JP2001078235A - Method and system for image evaluation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quantitatively evaluate the false color of a color image input device employing a solid-state image pickup element so as to be in matching with a subjective evaluation. SOLUTION: A display device characteristic correlative conversion section 103 converts an image to be evaluated received from a color image input device 200 in correlation with the characteristics of an image display device 300, a uniform color space component calculation section 104 converts the converted image into a uniform color space component, a statistic quantity calculation section 105 calculates the statistic quantity of the uniform color space component, a lightness difference detection characteristic correction section 106 multiplies a weight coefficient increasing exponentially in response to an average lightness with the statistic quantity of the lightness component so as to correct the statistic quantity corresponding to the lightness difference detection characteristic of a visual system, and a false color evaluation calculation section 107 calculates a false color evaluation value by using the statistic quantity corrected by the lightness difference detection characteristic correction section 106 and the statistic quantity of the chromaticity component.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the evaluation of image quality of digital images against false colors, and more particularly to an image evaluation method and image evaluation of image data captured by a color image input device using a solid-state image sensor. Related to the device.

[0002]

2. Description of the Related Art Conventionally, various attempts have been made as a method for evaluating the image quality of a digital image. There are mainly two types of image quality evaluation methods. The first type is an objective (physical) evaluation for measuring a deterioration factor of the image quality of an image itself, and the second type is an image quality evaluation method. On the other hand, it is a subjective (psychological) evaluation that quantifies the sensations felt by humans.

The objective evaluation can provide a quantitative evaluation, but has a problem that the correlation between the evaluation result and the human subjectivity is not always good. On the other hand, subjective evaluation is
This is an evaluation method that is performed by visually comparing the image to be evaluated with deterioration and the image without deterioration with a visual inspection.The inspection results are different when the inspector is different, or when the inspector is tired, etc. And a stable evaluation result cannot be obtained.

Therefore, in order to solve these problems, an evaluation device and an evaluation method for matching an objective evaluation with a subjective evaluation have been devised. Before describing the evaluation device and the evaluation method, a device to be evaluated will be described.

There are three types of images (hereinafter referred to as images to be evaluated) created by the device to be evaluated. Image output from copy machine or printer Image compressed / decompressed by irreversible image compression Image input from image input device such as scanner or digital camera

The above evaluation target is an image output device such as a copy machine or a printer. The images output from each evaluation target device are captured under the same conditions by an image input device such as a high-definition scanner, for example. The objective evaluation is made based on sharpness, which is the sharpness of the line, granularity, which is the degree of scattering of dots, gradation, which is the degree of change in gradation, and color reproducibility. On the other hand, the subjective evaluation is performed using an image output on paper such as a hard copy. In this field, various attempts have been made to obtain an objective evaluation having a good correlation with the above-described subjective evaluation.

The above evaluation target is an image compression method, and images compressed / expanded by various methods are evaluated.
In this field, importance is placed on the evaluation of "reproducibility" as to how efficiently an original image is compressed (by increasing the compression ratio) and faithful reproduction can be performed when the original image is expanded. The objective evaluation is performed using, for example, the S / N that is the difference between the original image and the processed image per pixel. On the other hand, the subjective evaluation uses both an image output on paper and an image displayed on an image display device such as a CRT display.

The above-mentioned evaluation target is an image input device such as a scanner or a digital camera. The same image is input to each evaluation target device (taken in the case of a scanner and taken in the case of a digital camera) to perform objective evaluation. on the other hand,
For the subjective evaluation, an image displayed on an image display device such as a CRT display or a liquid crystal display of a personal computer is mainly used.

The image quality of a digital camera using a solid-state image sensor among the above evaluation objects will be described. Before that, first, the configuration example of the digital camera is shown,
The schematic flow of captured image data will be described. FIG.
FIG. 2 is a schematic configuration diagram of a digital camera showing a flow of image data. The object is a solid-state image sensor 1 when the shutter is pressed.
2, the image data is taken into the digital camera as image data. Examples of the solid-state imaging device 12 include a CCD, a MOS-type imaging device, and the like. Here, it is assumed that a CCD is employed for convenience of description. The subject is imaged by the imaging lens 11, photoelectrically converted by the CCD 12, and output as an electric signal.

Subsequently, the analog image data output from the CCD 12 is converted by the A / D converter 13 into digital image data. Note that a part that processes a process until the captured image data becomes a digital signal through an optical system (not shown except for the imaging lens 11) is referred to as an imaging unit 14.

In the image processing unit 15, the color output processing from the digital output signal from the A / D conversion unit 13 is performed by the color interpolation processing unit 151, and the aperture processing is performed by the aperture processing unit 152. Various processes for improvement are performed.

Subsequently, the image compression section 16 compresses the image data output from the image processing section 15. The compression is performed for the purpose of reducing the data capacity and the like.
This makes it possible to effectively use the limited built-in memory in the digital camera and to shorten the transfer time to a personal computer or the like. Note that the compressed image data is referred to as a data stream. The image processing unit 15 and the image compression unit 16 are collectively referred to as an image processing compression unit 17.

The data stream is stored in the storage unit 18. Here, the storage unit 18 may be a memory card built in the digital camera, or may be a floppy disk, MO, or the like as an external storage medium. The data stream stored in the storage unit 18 can be reproduced on a display (PC display) 19 such as a personal computer.

Next, the image quality of a digital camera using a solid-state image sensor will be described. Generally, the image quality of a digital camera is evaluated based on three items: (i) sharpness or critical resolution characteristics, (ii) noise characteristics, and (iii) false color characteristics. Here, each item for determining the image quality will be described.

