JP2006331445A - Image processing apparatus and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method, which can simply adjust the contrast of a captured image using only the captured image having light and dark portions, and eliminate dependency on the input image and lost of color balance after conversion. <P>SOLUTION: The image processing apparatus includes an image input means 10 for converting input analogue image signals vi to digital image data, an image processing means 11 for executing desired image processing on the converted digital image data to produce processed image data v, an image synthesizing means 12 for synthesizing the processed image data with the digital image data to produce synthesized image data, and an image output means 13 for converting the synthesized image data to analogue synthesized image signals. The image processing means 11 is a contrast adjustment means or a color adjustment means. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method capable of automatically adjusting the contrast and color of a color image signal obtained by a camera or the like to obtain an image having a desired contrast and color tone.

  A color image taken with a digital camera is actually taken due to the influence of the SN level representing the noise ratio in the analog value obtained by the CCD element, which is an image sensor, and the conversion accuracy when the analog value is converted into a digital value. Since the natural image is limited to a range narrower than the dynamic range of the pixel density, there is a tendency that information in a shadowed detail is lost. This tendency is particularly significant when trying to photograph a sample in which a bright area and a dark area are mixed in the image. As an improvement, first, a method of performing contrast enhancement so as to expand the range of the brightness of the digital image from an image portion with higher brightness to an image portion with lower brightness can be considered. As a conventional method of contrast enhancement, a histogram showing the distribution of the luminance values of all the pixels that make up the original image is created, and the luminance value of the pixels in the original image is set to a new luminance using the histogram's cumulative curve as the luminance conversion curve. There is a histogram equalization technique that converts to a value and enhances the contrast of the image. In this method, since the luminance of the pixels in the entire original image region is converted to a new luminance using the same luminance conversion curve, a portion where contrast is lowered may occur. For this reason, when it is desired to perform contrast enhancement over the entire image, it is necessary to perform contrast enhancement processing suitable for the region.

  As a further improvement of this technique, many local histogram equalization techniques that divide an image into a plurality of rectangular areas and apply the histogram equalization technique for each area have been proposed (for example, Japanese Patent Laid-Open No. 2000-285230). Refer to the publication, reference 1), and its configuration diagram is as shown in FIG.

  FIG. 28 is a block diagram showing an image processing apparatus in Literature-1.

  FIG. 28 shows a contrast emphasizing unit, which includes an image data dividing unit 281 for dividing an image into rectangles, a histogram creating unit 282 for creating a histogram for each rectangle, and a contrast extending unit 283 for expanding the contrast for each rectangle. Become. However, when this method is used, problems have been pointed out, such as the occurrence of a rectangular region in which the contrast is too emphasized, or the contrast becoming discontinuous at the boundary between adjacent rectangular regions.

  On the other hand, as a solution to such a problem without using the histogram, the bright time and the dark portion are separately imaged by changing the shutter time and aperture of the digital camera for each field, and each information obtained is 1 A method has also been proposed in which a halftone density is realized by synthesizing with a single image so as to approach the dynamic range of the pixel density of an actually captured natural image. As an example, there is an example referred to in Japanese Patent Application Laid-Open No. 6-141229 (Document 2), and the configuration diagram of the apparatus is as shown in FIG.

  FIG. 29 is a block diagram showing an imaging apparatus in Document 2.

  In FIG. 29, reference numeral 290 denotes an image pickup device (CCD) that performs the photoelectric effect of a subject, 291 denotes a memory that records an image signal, 292 denotes multiplication means for multiplying the signal level by a constant, and 293 and 298 denote weights according to the level of the image signal. 295 is a speed converting means for converting the speed of the image signal, 296 is a level compressing means for compressing the level of the image signal, and 297 controls the timing of each block. Timing control means. This device performs weighted synthesis according to the signal level of two or more images having different charge accumulation periods in the image sensor, converts the resulting synthesized image output into the speed of a standard television signal, and also provides a reference for the television signal. Since the present invention relates to a television imaging apparatus that compresses to a level, it has a speed conversion means, a level compression means, and the like. Therefore, when applying to a digital camera, the speed conversion means and the level compression means are not necessary components. However, in the case of a method based on image synthesis obtained in a plurality of charge accumulation periods as in this device, contrast discontinuity is hardly generated in the synthesized image, but at least two images are taken continuously. Cannot take the same image. For this reason, when these images are combined, there is a possibility that an image in which details of the combined image are blurred or shifted is produced although it is affected by the shutter speed. Also, if the density range when shooting a bright part and the density range when shooting a dark part do not cover the entire density range in the image, there is a risk of discontinuity between the two intermediate density ranges. is there.

  By the way, when humans observe details and colors in such shadowed areas, human vision can perceive wide-range dynamics and colors of images without causing the above problems. Can do. The concept of central vision / peripheral vision Retinex centered on human vision is described by Edwin Land ("An Alternative Technology for the Design of the Designer in the Retinax Theorem Theorem Theorem Theorem Theorem Theorem Theorem in the Rethenology Theorem." of Science, vol. 84, pp. 3078 to pp. 3080 (1986)). In this, the Retinex concept of human vision is described by an inverse square function with a central visual field having a diameter of 2 to 4 basic units and a peripheral visual field having a diameter approximately 200 to 250 times the central visual field. ing. Then, the spatial average of the signal intensity in each of the central visual field and the peripheral visual field is defined as related to the perceived intensity. In recent years, a technique for improving the color and lightness expression in the dark part as described above has been proposed in accordance with these principles. This example is described in International Publication No. W097 / 45809 (published in Japanese translation table 2000-511315, Document 3).

  FIG. 30 is an explanatory diagram of the digital image improvement method in Document 3. Although a gray scale image is described here as an example, it can be extended to a color image. The pixel value I (i, j) in (i, j) of the image is adjusted by the processor 301 and the filter 302, and a filtering process is performed. For each pixel, the processor 301 calculates an adjusted pixel value I ′ (i, j) as in (Equation 1).

  Here, F (i, j) is a peripheral visual field function representing the peripheral visual field, and “*” indicates a convolution calculation process. Then, the normalization coefficient K is determined so that F (i, j) satisfies the condition of (Equation 2), whereby the second term of (Equation 1) is the average value of the pixel values in the peripheral visual field. It corresponds to.

  That is, (Equation 1) corresponds to a logarithmic conversion of the ratio of the pixel value of each pixel to the average pixel value in a large area. The peripheral visual field function F (i, j) is designed so that the contribution ratio increases as it approaches the target pixel from the correspondence with the human visual model, and a Gaussian function like (Equation 3) is applied.

  Here, c is a constant for controlling the adjusted pixel value I ′ (i, j) of each pixel value I (i, j).

  As described above, in the digital image improvement method of Document 3, the target pixel value with respect to the average pixel value in the peripheral visual field is calculated as the adjusted pixel value I ′ (i, j), and the display 303 Filtering is performed at 302 to generate the Retinex output R (i, j) used by. 302 converts I ′ (i, j) from a logarithmic area to a pixel value area of R (i, j) handled by the display 303, and is the same for all pixels for the sake of simplification of processing. A process that applies an offset and gain conversion function is used.

  However, this method is greatly affected by c that controls the peripheral visual field function. For example, if this c is a large value, the peripheral visual field contributing to the target pixel is increased, so that only color compensation in a large shadow is possible. On the other hand, if this c is a small value, only the vicinity of the target pixel is detected. Only an improvement in small shadow areas can be seen. Appropriate c must be taken into account according to the dynamic range of pixel values in the image to be handled in this way, and there is a problem of image dependency. As an improvement, a method for providing a plurality of peripheral visual field areas has been proposed in the same patent. Therefore, there is a problem that the processing time becomes enormous by preparing many small peripheral areas.

