the image noise Nz is calculated by the equations (4) to (7), but the same calculation can be performed in the polar coordinate system. That is, polar coordinate transformation is performed on the input spectrum F (u, v) and output spectrum G (u, v) represented by the horizontal and vertical spatial frequencies u, v, and the input spectrum F (r, r, Θ) and the output spectrum G (r, Θ) are multiplied by VTF (r) to obtain a corrected input spectrum F2 (r, Θ) and a corrected output spectrum G2 (r, Θ). ^{Incidentally, r = (u 2 + v} 2) 1/2, which is ^{Θ = tan -1 (v / u} ).
F2 (r, Θ) = F (r, Θ) × VTF (r) (14)
G2 (r, Θ) = G (r, Θ) × VTF (r) (15)
In the difference processing and the image noise calculation processing in the polar coordinate system, calculations are performed on the corrected input spectrum F2 (r, Θ) and the corrected output spectrum G2 (r, Θ) according to equations (16) and (17). Thus, the image noise Nz is obtained.
DS (r, Θ) = F2 (r, Θ) −G2 (r, Θ) (16)
Nz = Σ _{r ΣΘ} (DS (r, Θ)) ² (17)
The evaluation value calculation process can be performed according to the equation (8) similarly to the embodiment. As described above, the evaluation value HV can be obtained by applying VTF in the polar coordinate system.

Here, in general, human vision has a property of being less sensitive to an oblique direction than to a vertical / horizontal direction. Therefore, as a first modification, in consideration of the fact that the frequency characteristic of human vision also depends on the angle Θ, VTF (r, Θ) representing the frequency characteristic in the radial direction and the angular direction of the frequency is used. Can be used to evaluate. That is, the corrected input spectrum F2 (r, Θ) and the corrected output spectrum G2 (r, Θ) are obtained using the following equations (18) and (19). By performing operations using the equations (16), (17), and (8) on the acquired input spectrum F2 (r, Θ) and the corrected output spectrum G2 (r, Θ), Θ An evaluation value HV in consideration of the direction sensitivity characteristic is obtained.
F2 (r, Θ) = F (r, Θ) × VTF (r, Θ) (18)
G2 (r, Θ) = G (r, Θ) × VTF (r, Θ) (19)
In this way, by obtaining the evaluation value HV in consideration of the sensitivity characteristic in the Θ direction, it is possible to obtain the image processing evaluation apparatus 1 that can perform a suitable evaluation corresponding to the change in the sensitivity characteristic in the oblique direction. it can.

  (Modification 2) Although the calculation method of the evaluation value HV has been described in the above embodiment, the same processing is applied to the process of performing the difference between the input image f and the output image g or the process of applying the VTF. As long as a result is obtained, it is good also as processing in a different order. For example, the VTF can be applied to a spectrum obtained by subtracting the input spectrum F and the output spectrum G in the embodiment. Further, frequency conversion and VTF application processing may be performed on the difference image obtained by subtracting the input image f and the output image g.

(Modification 3) In the above embodiment, the evaluation value HV is calculated, and the magnitude of the value is evaluated, so that the performance evaluation of the halftone process can be performed on the output image F. In the third modification, frequency analysis is performed on the processing performance of the halftone process by calculating the evaluation value HV for each spatial frequency (u, v). That is, the evaluation value ΔHV ((u, v)) for each frequency (u, v) can be obtained by the following equation.
ΔHV (u, v) = ((F2 (u, v) −G2 (u, v)) ² ) ΔuΔv (20)
According to the third modified example, when the generation state of the image noise Nz has frequency dependency, the frequency of the evaluation value HV is calculated by calculating Expression (20) for a plurality of frequencies (u, v). Since the characteristics can be obtained, it is possible to easily know the frequency region where the halftone process can be suitably performed or the frequency region where the halftone process cannot be suitably performed.

  (Modification 4) In the above embodiment, the output image g is obtained from the halftone processing unit 3 of the image processing evaluation apparatus 1. The acquisition path of the output image g and the acquisition path of the input image f are not limited to this. For example, halftone processing is performed on the input image f outside the image processing evaluation apparatus 1, and the input image f and the output image g are You may make it input into the image processing evaluation apparatus 1. FIG.

  (Modification 5) Although the present invention is an image processing evaluation apparatus 1 that performs evaluation of halftone processing, the object to be evaluated is not limited to halftone processing, but for color conversion processing, data compression processing, and image processing. The present invention can also be applied to various image processing performed on an image, such as processing using various filters.

  (Modification 6) In the above embodiment, the halftone process for processing the input image expressed by the gradation value indicating the density level is evaluated. As a sixth modification, processing is not limited to an image representing a density level, and processing may be performed on a luminance level and a lightness level. For example, when evaluating image processing performed on a color image expressed in the YCbCr format, the evaluation value may be calculated using the Y luminance level. Also, when evaluating image processing performed on a color image expressed in the Lab format, an evaluation value may be calculated using the L lightness level. By doing so, it is possible to evaluate image processing performed on color images expressed in various formats.