(I) Sharpness or marginal resolution characteristics: The sharpness or marginal resolution characteristics can be examined using a resolution chart based on the flow of an analog camera. The resolution chart is a ladder pattern with different spatial frequencies, that is,
It consists of an image in which black and white stripes are alternately arranged (see FIG. 2). The image quality (sharpness) is evaluated by the number of stripes reproduced with respect to the length of the image taken in the designated framing and the vertical length of the image. The causes of the reduction in image sharpness include blurring of the image due to aberration of the optical system,
This is because a spatial frequency component higher than the Nyquist frequency of the number of captured pixels cannot be reproduced on an image.

(Ii) Noise characteristics: Noise of an image of a digital camera is generated by the influence of an electric shutter. Also,
The subsequent aperture processing unit 152 may further emphasize the image. Noise characteristic evaluation is NTSC YCb
This is performed using the value of S / N using the standard deviation in the Cr color system. Here, the NTSC YCbCr color system is converted from the RGB color system by the following formula.

Y = 0.299 · R + 0.587 · G + 0.114 · B Cb = (BY) / (2-2 * 0.114) Cr = (RY) / (2-2 * 0. 299) ・ ・ ・ (1)

Here, Y is a luminance component, and CbCr is a chromaticity component. In the NTSC YCbCr color system, the brightness component is conventionally referred to as a luminance component, but hereinafter, the brightness or the luminance component will be referred to as the brightness or the brightness component as appropriate.

The NTSC YCbCr color system is used for image compression / decompression processing because a luminance component and a chromaticity component can be obtained by a simple conversion from the RGB color system. However, the NTSC YCbCr color system is CIE 1976
It is not a "uniform color space" in which the color difference on the coordinates matches the color difference perceived by humans, as in the L * a * b * color system. Therefore,
Such a color system is suitable for image processing with reproducibility, but is not suitable as a color system used for determining image quality.

(Iii) False color characteristics: False color occurs in the following process. In general, a solid-state imaging device of a digital camera is a single-chip type CCD (Charge Coupled Dev).
ice) is used. The pixel arrangement of the single-chip CCD is shown in FIG.
In many cases, a Bayer array such as
In order to have all the GB signals, a color interpolation process (see a color interpolation processing unit 151 in FIG. 7) is performed to interpolate a missing color signal in a target pixel from surrounding pixels. For this reason, if the interpolation method is not appropriate, a color that is supposed to be an achromatic color will be colored, which will be a false color. Specifically, a false color is likely to occur at the edge portion of the ladder pattern as shown in FIG. 9, that is, at a portion where the reflectance changes greatly (a portion having a large contrast).

Techniques for suppressing the generation of false colors have been proposed in the past. For example, Japanese Patent Application Laid-Open No. 5-181006 discloses a "complementary color filter for CCD and a false color correction circuit using the same". A technique has been disclosed in which a new complementary color filter for a CCD sensor is proposed, and a false color is corrected using the color filter. In other words, four pixels on each side of a large number of color regions that are divided and arranged in a grid pattern vertically and horizontally,
In order to have the same color component at any position, two types of colors are alternately arranged for each pixel in the horizontal direction and the vertical direction of the color region.
A color difference signal is obtained based on the pixel. Further, two types of colors are alternately arranged for each pixel, and adjacent to the four pixels of interest so that no false color occurs in the color difference signal of the four pixels of interest, and
A correction value is calculated from the color information of four pixels for complementary colors overlapping by one line, and a false color generated in the target pixel is suppressed. Therefore, even if a false color area occurs at any position in the filter regardless of the horizontal or vertical direction, it is possible to always perform the correction of the false color using the color difference signal of the adjacent area.

Further, in Japanese Patent Application Laid-Open No. 10-108033, "Scanner and Interpolated Pixel Generation Method", a scanner comprising a green line sensor and a red / blue line sensor is used.
In a scanner having a line sensor, when generating image data representing an interpolation target pixel for which image data is missing, pixels (for example, R
1 and R5), the correction data (Δ3 = G3−3) is obtained from the pixels (G3, G1, G5) adjacent to both sides of the interpolation target pixel with respect to the data representing the arithmetic mean ((R1 + R5) / 2) of the interpolation target pixel.
A technique is disclosed in which (G1 + G5) / 2) is created and added to prevent generation of a false color in an edge portion.

In the "color image signal forming apparatus" disclosed in Japanese Patent Application Laid-Open No. Hei 10-304387, the generation of false colors is not suppressed by signal processing as in the above two techniques. The technology which suppresses is disclosed. That is, the light from the condenser lens is imaged on the color filter of the photoelectric conversion element of the CCD through the wedge prism. Here, the R, G, and B lights separated in the prism are collected on the corresponding R, G, and B color filters, respectively. As a result, the R, G, and B light quantity ratios of the incident light can be accurately detected, and the occurrence of false colors can be suppressed.

[0024]

However, the prior art relates to a technique for suppressing the generation of false colors, and an image quality evaluation relating to false colors has not been proposed. Therefore, there is a problem that a quantitative evaluation in which a subjective evaluation and an objective evaluation are correlated cannot be performed on a false color.
This is due to the fact that at which spatial frequency value of the ladder pattern a false color occurs depends on the number of pixels captured by the digital camera and the processing of the color interpolation processing unit 151, and is therefore unpredictable. . Further, when the image processing is different between the horizontal direction and the vertical direction of the image, the method of generating a false color also differs between the horizontal direction and the vertical direction, which is based on the internal processing. Therefore, for example, there is a problem that there is no objectivity when evaluating the image quality (false color) of a device (for example, a digital camera) using a solid-state imaging device.

The present invention has been made in view of the above, and has as its object to quantitatively evaluate a false color of a color image input device using a solid-state image sensor so as to conform to a subjective evaluation.