  Next, in the filter 302, processing for converting into actual pixel values used in the display 303 is performed. In this method, the same offset and gain conversion function processing is set for any image. Another problem is that empirical knowledge is required to set the optimum offset and gain conversion functions. Further, when the pixel value variation is extremely small within the maximum peripheral field of view set by a plurality of constants c, the adjusted pixel value I ′ (i, j) is I (i, j) even if a plurality of regions are prepared. ) Near 1.0. In such a case, I ′ (i, j) at the target pixel having a small variation in pixel value is often located in the vicinity of the average of I ′ (i, j) in the entire input image. Regardless of the conversion function, the actual pixel value tends to be near the center. In particular, in a uniformly wide image area having a highlight luminance, there is a problem that the luminance after adjustment is easily adjusted in a decreasing direction and visually deteriorates.

On the other hand, when color adjustment processing is used as image processing, the conventional color adjustment processing uses a method in which the color before conversion and the color after conversion are defined in advance in the color conversion table. In the case of a desire to make the displayed image more beautiful, the color before conversion is often not known until the input image is viewed, and cannot be determined uniformly in advance and color adjustment cannot be performed automatically. There is a point. As an improvement, a method of taking a color histogram in the input image and converting the higher-order color with the converted color in the color conversion table can be considered. However, simply taking the color histogram can provide a color balance after conversion. The possibility of collapse was great.
JP 2000-285230 A JP-A-6-141229 International Publication No. 97/45809 Pamphlet (Special Table 2000-511315) Edwin Land, “An Alternative Technology for the Computation of the Designer in the Retinex Theory of Color Vision, National Academy 84”. 3078-pp. 3080 (1986)

  As described above, the conventional contrast enhancement processing technique based on the histogram has a problem in that the contrast of a specific portion is excessively emphasized or discontinuity may be shown at a locally adjacent rectangular region boundary. It was. In addition, when a plurality of images captured under a plurality of aperture conditions are combined, in principle, the same subject cannot be captured, and there is a possibility that the composite image details may be blurred or color misalignment may occur. Had. Furthermore, with contrast adjustment technology based on the conventional human visual model, we have experience in designing filter processing when converting to constants for defining the peripheral visual field of humans and actual pixel values for final handling. Had the problem of including a lot of technical knowledge. In addition, particularly in a uniformly wide image area having highlight luminance, there is a problem that the luminance after adjustment is easily adjusted in a decreasing direction and visually deteriorates.

  In this image processing apparatus and image processing method, it is possible to easily adjust the contrast of the captured image using only the captured image having a bright and dark part, and to eliminate the dependence on the input image and the collapse of the color balance after conversion. It is required to be able to.

  In order to satisfy this requirement, the present invention can easily adjust the contrast of the captured image using only the captured image having a bright and dark part, and also eliminates dependency on the input image and collapse of the color balance after conversion. An image processing device that can be eliminated, and the contrast of the captured image can be easily adjusted using only the captured image having a bright and dark portion, and the dependency on the input image and the collapse of the color balance after conversion are eliminated. An object is to provide an image processing method.

  In an image processing apparatus that performs contrast adjustment on image data, means for obtaining a pixel value of a processing target pixel and pixel values of a plurality of pixels included in a peripheral region of the processing target pixel, and a pixel value of the processing target pixel And a contrast adjusting means for changing and outputting the pixel value of the pixel to be processed based on a comparison between the pixel values of the plurality of pixels included in the peripheral region or a converted value thereof, and image data from the output individual pixel values And the contrast adjusting means does not perform the contrast adjustment on the processing target pixel when a plurality of pixels included in the peripheral region uniformly show substantially the same pixel value. And outputting a pixel value of the pixel to be processed.

  With this configuration, the highlight (white) and black portions of the input image that are uniform in a large area are removed from the processing target, so that the brightness level that is likely to occur in the contrast-adjusted image in such an area as described above. Large fluctuations can be suppressed.

  In the following, pixel units are all used as the unit of pixel position (i, j). Further, the steps in the respective examples (embodiments) may be configured by hardware or software.

  FIG. 1 is a block diagram showing a basic configuration of an image processing apparatus according to an embodiment (embodiment) 1 of the present invention. 2A is a block diagram showing the image processing means in FIG. 1, FIG. 3 is a block diagram showing the contrast adjusting means in FIG. 2A, and FIG. 10 is a block diagram showing the image synthesizing means in FIG. FIG. 11 is an explanatory diagram schematically showing human vision.

  In FIG. 1, vi is an analog input image, vo is an output image that is finally output, and v is a processed image obtained by an image processing unit 11 described later. 10 is an image input means for converting an analog image signal vi from an image sensor such as a CCD element into digital image data, 11 is an image processing means for performing desired image processing on the digital image data obtained by the image input means 10, and 12 Is an image synthesizing means for synthesizing the digital image data from the image input means and the processed digital image data v obtained by the image processing means 11, and 13 is a final processed digital image data obtained by the image synthesizing means 12. Image output means for outputting a desired image (analog image signal) to a desired device (printer, display, etc.). In this embodiment (embodiment), the image processing in the image processing unit 11 is represented by a contrast adjusting unit 20 that performs contrast adjustment processing of an input image, as shown in FIG. . At that time, the contrast adjusting means is configured as shown in FIG. 3. In FIG. 3, reference numeral 30 denotes a color three-component value VPij (r (i, j), g (i, j) at the target pixel Pij (i, j). , B (i, j)) and contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) are calculated by comparison with surrounding pixels having a width c of Pij. Correction information deriving means 31 is an extracting means for limiting and extracting the effective range from the contrast adjustment amount VRPij obtained by the correction information deriving means 30, and 32 is the actual contrast adjustment amount VRPij selected by the extracting means 31. This is a pixel value conversion means for converting into an adjusted pixel value VPdij (rd (i, j), gd (i, j), bd (i, j)) in the pixel Pij. Further, as shown in FIG. 10, the image composition unit 12 gives priority to the input image vi or the adjusted image v obtained by the contrast adjustment unit 20 based on the luminance in the input image 1. Selection reference value determination means 100 for determining the image and the coupling coefficient ws (S = 1) applied to the input image vi and the adjusted image v obtained by the contrast adjustment means 20 based on the result of the selection reference value determination means 100. , 3; where 1 is a coupling coefficient applied to the input image vi, 3 indicates a coupling coefficient applied to the adjusted image v), and obtained by the coupling coefficient derivation means 101 The weighted average combining unit 102 generates a weighted average image of the input image vi and the adjusted image v obtained by the contrast adjusting unit using the coupling coefficients w0 and w1.

  The operation of the image processing apparatus configured as described above will be described.

  The color image vi is digitally input to the image processing unit 11 via the image input unit 10. In the case of a color image, the image input means 10 usually obtains component data of red r, green g, and blue b with the accuracy of the image input means 10 (with a value of 0 to 255 for 8 bits). The image input means 10 normalizes this value to a value from 0.0 to 1.0. Next, contrast adjustment processing for improving the contrast in the dark part of the input image is performed on the digital input image by the contrast adjustment means 20. The contrast adjusting unit 20 performs the process as shown in FIG. In human vision, as schematically shown in FIG. 11, the pixel information (color, contrast, etc.) of Pij is not recognized only by the pixel value perceived with respect to the target pixel Pij, but instead of the target pixel Pij. The pixel information of Pij is perceived by adjusting the pixel value according to the relative relationship with the information of the surrounding pixels. This is called the Retinex concept introduced by Edwin Land, as explained in the prior art, and the uneven illumination light that is only partially illuminated by such perception Even in a scene where the intensity of the pixel value changes extremely, the color of the object can be recognized with high accuracy. Also in the present embodiment (embodiment), the color and detail information in a dark part such as a shadow are clarified by using this concept.