The figure which showed the structure of the image processing evaluation apparatus. The figure which showed the structure of the image processing evaluation part. The flowchart which showed the flow of the process which an image process evaluation part performs. The figure which showed the structure of the image process evaluation part which concerns on 2nd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Image processing evaluation apparatus, 2 ... Input part, 3 ... Halftone processing part, 4 ... Image processing evaluation part, 5 ... Personal computer, 6 ... CD-ROM, 7 ... CD-ROM drive, 10 ... As an image acquisition means 11... Image conversion unit 11... Frequency conversion unit as conversion unit 12. VTF correction unit as correction unit 13. Difference value calculation unit as difference unit 14. Image noise calculation unit as image noise calculation unit 15 ... an evaluation value calculation unit as an evaluation value calculation means, 20 ... an inverse frequency conversion unit, f ... an input image as an original image, g ... an output image as a processed image and a halftone image, F ... an input spectrum in the frequency domain, G: Output spectrum on frequency domain, F2: Input spectrum after correction on frequency domain, G2: Output spectrum after correction on frequency domain, f2: Correction on real space Input image, g2 ... output image after correction in real space, DS ... difference spectrum as difference value, Nz ... image noise, HV ... evaluation value, DV ... difference value, freq ... spatial frequency, u ... in the x direction Spatial frequency, v ... spatial frequency in the y direction, r ... distance from the origin in the frequency space, Θ ... an angle indicating the inclination from the horizontal direction in the frequency space, M ... the number of pixels in the x direction of the input image, N ... The number of pixels in the y direction of the input image.

Claims (10)

An image acquisition step of acquiring an original image having a density level for each pixel and a halftone image that is an image obtained by performing a halftone process on the original image;
For the original image and the halftone image, a difference value acquisition step for acquiring a difference value by performing correction and difference based on sensitivity characteristics of a visual system;
An image noise calculation method comprising: calculating an image noise representing an error in density level generated by the halftone process based on the difference value. An image acquisition step of acquiring an original image having a density level for each pixel and a processed image that is an image obtained by performing predetermined image processing on the original image;
A difference value acquisition step for acquiring a difference value by performing correction and difference based on sensitivity characteristics of the visual system for the original image and the processed image;
Based on the difference value, an image noise calculating step for calculating image noise representing an error in density level generated by performing the image processing;
An image processing evaluation method comprising: an evaluation value calculating step for calculating an evaluation value for evaluating the image processing based on the image noise. The image processing evaluation method according to claim 2,
The difference value acquisition step includes:
A conversion step of performing frequency conversion for converting the original image and the processed image into a frequency space;
A correction step for performing an operation of correcting the frequency-converted original image and processed image by applying a spatial frequency characteristic representing the sensitivity characteristic of the visual system;
An image processing evaluation method comprising: a difference step for obtaining a difference value by subtracting the corrected original image and the processed image. The image processing evaluation method according to claim 2,
The difference value acquisition step includes:
A conversion step of performing frequency conversion for converting the original image and the processed image into a frequency space;
A correction step for performing an operation of correcting the frequency-converted original image and the frequency-converted processed image by applying a spatial frequency characteristic representing the sensitivity characteristic of the visual system;
An inverse transform step for performing an inverse frequency transform that transforms the corrected original image and processed image into real space;
An image processing evaluation method comprising: a difference step of obtaining a difference value by subtracting the original image subjected to inverse frequency conversion and a processed image. The image processing evaluation method according to any one of claims 2 to 4,
In the image noise calculation step, the image noise is calculated by integrating a value obtained by squaring the difference value. The image processing evaluation method according to any one of claims 2 to 5,
In the image processing evaluation method, the evaluation value calculation step calculates the evaluation value based on a value obtained by averaging the image noise and the density level of the original image. The image processing evaluation method according to any one of claims 2 to 6,
The image processing evaluation method, wherein the density level is a luminance level or a brightness level. An image acquisition means for acquiring an original image having a density level for each pixel and a processed image that is an image obtained by performing predetermined image processing on the original image;
A difference value acquisition unit that acquires a difference value by performing correction and difference based on sensitivity characteristics of a visual system for the original image and the processed image;
Image noise calculating means for calculating an image noise representing an error in density level generated by performing the image processing based on the difference value;
An image processing evaluation apparatus comprising: an evaluation value calculating unit that calculates an evaluation value for evaluating the image processing based on the image noise. Computer
An image acquisition means for acquiring an original image having a density level for each pixel and a processed image that is an image obtained by performing predetermined image processing on the original image;
A difference value acquisition unit that acquires a difference value by performing correction and difference based on sensitivity characteristics of a visual system for the original image and the processed image;
Image noise calculating means for calculating an image noise representing an error in density level generated by performing the image processing based on the difference value;
An image processing evaluation program that functions as an evaluation value calculation unit that calculates an evaluation value for evaluating the image processing based on the image noise. A computer-readable recording medium on which the program according to claim 9 is recorded.

Images