[0026]

According to an aspect of the present invention, there is provided an image evaluation method, comprising the steps of: inputting image data as an image to be evaluated from a color image input device using a solid-state imaging device; An image evaluation method for evaluating image quality related to a false color of the image to be evaluated, wherein the image to be evaluated is converted in accordance with characteristics of an image display device that inputs and displays image data from the color image input device. A display device characteristic correspondence conversion step, a uniform color space component calculation step of converting the evaluated image converted in the display device characteristic correspondence conversion step into a uniform color space component, and a uniform color space calculated in the uniform color space component calculation step A statistic calculation step of calculating the statistic of the component, and a statistic of the brightness component calculated in the statistic calculation step is multiplied by a weighting factor that increases exponentially according to the average brightness, and the lightness difference of the visual system A lightness difference detection characteristic correction step of performing correction corresponding to the output characteristic; a statistic corrected in the lightness difference detection characteristic correction step; and a false color calculating a false color evaluation value using the statistic of the chromaticity component. Evaluation value calculating step.

That is, according to the first aspect of the present invention, in the display device characteristic correspondence conversion step, the image to be evaluated is converted in accordance with the characteristics of the image display device for inputting and displaying image data from the color image input device, and the conversion is performed uniformly. In the color space component calculation step, the evaluated image converted in the display device characteristic conversion conversion step is converted into a uniform color space component, and in the statistic calculation step, the statistic of the uniform color space component calculated in the uniform color space component calculation step is calculated. And
In the brightness difference detection characteristic correction step, the statistic of the brightness component calculated in the statistic calculation step is multiplied by a weight coefficient that increases exponentially according to the average brightness, and correction corresponding to the brightness difference detection characteristic of the visual system is performed. In the false color evaluation value calculation step, a false color evaluation value is calculated using the statistic corrected in the brightness difference detection characteristic correction step and the statistic of the chromaticity component.

Further, the image evaluation method according to claim 2 is
An image evaluation method for inputting image data from a color image input device using a solid-state imaging device as an image to be evaluated, and evaluating image quality related to a false color of the image to be evaluated. A display device characteristic conversion step of converting the evaluated image in accordance with the characteristics of the image display device to be input and displayed; and converting the evaluated image converted in the display device characteristic conversion step into a uniform color space component. A uniform color space component calculating step, and at least an average value of lightness (average lightness value), a statistical deviation of lightness, and a statistic of chromaticity from each component of the uniform color space converted in the uniform color space component calculating step. A statistic calculation step of calculating a deviation,
A false color evaluation value calculating step of calculating an evaluation value for a false color based on the statistical deviation of the lightness and the statistical deviation of the chromaticity using a power of the average lightness value calculated in the statistical amount calculating step as a coefficient. .

That is, in the invention according to claim 2, in the display device characteristic correspondence conversion step, the image to be evaluated is converted in accordance with the characteristics of the image display device for inputting and displaying the image data from the color image input device, and the image data is uniformly converted. In the color space component calculation step, the evaluated image converted in the display device characteristic correspondence conversion step is converted into a uniform color space component, and in the statistic calculation step, at least from each component of the uniform color space converted in the uniform color space component calculation step, , The average value of the lightness (average lightness value) and the statistical deviation of the lightness and the chromaticity are calculated. In the false color evaluation value calculation step, the power of the average lightness value calculated in the statistic calculation step is calculated by a coefficient. The evaluation value regarding the false color is calculated based on the statistical deviation of the brightness and the statistical deviation of the chromaticity.

The image evaluation method according to claim 3 is
3. The image evaluation method according to claim 1, wherein in the display device characteristic correspondence conversion step, at least one of chromaticity of each primary color, chromaticity of a white point, and gamma characteristic of the image display device is used as a parameter. Used to convert the evaluated image.

That is, in the invention according to claim 3, in the display device characteristic conversion step, at least one of the chromaticity of each primary color, the chromaticity of a white point, and the gamma characteristic of the image display device is used as a parameter. To convert the evaluated image.

[0032] The image evaluation apparatus according to claim 4 is
An image evaluation device that inputs image data from a color image input device using a solid-state imaging device as an image to be evaluated, and evaluates image quality related to a false color of the image to be evaluated. Display device characteristic conversion means for converting the image to be evaluated in accordance with the characteristics of the image display device to be input and displayed; and display device characteristic conversion for converting the image to be evaluated in accordance with the characteristics of the image display device Means, a uniform color space component calculating means for converting the evaluated image converted by the display device characteristic correspondence converting means into a uniform color space component, and a statistic of the uniform color space component calculated by the uniform color space component calculating means. A statistic calculating means for calculating;
The statistic of the brightness component calculated by the statistic calculation means, multiplied by a weighting factor that increases exponentially according to the average brightness, brightness difference detection characteristic correction means for performing correction corresponding to the brightness difference detection characteristics of the visual system, And a false color evaluation value calculation means for calculating a false color evaluation value using the statistic corrected by the lightness difference detection characteristic correction means and the statistic of the chromaticity component.

That is, in the invention according to claim 4, the display device characteristic correspondence conversion means converts the image to be evaluated in accordance with the characteristics of the image display device which inputs image data from the color image input device and displays the image data. The device characteristic correspondence conversion means converts the evaluated image in accordance with the characteristics of the image display device, and the uniform color space component calculation means converts the evaluated image converted by the display device characteristic correspondence conversion means into a uniform color space component, The statistic calculation means calculates the statistic of the uniform color space component calculated by the uniform color space component calculation means, and the brightness difference detection characteristic correction means calculates the statistic of the lightness component calculated by the statistic calculation means, according to the average brightness. Multiplied by a weighting factor that increases exponentially, performs correction corresponding to the lightness difference detection characteristic of the visual system, and the false color evaluation value calculation means corrects the statistic amount corrected by the lightness difference detection characteristic correction means, Using the quantity To calculate the false color evaluation value.