  FIG. 12 is a flowchart showing the operation of the contrast adjusting means 20 of the image processing apparatus according to the present embodiment (embodiment).

  In FIG. 12, first, the correction information deriving unit 30 calculates the maximum value Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of each component of the pixel Pij in the input image (S1). . This is because when the minimum pixel value and the maximum pixel value in the output image vo are greatly different from those in the input image vi, the minimum pixel value of the pixel values in the output image vo This is a process for adjusting the maximum pixel value to the minimum pixel value and the maximum pixel value of the input image vi. Next, the pixel value VPij of the target pixel Pij is regarded as a central visual field in Land, and a region belonging to a rectangular region of c pixels around it is regarded as a peripheral visual field. Then, the correction information deriving means 30 obtains the weighted average pixel value VAPij (Ar (i, j), Ag (i, j), Ab (i, j)) of the pixel value in the peripheral visual field (S2), A contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) that leads to a relative relationship between VAPij and VPij is calculated (S3). As this VAPij, the pixel value VPij (r (i, j), g (i, j), b (i, j)) in the peripheral visual field of the c pixel in the two terms of (Equation 1) as in the prior art. And (Equation 2) and (Equation 3) can be defined by the convolution integral value of the peripheral visual field function F (x, y) defined by the Gaussian function, and the weighted average pixel value VAPij is also expressed by (Equation 1). ) Can be defined independently of the three components constituting the pixel value. However, in the present embodiment (embodiment), VAPij is a pixel value VPij (r (i, j) in the peripheral visual field of c pixel as in (Expression 4) in consideration of simplification and speeding up of processing. , G (i, j), b (i, j)) were defined. Further, the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) is not calculated by using the difference of the logarithmic conversion values as in (Equation 1) (Equation 5). Thus, the ratio of the pixel value VPij for each component to the weighted average pixel value VAPij is defined.

  The correction information deriving unit 30 performs the contrast adjustment amount VRPij for the target pixel Pij as described above on all the images in the input image (S4). Thereafter, the extraction unit 31 calculates an average value VaR (aRr, aRg, aRb) and a standard deviation amount VdR (dRr, dRg, dRb) for each component of the contrast adjustment amount VRPij (S5), and uses these values to adjust the contrast. A minimum value emin and a maximum value emax when extracted from the quantity VRPij are derived (S6). There are many methods for this derivation. Here, aRr + α × dRr, aRg + α × dRg, and aRb + α × dRb are obtained as emax candidates, and the maximum value among these three values is defined as emax. Then, aRr- [beta] * dRr, aRg- [beta] * dRg, and aRb- [beta] * dRb are obtained as candidates for emin, and the minimum value among these three values is defined as emin. By doing so, the necessary region is extracted so that the balance of the components of the extracted contrast adjustment amount VRPij is not lost. Next, the pixel value conversion means 32 uses the emax and emin to set the contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) to 0. The value is converted to a value within the range of 0 to 1.0 (S7), and a process of suppressing the converted value within Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of the input image is performed (S8). ). The contrast adjustment amount VRPij obtained in this way is regarded as a pixel value for contrast adjustment in the target pixel Pij, and the contrast adjustment processing is ended (S9). The pixel value conversion means 32 determines whether or not to end.

  By such a series of processing, only the portion near the center is extracted in the distribution of the ratio of the target pixel value to the weighted average pixel value in the peripheral visual field, the amount of variation from the center is emphasized, and the deviation from the center is greatly deviated. The contrast adjustment amount VRPij of the pixel having the ratio value is likely to be 1.0 or 0.0. For this reason, the region where the difference between the target pixel Pij as the central visual field and the pixels in the peripheral visual field is slightly enhanced, the difference is easily enhanced, and contrast enhancement is performed, and the details in the shadow and the range of the input device are insufficient. It becomes possible to emphasize and express the color information. In the case of the present embodiment (embodiment), the derivation of the contrast adjustment amount in each pixel is configured in a simpler form than the conventional method based on the Retinex concept. In the conventional method based on the Retinex concept, the setting of the filter processing (offset, gain conversion function) when converting the contrast adjustment amount in each pixel into the actual pixel value component requires an empirical knowledge. However, in the present embodiment (embodiment), it is mentioned as an advantage that it is not necessary. On the other hand, this contrast adjustment process is performed by extracting only the vicinity of the center in the distribution of the ratio of the target pixel value to the weighted average pixel value in the peripheral visual field, and saturating the area before and after that at 0.0 or 1.0. There is a possibility that the image vo after processing obtained in (1) may have a problem that the luminance level collects near the center as a whole. In this embodiment (embodiment), in order to improve this problem and the reduction of the luminance level in the highlight portion having a uniform color in a very large area described in the prior art, the input image vi and this By adaptively synthesizing the processed image vo, the reduction or increase in the luminance level inherent in the input image vi is suppressed.

  In the image synthesizing process in the image synthesizing unit 12, the processed image v and the input image vi in the contrast adjustment process are received and the process is executed as shown in FIG.

  FIG. 13 is a flowchart showing the operation of the image synthesizing means 12 of the image processing apparatus according to Embodiment 1 of the present invention.

  Here, first, the selection reference value determination unit 100 calculates the luminance y (i, j) of each pixel Pij in the input image (S11). Then, threshold values Th_h and Th_l prepared in advance are compared with each pixel luminance y (i, j) (S12). At this time, if Th_l> y (i, j) or y (i, j)> Th_h, the selection reference value determination unit 100 further calculates the average luminance Ay (i, j in the rectangular peripheral visual field region of c in Pij. ) Is obtained (S13). If y (i, j) satisfies | Ay (i, j) −y (i, j) | <Th_len (S14), the pixel value VPij (r (r (r) in the input image vi in the target pixel Pij). i, j), g (i, j), b (i, j)) is the value of the pixel Pij of the processed image vo, VWPij (WRr (i, j), WRg (i, j), WRb (i) , J)) is replaced (S15). As described in the related art, this process is a process that takes into account that the contrast adjustment amount in Pij when the fluctuation is very small in the area larger than the peripheral visual field area c is likely to be concentrated in the vicinity of 1.0. By removing the highlight (white) and black parts of the input image that are uniform over a large area as a processing object, this process removes the brightness level that is likely to occur in the contrast-adjusted image in such an area as described above. The purpose is to suppress large fluctuations.

  Next, in the case where the above condition is not satisfied, the coupling coefficient deriving unit 101 determines the luminance value y (i, j) in Pij of the input image vi and the luminance value yd (i, i, i) in the corresponding pixel of the processed image v. j) are compared (S16), and the absolute value len = | y (i, j) -yd (i, j) | of the error between the two is input to the prepared w3 decision function, thereby processing the processed image. The coupling coefficient w3 occupied by v is determined (S17). The coupling coefficient w1 occupied by the input image vi is determined from w1 = 1.0−w3. Although this w3 determination function is not uniformly determined, a threshold function g (X) as shown in (Equation 6) is used here for simple purposes.

  The weighted average combining means 102 uses the w1 and w3 to calculate the weighted average pixel value VWPij (WRr (i, j), WRg) of the pixel value VPij of the input image vi and the pixel value VRPij of the processed image v using Pij. (I, j), WRb (i, j)) is obtained (S17) and embedded in the corresponding pixel position of the image v (S18). By performing such a series of determination and weighted average processing for all the pixels (S19), an output image vo that is finally output by the image output means 13 is generated.

  By performing the configuration and processing as described above, the image processing apparatus according to the present embodiment (embodiment) solves the problems of the conventional technology and takes advantage of the advantages of the input image easily and accurately. Contrast adjustment can be performed.

  Note that the definition of the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) is not uniform. For example, as in (Equation 7), the pixel value for each component The value obtained by subtracting 1.0 from the ratio of VPij to the weighted average pixel value VAPij multiplied by a preset positive constant γ is the pixel value VPij (r (i, j), g (i, j)) of the input image. , B (i, j)) can be redefined as the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)).

  By doing so, the contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j), Rb () when the pixel value variation is very small in the peripheral visual field of the c pixel, which is a problem in the conventional technique, i, j)) each component of VPij (r (i, j), g (i, j), b (i, j)) can be avoided from being in the vicinity of 1.0. Although there is an advantage, it should be noted that there is a dependency due to the set γ. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the present embodiment (embodiment). Can be realized.

  The basic configuration of the image processing apparatus according to the embodiment (embodiment) 2 of the present invention is the configuration of FIG. 1 as in the embodiment (embodiment) 1, and the image processing means 11 is also as shown in FIG. Thus, the contrast adjusting means 20 is obtained. At that time, the contrast adjusting means 20 is configured as shown in FIG.

  In FIG. 4, the correction information deriving means 30, the extracting means 31, and the pixel value converting means 32 are the same as those in FIG. Reference numeral 40 denotes an initial setting means for setting initial conditions for performing correction information derivation (calculation of the contrast adjustment amount of the target pixel). The initial setting means 40 performs a process of setting the initial peripheral area size c0 to the size c of the peripheral visual field part to be compared with the target pixel Pij that is mainly the central visual field. Reference numeral 41 denotes an end determination condition for determining whether or not the contrast adjustment amount has been calculated for all of a plurality of peripheral visual field areas prepared in advance. Reference numeral 42 denotes a current process when the end determination condition 41 does not regard the end determination. Correction range changing means for changing the size c of the peripheral visual field area to the next candidate. Note that the image synthesizing means 12 is configured as shown in FIG. 10 as in the first embodiment (embodiment).

  The operation of the image processing apparatus configured as described above will be described with reference to FIG. 10, FIG. 14, and FIG. FIG. 14 is an explanatory diagram schematically showing the peripheral visual field region, and FIG.

  The color image vi is digitally input to the image processing unit 11 via the image input unit 10. In the case of a color image, the image input means 10 usually obtains component data of red r, green g, and blue b with the accuracy of the image input means 10 (with a value of 0 to 255 for 8 bits). The image input means 10 normalizes this value to a value from 0.0 to 1.0. Next, contrast adjustment processing for improving the contrast in the dark part of the input image is performed on the digital input image by the contrast adjustment means 20. The contrast adjusting unit 20 performs the process as shown in FIG. The feature of the present embodiment (embodiment) is that contrast adjustment is performed in a plurality of peripheral visual field regions prepared for the embodiment (embodiment) 1 as schematically shown in FIG. This is because the influence of the size of the dark part (shadow) existing in the image is reduced.

  Next, the operation of the contrast adjusting unit 20 will be described with reference to FIG.

  In FIG. 15, first, the initial setting means 40 calculates the maximum value Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of each component of the pixel Pij in the input image (S21). Next, the pixel value VPij of the target pixel Pij is regarded as a central visual field in Land, and a region belonging to a rectangular region of c pixels around it is regarded as a peripheral visual field. At that time, the initial setting means 40 first sets c0 as the peripheral visual field area within a plurality of peripheral visual field area sizes ck (k = 0, 1,..., Cnum−1) prepared in advance, such as c = c0. Set as. In this case, ck may be prepared in ascending order from the minimum size area, or may be prepared in descending order from the maximum size, but it is better to align the direction of size change. Here, assuming that the images are prepared in descending order from the maximum size, it is assumed that the detail improvement in the input image is performed while the peripheral visual field region is sequentially reduced. Next, in the correction information deriving unit 30, the weighted average pixel value VAPij_k (Ar_k (i, j), Ag_k (i) of the pixel value in the peripheral visual field with respect to the peripheral visual field in the rectangular area of c = ck currently set. , J), Ab_k (i, j)) is calculated (S22). Then, the end determination unit 41 determines whether or not the weighted average pixel value calculation in the peripheral field of view of the pixel Pij in the correction information deriving unit 30 is completed for all ck (S23). If it is determined that the processing has not been completed, the process proceeds to the correction information range changing unit 42, the currently set peripheral visual field region c is changed to the next candidate, and the weighted average pixel in the correction information deriving unit 30 is again displayed. Value calculation is performed. On the other hand, when the end determination unit 41 determines the end, the correction information deriving unit 30 calculates the weighted average pixel value VAPij_k (Ar_k (i, j), Ag_k (i, j)) of Pij for each peripheral visual field region ck. A weighted average value of Ab_k (i, j)) is obtained, and the value is set as all weighted average pixel values VAPij (Ar (i, j), Ag (i, j), Ab (i, j)) of Pij (S24). At this time, it is conceivable to add a weight according to the size of the peripheral visual field region ck to the weighted average pixel value of Pij by each ck. The average pixel value is adopted as the total weighted average pixel value of Pij. Then, the correction information deriving means 30 determines the ratio of the pixel value VPij of each component in Pij to the total weighted average pixel value VAPij as contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j), Rb (i , J)) (S25).

  In the subsequent processing, as in Example (Embodiment) 1, first, a minimum value emin and a maximum value emax are derived when extracted from the contrast adjustment amount VRPij (S26, S27, S28). Next, using emax and emin, each component of the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) is in the range of 0.0 to 1.0. (S29), and the converted value is suppressed to Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of the input image (S30). The contrast adjustment amount VRPij obtained in this way is regarded as the pixel value for contrast adjustment in the target pixel Pij (S30), and the contrast adjustment processing is ended (S30a).

  Similarly to the embodiment (embodiment) 1, the image composition processing unit also uses the luminance value y (i, j) in Pij of the input image 1 and the luminance value yd (i, i, i) in the corresponding pixel of the processed image 3. j) The weighted average pixel values VWPij (WRr (i, j), WRg (i) of the pixel value VPij of the input image vi and the pixel value VRPij of the processed image v using the coupling coefficients w1 and w3 obtained from j). , J), WRb (i, j)) are determined and embedded in the corresponding pixel position Pij of the processed image v. By performing such processing for all the pixels, an output image vo that is finally output by the image output unit 13 is generated.

  By performing the configuration and processing as described above, the image processing apparatus according to the present embodiment (embodiment) takes advantage of the characteristics of the image processing apparatus according to the embodiment (embodiment) 1 and changes the pixel values in the input image. The contrast of the input image can be automatically adjusted without being greatly affected by the size of the dark part such as the dynamic range and shadow, which leads to the efficiency of the image contrast adjustment. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the present embodiment (embodiment). Can be realized.

  The basic configuration of the image processing apparatus according to the embodiment (embodiment) 3 of the present invention is the same as that of the embodiment (embodiment) 1 as shown in FIG. The image processing means 11 also employs contrast adjustment means 20 as shown in FIG. At that time, the contrast adjusting means 20 is configured as shown in FIG. FIG. 5 is a block diagram showing the contrast adjusting means 20 constituting the image processing apparatus according to the present embodiment (embodiment).