[0034] The image evaluation device according to claim 5 is
An image evaluation device that inputs image data from a color image input device using a solid-state imaging device as an image to be evaluated, and evaluates image quality related to a false color of the image to be evaluated. A display device characteristic conversion unit for converting the image to be evaluated in accordance with the characteristics of the image display device to be input and displayed; and converting the evaluated image converted by the display device characteristic conversion unit into a uniform color space component. A uniform color space component calculating means, and at least an average value of lightness (average lightness value), a statistical deviation of lightness, and a statistic of chromaticity from each component of the uniform color space converted by the uniform color space component calculating means. A statistic calculation means for calculating the deviation,
False color evaluation value calculation means for calculating an evaluation value relating to a false color based on the statistical deviation of the lightness and the statistical deviation of the chromaticity using a power of the average lightness value calculated by the statistical amount calculation means as a coefficient. Things.

That is, in the image evaluation apparatus according to a fifth aspect, the display device characteristic correspondence conversion means converts the image to be evaluated in accordance with the characteristics of the image display device that inputs and displays image data from the color image input device. The uniform color space component calculating means converts the evaluated image converted by the display device characteristic correspondence converting means into a uniform color space component, and the statistic calculating means converts each component of the uniform color space converted by the uniform color space component calculating means. Calculate at least the average value of the lightness (average lightness value) and the statistical deviation of the lightness and the statistical deviation of the chromaticity, and calculate the power of the average lightness value calculated by the false color evaluation value calculating means by the statistical amount calculating means. An evaluation value relating to a false color is calculated based on the statistical deviation of lightness and the statistical deviation of chromaticity as coefficients.

The image evaluation device according to claim 6 is
6. The image evaluation device according to claim 4, wherein the display device characteristic conversion unit uses at least one of a chromaticity of each primary color, a chromaticity of a white point, and a gamma characteristic of the image display device as a parameter. Used to convert the evaluated image.

That is, in the image evaluation device according to the present invention, the display device characteristic correspondence conversion means sets at least one of the chromaticity of each primary color, the chromaticity of a white point, and the gamma characteristic of the image display device as a parameter. To convert the evaluated image.

[0038]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described with reference to the drawings with reference to a color image input device, an image evaluation device, a display device characteristic conversion unit, and a uniform color space component calculation unit. ] [Statistical amount calculation unit] [Lightness difference detection characteristic correction unit]
This will be described in detail in the order of [false color evaluation value calculation unit].

FIG. 1 is a block diagram showing a configuration example of the image evaluation apparatus of the present invention. In the figure, 100 denotes an image evaluation device, 200 denotes a color image input device, and 300 denotes an image display device.

[Color Image Input Device] In this embodiment, the color image input device 200 is assumed to be a digital camera. The digital camera 200 shown in FIG.
, A series of patterns in which the low reflectance and the high reflectance are periodically repeated and have various continuous frequency components from a low spatial frequency component to a high spatial frequency component (hereinafter referred to as a ladder pattern as appropriate) ), And an image to be evaluated is obtained.

As a specific input image, an "ISO standard electronic still camera resolution chart" (hereinafter referred to as an ISO chart) can be given. FIG. 4 is a diagram showing a resolution chart (ISO chart) for an ISO standard electronic still camera, and FIG. 5 is a diagram showing names of respective parts of the ISO chart. In the P1 and P2 parts of the ISO chart, 1
Ladder patterns from 00TV to 1000TV are described. For example, 100 TV books means the vertical direction 20 of the active area of the chart (white area of the chart).
Refers to the number of black and white lines that fit in cm. However, two black and white lines are counted as two.

If the imaging condition of the input image to be used is determined, it is followed. In the case of an ISO chart, the framing point and the focus setting point for each aspect ratio of the number of pixels captured by the digital camera 200 are determined, such as the uniformity of the luminance distribution on the chart. The evaluated image captured by the digital camera 200 is an evaluation target of the image evaluation device 100.

Although a digital camera is used as an example of the color image input device 200 using a solid-state image pickup device for convenience of the following description, a color image input device such as a scanner may be used. Good. Further, as the input image, the chart shown in FIGS. 2 and 3 using an ISO chart may be used, or various other evaluation charts may be used.

[Image Evaluation Apparatus] The image evaluation apparatus 100
An image storage unit 101 that stores an image captured by the digital camera 200 as an image to be evaluated, a selection unit 102 that selects (cuts out) a part for evaluating image quality (false color),
A display device characteristic correspondence conversion unit 103 that converts the evaluated image in accordance with the characteristics of the image display device 300 described below;
Uniform color space component calculation unit 1 that calculates a uniform color space component from the evaluated image converted by the display device conversion unit 103
04, a statistic calculation unit 105 that calculates statistics such as an average value and a deviation of the uniform color space components calculated by the uniform color space component calculation unit 104, and a uniform color space component calculated by the statistic calculation unit 105 The brightness difference detection characteristic correction unit 106 that performs correction corresponding to the lightness difference detection characteristic of the visual system, which will be described later, the statistic corrected by the brightness difference detection characteristic correction unit 106, and calculated by the statistics calculation unit 105 A false color evaluation value calculation unit 107 that calculates the image quality of the digital camera 200 with respect to the false color from the obtained statistics.

The image evaluation device 100 includes an image display device coefficient storage unit 1 which is referred to when the display device characteristic conversion unit 103 performs conversion corresponding to the characteristics of the image display device 300.
08, the image display device coefficient storage unit 108 stores a tristimulus value of a white point described later, RGB xy chromaticity coordinates, parameters related to gamma (input / output characteristics of the image display device 300), and the like. . Similarly, the image evaluation device 100
The lightness difference detection characteristic correction unit 106 stores a lightness difference detection characteristic correction coefficient storage unit 109 that stores an index used when correcting the perception characteristic of the human visual system for the lightness difference, and the false color evaluation value calculation unit 107 False color evaluation value correction coefficient storage unit 110 that stores coefficients used when finally calculating a false color evaluation value
And

Next, an outline of an image evaluation method by the image evaluation apparatus 100 will be described. FIG. 6 is a flowchart illustrating an outline of the image evaluation method for false colors.