  In FIG. 5, 31 is extraction means similar to FIG. 3, 50 is signal conversion means for converting the pixel value in the input image vi into a signal (target signal) to be subjected to contrast adjustment, and 51 is signal conversion means 50. Target correction information deriving means for obtaining the contrast adjustment amount of the target pixel Pij with respect to the obtained target signal, 52 is a target signal obtained by the signal conversion means 50 based on the contrast adjustment amount extracted by the extraction means 31 Is converted into an actual target signal (adjusted target signal), 53 is an adjusted target signal obtained by the target signal conversion means 52 and other signals obtained by the signal conversion means 50 ( Signal reverse conversion means for performing reverse conversion processing of the target signal adjusted with the target signal) to the actual pixel value.

  A feature of the present embodiment (embodiment) is that the pixel value VPij (r (i, j), g (i, j), b (i, j)) in the pixel Pij of the color image is converted into other plural signals. Then, the contrast adjustment amount is calculated for one of the signals in the same manner as in the first embodiment (embodiment), and is maintained as it is at the time of conversion from the contrast adjustment amount of the last adjusted target signal and the input image vi. The pixel value in the pixel Pij of the processed image in FIG. 1 is obtained by performing an inverse conversion process of the conversion process performed by the signal conversion unit 50 from other signals. Here, in the signal conversion means 50, for example, the lightness L (i, j) and hue a * () from the pixel value VPij (r (i, j), g (i, j), b (i, j)) of Pij. i, j), b * (i, j), and this lightness L (i, j) is used as a target signal during contrast adjustment. In this embodiment (embodiment), the color information in the input image can be held as it is while controlling the brightness level by holding the other hue signals as they are in the input image. The color balance of the processed image v obtained by the means 20 can reproduce the input image as it is, and can reduce the amount of comparison processing between the peripheral visual field area and the target pixel in the target correction information deriving means 51. .

  The operation of the image processing apparatus configured as described above will be described with reference to FIG. FIG. 16 is a flowchart showing the operation of the contrast adjusting means 20 of FIG.

  In FIG. 16, first, VPij (r (i, j), g (i, j), b (i, j)) in the pixel Pij of the input image 1 is converted into a signal VPLij (L (i, j, j), a * (i, j), b * (i, j)) by the signal conversion means 50, and hues a * (i, j) and b * (i) among the converted value signals in Pij. , J) are held (S31). Thereafter, the processing from the target correction information deriving unit 51 to the target signal converting unit 52 is performed using the lightness L (i, j) as the target signal. Note that L (i, j) is normalized from 0.0 to 1.0. Thereby, it is considered that the balance of the color information of the input image vi can be accurately maintained even in the processed image v. The target correction information deriving unit 51 calculates the maximum value Lx and the minimum value Ln of the lightness L (i, j) of the pixel Pij of the input image (S32). Next, the weighted average brightness AL (i, j) of the brightness L (i, j) in the peripheral visual field of the target pixel Pij is obtained (S33), and the ratio of L (i, j) to AL (i, j) is calculated. The brightness contrast adjustment amount RL (i, j) is calculated (S34). The calculation of the brightness contrast adjustment amount RL (i, j) for the target pixel Pij as described above is performed by the target correction information deriving unit 51 for all the pixels in the input image. The extraction unit 31 obtains the average value aRL and the standard deviation dRL of the lightness contrast adjustment amount RL (i, j) (S36), and extracts the lightness contrast adjustment amount RL (i, j) based on the values. The minimum value emin and the maximum value emax are determined (S37). The target signal converting means 52 converts the brightness contrast adjustment amount RL (i, j) to a value within the range of 0.0 to 1.0 using the emax and emin (S38), and converts the converted value into a value. A process of suppressing the input image to the maximum value Lx and the minimum value Ln is performed (S39). In the signal inverse conversion means 53, the brightness contrast adjustment amount RL (i, j) thus obtained is regarded as the brightness contrast adjustment value in the target pixel Pij, and the retained hues a * (i, j) and b * The pixel value VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) in the pixel Pij of the processed image v is used by the signal conversion means 50 using (i, j). The contrast-adjusted image v is generated (S40a). Since the lightness contrast adjustment amount RL (i, j) is converted to a value in the range of 0.0 to 1.0 in the signal reverse conversion means 53, the hues a * (i, j) and b are once. * It is also necessary to return to the signal level at the time of reverse conversion together with (i, j).

  By taking such a series of processes, the image processing apparatus according to the present embodiment (embodiment) can adjust contrast independently for three pixels in a color image as in the prior art and embodiments (embodiments) 1 and 2. Compared to the case, it becomes easier to maintain the color balance of the pixels in the input image, and the stability of the color in a dark part such as a shadow in the image obtained by the adjustment can be further maintained. Further, the contrast adjustment amount calculated by the relative comparison between the target pixel in the target correction information deriving means 51 and the pixel in the peripheral visual field is independent of the three components in the conventional technique and the embodiments (embodiments) 1 and 2. Although it was necessary to calculate, only one component of brightness is required, leading to a significant reduction in processing time. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the third embodiment (embodiment). Can be realized.

  The basic configuration of the image processing apparatus according to the embodiment (embodiment) 4 of the present invention is the same as that of the embodiment (embodiment) 1 as shown in FIG. The image processing means 11 also employs contrast adjustment means 20 as shown in FIG. At that time, the contrast adjusting means 20 is configured as shown in FIG.

  FIG. 6 is a block diagram showing the contrast adjusting means 20 constituting the image processing apparatus according to the present embodiment (embodiment). In FIG. 6, the signal conversion means 50, the supplementary note setting means 40, the target correction information derivation means 51, the correction range change means 42, the end determination means 41, the extraction means 31, the target signal conversion means 52, and the signal reverse conversion means 53 are shown in FIG. Since it is the same as that shown in FIG. As apparent from FIG. 6, the contrast adjustment processing in the contrast adjustment means in Example (Embodiment) 3 is limited to the target signal only in contrast adjustment processing having a plurality of peripheral visual fields in Example (Embodiment) 2. It is something that has been added.

  The operation of the contrast adjusting unit 20 in FIG. 6 will be described with reference to FIG. FIG. 17 is a flowchart showing the operation of the contrast adjusting means 20 of FIG.

  In FIG. 17, first, VPij (r (i, j), g (i, j), b (i, j)) in the pixel Pij of the input image vi inputted by the image input means 10 is obtained by the signal conversion means 50. It is converted into a signal VPLij (L (i, j), a * (i, j), b * (i, j)) in the La * b * space, and the hue a * (i, j) and b * (i, j) are retained, and L (i, j) is normalized from 0.0 to 1.0 (S41). The target correction information deriving unit 51 calculates the maximum value Lx and the minimum value Ln of the lightness L (i, j) of the pixel Pij of the input image (S42). Next, the target correction information deriving means 51 uses the lightness L (i, j) as a target signal, as in the second embodiment (embodiment) 2, and a plurality of peripheral visual field regions ck (k = 0, 1,. .., Cnum-1) is obtained (S43). Then, when the end determination unit 41 determines that the calculation of the weighted average brightness for all the predetermined peripheral visual field regions has ended (S44), the target correction information deriving unit 51 determines that the weighted average value is the total weighted average. The brightness AL (i, j) is set (S45), and the ratio of L (i, j) to the value is calculated as the brightness contrast adjustment amount RL (i, j) (S46). When the end determination unit 41 determines that all pixels have ended (S47), the extraction unit 31 obtains the average value aRL and the standard deviation dRL of the lightness contrast adjustment amount RL (i, j) (S48). Using the minimum value emin and maximum value emax of the determined brightness contrast adjustment amount RL (i, j), the brightness contrast adjustment amount RL (i, j) is a value within the range of 0.0 to 1.0. (S49, S50). The conversion value is suppressed within the maximum value Lx and the minimum value Ln of the input image, and the signal inverse conversion unit 53 uses the lightness contrast adjustment amount RL (i, j) thus obtained as the lightness contrast in the target pixel Pij. The pixel value VRPij (Rr (i, j) in the pixel Pij of the processed image 3 is used by using the hues a * (i, j) and b * (i, j) that are held as an adjustment value (S51). ), Rg (i, j), Rb (i, j)) (S52), a contrast-adjusted image v is generated (S53).

  As described above, the image processing apparatus according to the present embodiment (embodiment) uses a plurality of contrast adjustment amounts calculated by comparing the target pixel value with the average pixel value in the surrounding area in the embodiment (embodiment) 3. This is an extension to the weighted average value of the contrast adjustment amount in the peripheral area, and there is an advantage that it is not necessary to predetermine the peripheral visual field area, which is a feature of the embodiment (embodiment) 2, and the embodiment (embodiment). As shown in FIG. 3, the color balance of the pixels can be easily maintained, and the color stability in a dark portion such as a shadow in the image obtained by the adjustment can be further maintained. In addition, there is an advantage that processing is simplified and processing time can be shortened.

  Further, these processes are performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the color image processing method according to the fourth embodiment (embodiment). It can be realized similarly.

  The basic configuration of an image processing apparatus according to Example (Embodiment) 5 of the present invention is the same as that of Example (Embodiment) 1 as shown in FIG. Further, the image processing means 11 employs the contrast adjusting means 20 as shown in FIG. 2A, and the contrast adjusting means 20 has a configuration as shown in FIG. This embodiment (embodiment) is different from embodiment (embodiment) 1 in that the pixel value conversion means 32 in the contrast adjustment means 20 in FIG. 3 is configured as shown in FIG. It is.

  FIG. 7A is a block diagram showing the pixel value conversion means 32 according to the present embodiment (embodiment). As shown in FIG. 7A, the pixel value conversion unit 32 is based on the average luminance calculation unit 70 that calculates the average luminance in the input image vi and the average luminance obtained by the average luminance calculation unit 70. In accordance with the conversion method classification means 71 for selecting the conversion method for converting the contrast adjustment amount obtained by the extraction means 31 into actual pixel values, and the conversion method obtained by the conversion method classification means 71, the extraction means 31 A pixel value estimation unit 72 that converts the obtained contrast adjustment amount into a pixel value of an actual pixel is included as a component. The feature of the present embodiment (embodiment) is that, in the embodiment (embodiment) 1, the value obtained by relative comparison with the pixels in the peripheral visual field based on the luminance of the input image is used as the actual pixel. A conversion method for converting to a value is selected.

  In the contrast adjusting unit 20 in FIG. 3, the flow of processing other than the pixel value converting unit 32 is the same as in the first embodiment (Embodiment) 1, and the pixel value of the pixel Pij in the input image is changed to the peripheral visual field region of the c pixel. First, the contrast adjustment amount is obtained by dividing by the weighted average pixel value at. An effective region is extracted from the contrast adjustment amount based on the histogram distribution of each component constituting the contrast adjustment amount, and each component is converted to a value within 0.0 to 1.0. Thereafter, pixel value conversion processing in the pixel value conversion means 32 is executed instead of step S8 in FIG. 12, and the processing is as shown in FIG.

  Next, the operation of the pixel value conversion means 32 in FIG. 7A will be described with reference to FIG. FIG. 18 is a flowchart showing the operation of the pixel value conversion means 32 in FIG.

  In FIG. 18, first, the average luminance calculating means 70 calculates the luminance y (i, j) of the pixel Pij in the input image and obtains the average luminance ave_y (S61). Next, the conversion method classification unit 71 selects a method in which the contrast adjustment amount is converted into an actual pixel value with the value of ave_y. In this case, many methods are conceivable. Here, a low luminance threshold Th_low and a high luminance threshold Th_high are provided, and the contrast adjustment amount is converted into an actual pixel value component as shown in (Equation 8). The conversion function gg (x) is selected (S62 to S66).

  In Equation 8, k_l, g_l, g_n, k_h, and g_h are preset constants, and X represents a target component value of the contrast adjustment amount before conversion. This determines the portion to be emphasized based on the average luminance in the input image before processing. For example, when the entire input image has a low luminance, it is better from the viewpoint of stereoscopic effect to leave as low a luminance portion as possible rather than the entire image suddenly brightening. In addition, when the entire input image has high luminance, human attention to the high luminance portion is high, and it is necessary to hold that portion as much as possible. When the image has intermediate luminance as in a natural image, a method of making the luminance portion lower than the center slightly darker and making the luminance higher than the center slightly brighter is performed by gamma conversion of input / output values at the time of printer output or the like. The processing here is in accordance with these findings, and if it is a conversion method that conforms to such knowledge, it is considered that the method shown here is not uniformly determined but can be used similarly. Finally, the contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j), extracted by the extraction means 31 using the conversion function selected by the conversion method classification means 71 by the pixel value estimation means 72. The conversion value VRPoij (Rro (i, j), Rgo (i, j), Rbo (i, j)) of each component of Rb (i, j)) is obtained (S67, S68). This is suppressed within the maximum value Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of each component of the pixel Pij in the input image, and the adjusted pixel value in the pixel Pij of the processed image v As a result, the processed image v is obtained (S69, S70). The image synthesizing unit 12 obtains a weighted average image of the image v and the input image vi, and this is the output image vo that is finally output.

  As described above, the image processing apparatus according to Example (Embodiment) 5 seems to be effective from the contrast adjustment amount obtained by comparison with surrounding pixels in the pixel value conversion unit 32 of Example (Embodiment) 1. The pixel value estimating means 72 is provided for selecting and applying a conversion method for converting the extracted contrast adjustment amount based on the average luminance in the input image after the extracted region is extracted. By doing so, it is possible to select an appropriate pixel value conversion according to the luminance distribution in the input image, and it is possible to enhance contrast adjustment in the image after adjustment.

  In this embodiment (embodiment), the contrast adjusting means is described in the case of the configuration of FIG. 3 described in embodiment (embodiment) 1. However, as shown in FIG. It is also possible to combine the pixel value conversion means in this example (embodiment) with the contrast adjustment means used in Example (embodiment) 2. Further, these processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the embodiment (embodiment) 5. Can be realized.

  The basic configuration of an image processing apparatus according to Example (Embodiment) 6 of the present invention is the same as that of Example (Embodiment) 1 as shown in FIG. The contrast adjusting means set as the image processing means 11 has a configuration as shown in FIG. The difference between Example (Embodiment) 6 and Example (Embodiment) 3 is that target signal conversion means 52 in contrast adjustment means 20 in FIG. 5 is configured as shown in FIG. 7B. It is.

  FIG. 7B is a block diagram showing the target signal conversion means 52 of FIG.

  As shown in FIG. 7B, the target signal conversion unit 52 is obtained by an average target signal calculation unit 73 that calculates the average value of the flat target signal in the input image vi and the average target signal calculation unit 73. A conversion method classification unit 71 that selects a conversion method for converting the contrast adjustment amount of the target signal obtained by the extraction unit 31 into an actual target signal value based on the average value of the target signal, and a conversion method classification unit In accordance with the conversion method obtained in 71, the target signal estimation means 74 for converting the contrast adjustment amount of the target signal obtained by the extraction means 31 into an actual target signal value is included as a component. The feature of this example (embodiment) is that, in Example (embodiment) 3, based on the target signal of the input image, the value obtained by relative comparison with the target signal in the peripheral visual field is actually The conversion method for converting to the target signal value is selected.