First, an image for input (image of a subject) is captured by the digital camera 200 (step S).
11). The captured image is stored in the image storage unit 101 of the image evaluation device 100 as an image to be evaluated (step S
12). Subsequently, a portion where the false color occurs most is cut out of the evaluated image (step S13). This is performed by cutting out using a commercially available image tool or the like. The cutting process is performed by the selection unit 102 of the image evaluation device 100. The cut portion is stored again in the image storage unit 101 as an image to be evaluated (Step S).
14).

By the way, the normally picked up image is displayed on an image display device 300 such as a CRT display or a liquid crystal display.
Will be displayed. The display device characteristic conversion unit 103 objectively evaluates the evaluated image in correlation with human subjectivity,
The image to be evaluated stored in the image storage unit 101 is converted in accordance with the characteristics of the image display device 300 to be used, and output to the uniform color space component calculation unit 104 (step S15).
The detailed processing content of the display device characteristic correspondence conversion unit 103 will be described later.

The uniform color space component calculation unit 104 converts the output of the display device characteristic correspondence conversion unit 103 into a uniform color space component which is supposed to quantitatively represent a perceptual difference with respect to human colors, Output to the amount calculation unit 105 (step S1
6). The detailed processing content of the uniform color space component calculation unit 104 will be described later.

The statistic calculation unit 105 calculates an average value and a deviation of the color space components based on the output of the uniform color space component calculation unit 104 (step S17). The lightness difference detection characteristic correction unit 106 performs exponentiation calculation based on the average lightness value output from the statistic calculation unit 105 from the lightness difference detection characteristic correction coefficient storage unit 109, and considers human visual characteristics. The calculated correction value is calculated (step S18). The processing contents of the statistic calculation unit 105 and the brightness difference detection characteristic correction unit 106 will be described later.

The false color evaluation value calculation unit 107 determines in step S1
8 using the correction value calculated in step 8 and taking into account the coefficient stored in the false color evaluation value correction coefficient storage unit 110.
The false color evaluation value is calculated on the basis of the statistics output by the counter 5 (step S19).

Next, the display device characteristic correspondence conversion unit 103, the uniform color space component calculation unit 104, the statistic calculation unit 105, the lightness difference detection characteristic correction unit 106, and the false color evaluation value calculation unit 10
7 will be described in detail.

[Display device characteristic conversion unit] The display device characteristic conversion unit 103, as described above, evaluates the image to be evaluated stored in the image storage unit 101 in order to perform an objective evaluation with good correlation with human subjectivity. Image display device 300 using
(Step S1 in FIG. 6).
5). This is because the captured image is usually displayed on an image display device such as a CRT display. The specific processing contents will be described.

The display device characteristic correspondence conversion unit 103 stores (R, G, B) 0 to 25 stored in the image storage unit 101.
The data of the five-valued image to be evaluated is read out for each pixel, and converted into tristimulus values (X, Y, Z) according to the characteristic parameters of the display for display, for example, by a conversion matrix expressed by the following equation (2). And outputs the result to the uniform color space calculation unit 103. Since the method of deriving the transformation matrix is a well-known technique, the description is omitted here. 4 shows an example of a conversion matrix.

For example, a conversion matrix in the case of a display calibrated to a color temperature of 9300 K ° (substantially the same as the color temperature of a display of a general personal computer) and having the following conditions can be expressed by the following equation (2). .

Tristimulus value (Xn, Yn, Zn) at white point = (95.25, 100, 14)
1.25) RGB xy-chromaticity coordinates (xR, yR, zR) = (0.625, 0.339, 1
−xR−yR) (xG, yG, zG) = (0.283, 0.606, 1
−xG−yG) (xB, yB, zB) = (0.150, 0.063, 1
−xB−yB) Gamma (input / output characteristics of the image display device 300) = 1.0

[0057]

(Equation 1)

Note that the XYZ data is 9300
The present invention is not limited to the CRT display calibrated to K °, but may use a conversion matrix according to the characteristics of the image display device 300.

As described above, the display device characteristic correspondence conversion unit 103 uses at least one of the chromaticity of each primary color, the chromaticity of the white point, and the gamma characteristic of the image display device 300 as a parameter, and Since the conversion corresponding to the characteristic (environment) when displayed on the image display device 300 is performed on the evaluation image, it is possible to obtain an evaluation value of the image quality in consideration of the actual observation environment. .

[Equivalent Color Space Component Calculator] As described above, the uniform color space component calculator 104 quantitatively expresses the output of the display device characteristic correspondence converter 103 as a perceptual difference with respect to human colors. (See FIG. 6)
Step S16). The specific processing content will be described in detail.

The uniform color space component calculation unit 104 converts the output of the display device characteristic correspondence conversion unit 103 into, for example, an L * a * b * uniform color space. The conversion formula for converting to the L * a * b * uniform color space is represented as the following formula (3). L * a *
b * data (CIE1976L * a * b * data) is a color space that reflects human color perception, and includes a lightness component and a chromaticity component.

L * = 116f (Y / Yn) -16 a * = 500 {f (X / Xn) -f (Y / Yn)} b * = 200Δf (Y / Yn) -f (Z / Zn )｝ ・ ・ ・ (3)

Here, (Xn, Yn, Zn) are the tristimulus values at the white point of the display for display described above.
F (ω) is represented by the following equation (3 ′) using ω as a variable (ω = X / Xn, Y / Yn, Z / Zn).

F (ω) = ω ^1/3 (ω> 0.008856) f (ω) = 7.787ω + 16/116 (ω ≦ 0.008856) (3 ′)

As described above, when the uniform color space component calculating unit 104 calculates the lightness component (L *), the human visual system is most sensitive to lightness. The statistic calculation unit 105 to be described can calculate the false color evaluation value in consideration of the deviation based on L *. Here, the L * a * b * uniform color space has been described as an example, but a Luv uniform color space may be used.

When the false color evaluation value is obtained, if the NTSC YCbCr space is used similarly to the noise characteristic,
Since the origin of the bCr chromaticity coordinate system is assumed to be an achromatic point (however, strict proof has not been performed), the deviation from the origin can be interpreted as a false color component, and this point is useful. It can be said that it is a color space. However, humans are more sensitive to changes in the lightness component (however, they are customarily called “brightness components” in the NTSC YCbCr color system) than changes in the chromaticity component. When the color space is not used, and there is no luminance component, there is a drawback that an evaluation value having a good correlation with the subjectivity cannot be obtained. According to the inventor's subjective evaluation experiment, NTSC YCb
The contribution ratio between the average radius and the subjectivity in the Cr color system, that is, the square of the correlation coefficient was 0.414.

[Statistics Calculation Unit] The statistics calculation unit 105
As described above, the statistic of the color space component is calculated based on the output of the uniform color space component calculation unit 104 (step S17 in FIG. 6). Originally, the captured ISO chart is an achromatic color, but a false color is generated by image processing in the color complement processing unit 151 (see FIG. 7) and the like. The value affected by the generation of the false color is the change amount of the chromaticity component a * b * and the change of the lightness component L * accompanying the change of the chromaticity component in the L * a * b * uniform color space. Quantity.

However, for example, in an image subjected to compression / expansion processing, there exists digital image data (original image) before processing and digital image data after compression / expansion. In the apparatus, there is no original image of the evaluated image. Therefore,
It is not possible to calculate the amount of change of the lightness component L * or the chromaticity components a * b * "from the original image".

Therefore, for the lightness component L *, the “average deviation from the average lightness” is used as the change ΔL * of the lightness component, and for the chromaticity component a * b *, The average value of the distances ”is defined as the chromaticity component change amount Δa * b *. Specifically, each change amount is defined by the following equation (4).

ΔL * ≡ (Σsqrt ((Li * −ave (Li *)) ² )) / N Δa * b * ≡Σ (sqrt ((ai *) ² + (bi *) ² )) / N・ (4)

Accordingly, the color generation amount Δcolor on the L * a * b * uniform color space can be expressed by the following equation (5).

Δcolor≡sqrt ((ΔL *) ² + (Δa * b *) ² ) (5)

[Brightness Difference Detection Characteristic Correction Unit] The brightness difference detection characteristic correction unit 106 refers to the exponent from the brightness difference detection characteristic correction coefficient storage unit 109 based on the average brightness value output from the statistic calculation unit 105, A power calculation is performed to calculate a correction value in consideration of human visual characteristics (step S18).

One of the human visual characteristics is that even if the numerical amount of noise in an image (for example, the amount of color generation Δcolor) is the same, the perceived noise is different. Experiments have confirmed that false colors are emphasized and perceived as the average brightness of the evaluated image increases. Even if the false color evaluation value is the same, a correct objective evaluation cannot be made if the false color actually perceived is different, so the false color evaluation value is a value that does not depend on the average brightness of the image. Must-have. Through experiments, it has been found that a value that increases exponentially according to the average brightness of the evaluated image is appropriate for the brightness correction (brightness difference detection characteristic) for the false color. That is, the brightness difference detection characteristic is represented by Expression (6).

Lightness difference detection characteristic ≡ (ave (L *)) θ (6) Here, θ is an index obtained by experiment.

[False Color Evaluation Value Calculation Unit] False Color Evaluation Value Calculation Unit 1
07 is a false color evaluation value in which the subjective evaluation and the objective evaluation are finally matched in consideration of the statistic output from the statistic calculation unit 105 and the lightness difference detection characteristic calculated by the lightness difference detection characteristic correction unit 106. Is calculated (step S19 in FIG. 6). The false color evaluation value is represented by Expression (7).

[0077]

(Equation 2)

In equation (7), ave is a bar and sq
rt is represented by the square root.

The coefficients α, β, γ, ζ, ξ, φ, and で あ are the values stored in the false color evaluation value correction coefficient storage unit 110, and the value obtained by the subjective evaluation experiment is α = −0. 8273, β
= -54.1404, γ = -0.8221, ζ = -0.
0208, ξ = 4.0112, φ = 3.1891, ψ =
1.0663. The coefficients are determined according to the characteristics of the uniform color space used in order to obtain an objective evaluation value in good correlation with human subjectivity.

Next, the meaning of each term in equation (7) will be described. In the equation (7), (ave (L *)) γ in the first term is a brightness correction (brightness difference detection characteristic) relating to the color generation amount Δcolor, and (ave (L *)) φ in the second term is (Δcolo).
r) is a coefficient related to ξ. Since the image to be evaluated is composed of the ladder pattern as shown in FIG. 2, the false color becomes difficult to see due to the contrast of the ladder pattern itself. The fact that a small stimulus disappears due to the presence of such a large stimulus is called "masking".

In the equation (7), when the contrast of the ladder pattern is small, the second term ΔL * → 0, so that the effect of the first term increases. If there is no false color, ΔL *, Δa * b *
→ 0, false color evaluation value → 1 / α + ψ = -0.14
In the case where the false color is large, Δa * b * → ∞, so that the denominator of the first term of Expression (7) also becomes large, and the false color evaluation value → ψ = 1.0663.