  The flow of contrast adjustment processing in the present embodiment (embodiment) is processed in substantially the same manner as in FIG. 16 representing the embodiment (embodiment) 3. Compared to Example (Embodiment) 3, the processing of the target signal conversion means 74 of FIG. 7B is added instead of step S40 of FIG. The processing in the target signal conversion means 74 in FIG. 7B is performed by changing the processing in the pixel value conversion means 32 in FIG. 7A used in the embodiment (embodiment) 5 to the lightness that is the target signal. What has been changed to be performed on is performed. That is, in contrast to the pixel value conversion processing of Example 5 (Embodiment) 5 shown in FIG. 18, the conversion function is applied only to the target signal in this Example (Embodiment), and extraction is performed. In the embodiment (embodiment) 5, the average luminance of the input image is used to select the conversion method of the contrast adjustment amount extracted by the means 31. However, in this embodiment (embodiment), the input image is input by the signal conversion means 50. The difference is that an average value of brightness as a target signal is used among brightness and hue obtained. The reason why the average value of the target signal is used for selecting the conversion method of the contrast adjustment amount extracted by the extraction means 31 is that the calculation of the contrast adjustment amount is originally performed in the third embodiment (embodiment). The brightness RL (i, j) of the adjusted pixel (i, j) obtained through the target signal conversion means 52 and the lightness RL (i, j) obtained through the target signal conversion means 52 and the signal conversion means 50 are obtained. Since the desired contrast-adjusted image v is obtained by inversely transforming the obtained hue, there is a point that the brightness has the same meaning as the luminance from the viewpoint of simplification of processing.

  Such an image processing apparatus according to the present embodiment (embodiment) is effective from the contrast adjustment amount of brightness obtained by comparison with surrounding pixels in the target signal conversion means 50 of the embodiment (embodiment) 3. After extracting a possible region, a conversion method for converting the contrast adjustment amount of the extracted target signal is selected and applied based on the average brightness in the input image. Therefore, the present embodiment (embodiment) is for improving the color stability of the pixel as in the embodiment (embodiment) 3 and further improving the accuracy of the contrast in the embodiment (embodiment) 5. It is.

  In the present embodiment (embodiment), the contrast adjusting means 20 has been described in the case of the configuration of FIG. 5 described in the embodiment (embodiment) 3. However, as shown in FIG. It is also possible to combine the target signal conversion means 52 in the present embodiment (embodiment) with the contrast adjustment means 20 used in the embodiment 4 (embodiment). Further, these processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the present embodiment (embodiment). Can be realized.

  The basic configuration according to the embodiment (embodiment) 7 of the present invention is the configuration shown in FIG. The contrast adjusting means 20 which is the image processing means 11 has a configuration as shown in FIG. As shown in FIG. 3, the contrast adjusting unit 21 includes a correction information deriving unit 30, an extracting unit 31, and a pixel value converting unit 32.

  FIG. 8A is a block diagram showing the image conversion means 32. As shown in FIG. 8A, the image conversion means 32 uses the reference intensity calculation means 80 for calculating the luminance edge intensity representing the contrast information in the input image vi, and the contrast adjustment amount obtained by the extraction means 31 as the actual pixel. A conversion curve estimation means 81 for estimating a conversion curve at the time of conversion into a value by a genetic algorithm, and a contrast adjustment obtained by the extraction means 31 based on the conversion curve of the pixel value estimated by the conversion curve estimation means 81 A pixel value estimation unit 72 that converts the quantity into an actual pixel value is included as a component.

  FIG. 9A is a block diagram showing a conversion curve estimation means 81 prepared for genetic algorithm estimation. As shown in FIG. 9A, the conversion curve estimation means 81 includes initial candidate setting means 90 for preparing a chromosome set indicating a plurality of conversion curves and initial conditions necessary for estimation, and conversion obtained from each chromosome. A pixel value conversion candidate calculation unit 91 for obtaining an actual pixel value based on a curve, and a luminance edge of an image generated by each chromosome using pixel values from each staining obtained by the pixel value conversion candidate calculation unit 91 The evaluation value calculation means 92 for calculating the evaluation value of each conversion curve from the intensity and the luminance edge intensity obtained from the input image, and the fitness for calculating the fitness of each chromosome from the evaluation value obtained by the evaluation value calculation means 92 Degree calculation means 93, selective selection of each chromosome set obtained by fitness calculation means 93, recombination operation means 94 for generating the next chromosome group by performing crossover and mutation recombination operations, and conversion curve optimization Estimation is finished Or performed if the determination has, in the time of completion having as a component the estimated end determining unit 95 to pass the optimum conversion curve to the next pixel value estimation unit 72.

  The feature of the present embodiment (embodiment) is that the value obtained by relative comparison with the target signal in the peripheral visual field obtained in embodiment (embodiment) 1 is converted into an actual pixel value. The image is obtained by further improving the sharpness of the edge portion of the image obtained using the genetic algorithm with the contrast adjustment amount. Example 5 is prepared in advance with the average luminance of the input image. Compared to selecting from a plurality of conversion methods (conversion functions), it is possible to automatically estimate a more appropriate conversion function, so it is possible to set conditions for classifying the average luminance in advance and multiple conversions There is no need to select a function, and there is an advantage that the dependence on the input image is very small.

  In the flow of processing in the present embodiment (embodiment), the image input means 10, the image composition means 12, and the image output means 13 are the same as those in the embodiment (embodiment) 1 and will not be described.

  FIG. 19 is a flowchart showing a process flow in the contrast adjustment unit 20 as the image processing unit 11, and FIG. 20 is a flowchart showing a process flow in the conversion curve estimation unit 81 in the contrast adjustment process. Contrast adjustment processing in the present embodiment (embodiment) is shown in accordance with these drawings.

  In FIG. 19, first, the correction information calculation means 30 calculates the maximum value Vmax (rx, gx, bx) and the minimum value Vmin (rn, gn, bn) of each component of the pixel Pij in the input image (S71). . Next, the pixel value VPij of the target pixel Pij is regarded as a central visual field in Land, and a region belonging to a rectangular region of c pixels around it is regarded as a peripheral visual field. Then, the correction information calculation means 30 obtains a weighted average pixel value VAPij (Ar (i, j), Ag (i, j), Ab (i, j)) of the pixel value in the peripheral visual field (S72). A contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) that leads to a relative relationship between VAPij and VPij is calculated (S73). As VAPij, as in the prior art (embodiment) 1, as in the prior art, pixel values Vpij (r (i, j), g (i, j), b (i , J)), and the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) is a pixel for each component as shown in (Equation 5). The ratio of the value VPij to the weighted average pixel value VAPij is defined.

  The calculation of the contrast adjustment amount VRPij for the target pixel Pij as described above is performed for all images in the input image (S74). Thereafter, the extraction unit 31 extracts the minimum value when the contrast adjustment amount VRPij is extracted from the contrast adjustment amount VRPij from the average value VaR (aRr, aRg, aRb) and the standard deviation amount VdR (dRr, dRg, dRb) for each component of the contrast adjustment amount VRPij. The emin and the maximum value emax are derived, and the components of the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) are set to 0.0 using the emax and the emin. To a value within the range of 1.0 (S75 to S77). The steps so far are the same as those in the first embodiment (embodiment). Thereafter, conversion curve estimation processes S78 and S79 of the extracted contrast adjustment amount are performed using a genetic algorithm as shown in FIG.