On the contrary, if the contrast of the ladder pattern is large, ΔL * ≫0, and the effect of the second term is increased. That is, the second term represents the contribution of the magnitude of the contrast of the ladder pattern to the false color evaluation value. When there is no false color, ΔL * ≫0 and Δa * b * → 0, so the false color evaluation value → ζ / ((ave (L *)) φ) + ψ, and the average lightness (ave (L * )) Is small, the false color evaluation value → −∞, and if the average lightness (ave (L *)) is large, the false color evaluation value → φ
= 1.0663. If there is a false color, ΔL * ≫
0, Δa * b * → ∞, so that the false color evaluation value = ζ / ((s
qrt2) ξ · (ave (L *)) φ) + ψ, and if the average lightness (ave (L *)) is small, the false color evaluation value → −
け れ ば, if the average brightness (ave (L *)) is large, the false color evaluation value → φ = 1.0663. However, since the condition that the contrast is large and the average brightness is small is not actually assumed, in this case, the false color evaluation value → 0.

The reason for constructing the evaluation formula in this way is that “when the subject is strictly achromatic and has no contrast,
This is because no false color is generated by the color interpolation processing (see the color interpolation processing unit 151 in FIG. 7). On the other hand, "the smaller the contrast of the evaluated image, the more noticeable the false color". Equation (7) is valid even if the evaluated image has an arbitrary contrast. Note that the contribution rate to the subjective evaluation value (correlation coefficient of 2
Squared) was 0.929.

As described above, the image quality method and the image quality apparatus of the present invention take into account human visual characteristics, particularly, visual characteristics based on lightness. It is possible to calculate an evaluation value that matches the evaluation.

Expression (7) shows an example of the false color evaluation value, and may be derived by the following expression (8).

[0086]

(Equation 3)

[0087]

As described above, in the image evaluation method of the present invention (claim 1), in the conversion process corresponding to the display device characteristics, the characteristics of the image display device for inputting and displaying image data from the color image input device are displayed. In the uniform color space component calculation step, the evaluated image converted in the display device characteristic correspondence conversion step is converted into a uniform color space component, and in the uniformity color space component calculation step, the statistic calculation step is performed. Calculate the calculated statistic of the uniform color space component, and in the brightness difference detection characteristic correction process, multiply the statistic of the brightness component calculated in the statistic calculation process by a weighting factor that increases exponentially according to the average brightness. A correction corresponding to the lightness difference detection characteristic of the system is performed, and a false color evaluation value is calculated using a statistic corrected in the lightness difference detection characteristic correction step and a statistic of the chromaticity component in a false color evaluation value calculation step. For solid-state imaging device False color of a color image input apparatus using it becomes possible to quantitatively evaluate to match the subjective evaluation.

Further, the image evaluation method of the present invention (Claim 2)
Converts the image to be evaluated in accordance with the characteristics of the image display device which inputs and displays image data from the color image input device in the display device characteristics conversion process, and converts the display device characteristics in the uniform color space component calculation step. The evaluated image converted in the step is converted into a uniform color space component, and in the statistic calculation step, at least an average value (average lightness value) and a lightness value of each component of the uniform color space converted in the uniform color space component calculation step are calculated. ,
Calculate the statistical deviation of the lightness and the statistical deviation of the chromaticity, and in the false color evaluation value calculating step, use the power of the average lightness value calculated in the statistical amount calculating step as a coefficient to calculate the statistical deviation of the lightness and the statistical Since the evaluation value relating to the false color is calculated based on the deviation, it is possible to quantitatively evaluate the false color of the color image input device using the solid-state imaging device so as to match the subjective evaluation.

Further, the image evaluation method of the present invention (Claim 3)
In the conversion process corresponding to the display device characteristics, the evaluated image is converted by using at least one of the chromaticity of each primary color, the chromaticity of the white point, and the gamma characteristic as a parameter in the image display device. It is possible to obtain an evaluation value of image quality in consideration of an observation environment (display device).

Further, the image evaluation apparatus of the present invention (Claim 4)
The display device characteristic conversion unit converts the image to be evaluated according to the characteristics of the image display device that inputs and displays image data from the color image input device, and the display device characteristic conversion unit converts the image to be evaluated into an image. Converted in accordance with the characteristics of the display device, the uniform color space component calculating means converts the evaluated image converted by the display device characteristics corresponding conversion means into a uniform color space component,
The statistic calculation means calculates the statistic of the uniform color space component calculated by the uniform color space component calculation means, and the brightness difference detection characteristic correction means calculates the statistic of the lightness component calculated by the statistic calculation means, according to the average brightness. Multiplied by a weighting factor that increases exponentially, performs correction corresponding to the lightness difference detection characteristic of the visual system, and the false color evaluation value calculation means corrects the statistic amount corrected by the lightness difference detection characteristic correction means, Since the false color evaluation value is calculated using the quantity, the false color of the color image input device using the solid-state imaging device can be quantitatively evaluated so as to match the subjective evaluation.

Further, the image evaluation device of the present invention (claim 5)
The display device characteristic conversion unit converts the image to be evaluated in accordance with the characteristics of the image display device that inputs and displays the image data from the color image input device, and the uniform color space component calculation unit converts the display device characteristic conversion. The evaluated image converted by the means is converted into a uniform color space component, and the statistic calculating means converts at least the average value of the lightness (average lightness value) and the respective components of the uniform color space converted by the uniform color space component calculating means. Calculate the statistical deviation of the brightness and the statistical deviation of the chromaticity, and calculate the statistical deviation of the brightness and the statistical Since the evaluation value relating to the false color is calculated based on the statistical deviation, it is possible to quantitatively evaluate the false color of the color image input device using the solid-state imaging device so as to match the subjective evaluation.

Further, the image evaluation device of the present invention (claim 6)
Since the display device characteristic conversion unit converts the evaluated image using at least one of the chromaticity of each primary color, the chromaticity of the white point, and the gamma characteristic as a parameter in the image display device, It is possible to obtain an evaluation value of image quality in consideration of an observation environment (display device).

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of an image evaluation device according to the present invention.

FIG. 2 is a diagram illustrating an example of a pattern having continuous frequency components, which is imaged when performing false color evaluation.

FIG. 3 is a diagram illustrating an example of a pattern having continuous frequency components, which is imaged when performing false color evaluation.

FIG. 4 is a diagram showing a resolution chart (ISO chart) for an ISO standard electronic still camera.

FIG. 5 is a diagram showing names of respective parts of the ISO chart.

FIG. 6 is a flowchart illustrating an image evaluation method relating to a false color.

FIG. 7 is a schematic configuration diagram of a digital camera showing a flow of image data.

FIG. 8 is a diagram showing a pixel arrangement (Bayer arrangement) of a CCD.

FIG. 9 is a diagram illustrating a boundary having a large contrast in which a false color easily occurs.

[Explanation of symbols]

 REFERENCE SIGNS LIST 11 imaging lens 12 solid-state imaging device 15 image processing unit 16 image compression unit 100 image evaluation unit 101 image storage unit 102 selection unit 103 display device characteristic correspondence conversion unit 104 uniform color space component calculation unit 105 statistic calculation unit 106 brightness difference detection characteristics Correction unit 107 False color evaluation value calculation unit 151 Color interpolation processing unit 152 Aperture processing unit

Claims (6)

[Claims] 1. An image evaluation method for inputting image data as an image to be evaluated from a color image input device using a solid-state imaging device and evaluating image quality relating to a false color of the image to be evaluated. A display device characteristic corresponding conversion step of converting the image to be evaluated in accordance with the characteristics of the image display device for inputting and displaying image data from the device; and a uniform color conversion of the image to be evaluated converted in the display device characteristic corresponding conversion step. A uniform color space component calculating step of converting to a space component; a statistic calculating step of calculating a statistic of the uniform color space component calculated in the uniform color space component calculating step; and a lightness component calculated in the statistic calculating step. A brightness difference detection characteristic correction step of multiplying the statistic by a weighting factor that increases exponentially according to the average brightness, and performing correction corresponding to the brightness difference detection characteristic of the visual system; and Image evaluation method of the statistic that is corrected by a positive step, and false color evaluation value calculating step of calculating a false color evaluation values using a statistical amount of the chromaticity components, characterized in that it comprises. 2. An image evaluation method for inputting image data as an image to be evaluated from a color image input device using a solid-state imaging device and evaluating image quality relating to a false color of the image to be evaluated. A display device characteristic corresponding conversion step of converting the image to be evaluated in accordance with the characteristics of the image display device for inputting and displaying image data from the device; and a uniform color conversion of the image to be evaluated converted in the display device characteristic corresponding conversion step. A uniform color space component calculating step of converting to a spatial component; and, from each component of the uniform color space converted in the uniform color space component calculating step, at least an average value of lightness (average lightness value) and a statistical deviation of lightness. A statistical amount calculating step of calculating a statistical deviation of the chromaticity; and a power of the average lightness value calculated in the statistical amount calculating step as a coefficient, based on the statistical deviation of the lightness and the statistical deviation of the chromaticity. Image evaluation method characterized by comprising a pseudo color evaluation value calculating step of calculating an evaluation value relating to false color can, a. 3. In the display device characteristic conversion step, the evaluated image is converted using at least one of chromaticity of each primary color, chromaticity of a white point, and gamma characteristic of the image display device as parameters. The image evaluation method according to claim 1, wherein the image evaluation is performed. 4. An image evaluation device for inputting image data as an image to be evaluated from a color image input device using a solid-state imaging device and evaluating image quality relating to a false color of the image to be evaluated. A display device characteristic correspondence conversion unit that converts the image to be evaluated in accordance with the characteristics of the image display device that inputs and displays image data from the device; and converts the image to be evaluated in accordance with the characteristics of the image display device. A display device characteristic correspondence conversion unit; a uniform color space component calculation unit for converting the evaluated image converted by the display device characteristic correspondence conversion unit into a uniform color space component; and a uniform color space calculated by the uniform color space component calculation unit. A statistic calculating means for calculating the statistic of the component; and a statistic of the brightness component calculated by the statistic calculating means multiplied by a weighting factor which increases exponentially in accordance with the average brightness. Calculating a false color evaluation value using a brightness difference detection characteristic correction unit that performs correction corresponding to the brightness difference detection characteristic, a statistic corrected by the brightness difference detection characteristic correction unit, and a statistic of the chromaticity component. And a false color evaluation value calculating means. 5. An image evaluation device for inputting image data as an image to be evaluated from a color image input device using a solid-state imaging device, and evaluating image quality relating to a false color of the image to be evaluated. A display device characteristic corresponding conversion unit for converting the image to be evaluated in accordance with the characteristics of the image display device for inputting and displaying image data from the device; and converting the image to be evaluated converted by the display device characteristic corresponding conversion unit into a uniform color. Uniform color space component calculating means for converting into a space component; and from each component of the uniform color space converted by the uniform color space component calculating means, at least an average value of lightness (average lightness value) and a statistical deviation of lightness. A statistic calculating means for calculating a statistical deviation of the chromaticity; and a power of the average lightness value calculated by the statistic calculating means as a coefficient, based on the statistical deviation of the lightness and the statistical deviation of the chromaticity. Image evaluation apparatus characterized by comprising a false color evaluation value calculating means for calculating an evaluation value relating to false color can, a. 6. The display device characteristic correspondence conversion unit converts the evaluated image using at least one of chromaticity of each primary color, chromaticity of a white point, and gamma characteristic of the image display device as a parameter. The image evaluation device according to claim 4, wherein conversion is performed.