  The conversion curve estimation unit 81 includes an initial candidate setting unit 90 to an estimation end determination unit 95, and processing in the middle corresponds to a method in which local search capability is supplemented to an optimal solution search method called a genetic algorithm. The genetic algorithm is described in, for example, documents such as “Genetic Algorithms in Search, Optimization and Machine Learning” (David E. Goldberg, Addison Wesley), etc. Even if differential information of the evaluation function to be obtained is not obtained, if the evaluation value itself is obtained, the search can be performed, and it can be applied to a target problem that has only a human subjective evaluation criterion and is effective. Sex is a technique that has been reported so far. Specifically, the process proceeds as follows. The process will be described for each stage with reference to FIGS. FIG. 21 is a schematic diagram schematically showing the chromosome structure used in the genetic algorithm in the conversion curve estimation means 81, and FIG. 22 is a schematic diagram schematically showing the outline of the roulette selection method in the conversion curve estimation means 81, FIG. 23 is a schematic diagram schematically showing a recombination operation process of the genetic algorithm used in the conversion curve estimation means 81.

{Step S81 to Step S83}
First, consider a set Gv consisting of Nch elements based on the search vector Vch [A] composed of the n parameters with g0 to gn-1 in FIG. 21 as search parameters. Each element g−k [A] (k = 0,..., N−1) of the search vector Vch [A] is called a gene because of the relationship with the organism, and the search vector Vch [A] be called. When using the genetic algorithm, first, the initial candidate setting means 90 performs a process for appropriately generating the initial set Gv of search vectors. In the initial candidate setting means 90, various methods are conceivable as chromosome setting methods. Here, for the contrast adjustment amount ik before conversion existing in the range of 0.0 to 1.0 as shown in FIG. Then, a conversion curve is set on the assumption that the converted output amount ok is converted to a value in the range of 0.0 to 1.0. Then, the chromosome notation of the search vector is performed using a real number coding method in which n pieces of g-k obtained from g-k = o, k-o, k-1 are used as they are. g-0 is assumed to be g-0 = o-0. That is, the conversion curve from 0.0 to 1.0 is quantized into n pieces of data, and the value obtained by each quantization is used as a search parameter that is estimated by a genetic algorithm. However, conversion of the search parameter for the conversion curve is not uniquely determined, and other conversion is possible. In addition, the chromosomal structure used here can also handle a binary coding method in which each gk is displayed in binary and the binary numbers are arranged.

  Initial candidate setting means 90 first sets one initial search vector v by randomly selecting n real numbers from within a range of −1 to 1 using uniform random numbers. By preparing Nch number of preset chromosome sets, an initial search vector set Gv = {Vch [A]} (A = 0,..., Nch−1) is set. The optimum solution search is started from the initial search vector set Gv. Then, in order to evaluate each search vector Vch [A], the luminance y (i, j) and color differences u (i, j), v (i, j) in all the pixels Pij of the input image 1 are calculated. Then, the edge strength Ec of the input image is obtained from this luminance. This Ec is defined as (Equation 9), and is defined as the sum of the edge intensities ec (i, j) in the pixel Pij calculated by the Sobel operator.

{Step S84}
Using the conversion curve obtained from each element (chromosome) of the set Gv, the contrast adjustment amounts VRPij (Rr (i, j), Rg (i, j) obtained by the extraction means using each search vector Vch [A]. , Rb (i, j)) is output to VRPoij [A] (Rro (i, j, A), Rgo (i, j, A), Rbo (i, j, A)). Then, the adjusted luminance value yo (i, j, A) is calculated from this value. Then, the edge intensity E [A] of the adjusted image v candidate is obtained from the adjusted luminance. This E [A] is obtained by (Equation 9) similarly to Ec. These processes are performed by the pixel value conversion candidate means 91.

{Step S84}
The goodness as a solution of each search vector Vch [A] obtained in the previous step is evaluated according to a preset evaluation scale, and the evaluation result of the search vector Vch [A] is expressed as an evaluation value E [A]. Here, the evaluation scale set in advance is referred to as an evaluation function (fitness function). This process is performed by the evaluation value deriving unit 92. Many evaluation functions E [A] are conceivable. As shown in (Equation 9), the edge strength of the adjusted image v candidate obtained with each search vector is used. This is because an image that is visually perceived as having good contrast is not only bright and dark, but also has a sharp edge portion of the image.

{Step S85, Step S86}
In the fitness calculation means 93, fitness is calculated from the evaluation value E [A] obtained by the evaluation value deriving means 92 in order to see the fitness of each search vector. Various functions are conceivable as functions for deriving the fitness f [A]. The brightness yo at each pixel of the adjusted image v candidate is added to the edge intensity E [A] of the adjusted image v candidate obtained by each search vector. A value obtained by dividing the value obtained by adding the second term relating to the distance between the maximum value yox [A] and the minimum value yon [A] of (i, j, A) by the edge strength Ec of the input image is the fitness f [A]. Used as. Here, φ is a constant.

  A larger value of f [A] is determined as a better search vector, that is, a conversion curve. The second term not only improves the edge strength, but also has as many conversion output regions as possible.

{Step S87 to Step S90}
The search vector set selected in step S86 is subjected to genetic recombination operations such as crossover and mutation to create a new search vector set. The number Nch of search vectors included in the search vector set is constant here, but may increase or decrease. This process is performed by the rearrangement operation means 94. Here, selective selection is performed by a roulette selection method in which a search vector is selected with a probability proportional to the figure matching degree. The outline of the roulette selection method is as follows.

{Roulette selection method}
(I) The fitness f [A] of the search vector Vch [A] (A = 0,..., Nch−1) belonging to the set Gv and the sum F of the fitness of all search vectors are obtained.

  (Ii) The selection probability h [k] that is selected as a parent for generating the next generation search vector by Vch [A] is determined as shown in FIG. In order to assign this probability to the search vector Vch [A], for example, the following method can be considered.

  (Iii) The selection range I [A] of each search vector is assigned to the interval in [0, 1] as shown in FIG. Here, a set RR = (ra [0], ra [1],..., Ra [Nch-1]) of uniform random numbers ra [A] is generated in [0, 1]. Then, as shown in FIG. 22D, a set of num [B] = I [A] satisfying (Expression 11) for ra [B] in the RR Num = (num [0], num [1], .., Num [Nch-1]), a set of Nch search vectors corresponding to this Num is selected.

  By such a roulette selection method, the search vector Vch [A] in the current search vector group Gv is selected.

  Next, crossover processing and mutation processing as shown in FIG. 23 are performed on the new search vector group Gv thus obtained. Crossover is an operation of creating a new search vector by replacing a part of the search vector represented by a finite symbol with a part of another adjustment vector as shown in FIG. Further, the mutation is an operation of changing a part of the component of the search vector selected from the search vector set to another symbol with a certain low probability as shown in FIG. Crossover processing corresponds to a search at a position far away from the current solution vector, and mutation corresponds to a search in the vicinity of the current search vector. Through these two processes, the genetic algorithm creates a new set of search vectors. Various methods have been proposed for processing such as crossover and mutation. In this embodiment (embodiment), one-point crossover or two-point crossover that replaces two points as shown in FIG. 23 is used. . Further, as a mutation, a mutation is performed by selecting each gene (g−k) constituting the search vector with a low probability from the new search vector group Gv obtained through the crossover process, and adding a random constant. Do it. At that time, we tried to maintain the diversity of search vectors by changing the probability of mutation in the half and the other half of the search vector group.

  It is determined whether or not the number of iterations exceeds the allowable number of iterations as a convergence condition for the search vector optimization process that represents the transformation curve by these genetic algorithms. The process moves to. By repeatedly executing the above processing steps until the convergence condition is satisfied, the search parameters constituting the optimum conversion curve are estimated. In this way, a large number of candidates are prepared globally, and extracted accurately and automatically by repeatedly performing symbolic recombination operations such as crossover and mutation that are difficult to trap in the local optimal solution (local minimum). It is possible to estimate an optimal conversion curve for converting the contrast adjustment amount obtained by the means 31.

  Using the conversion curve obtained as described above, the pixel value estimating unit 72 extracts the contrast adjustment amount VRPij (Rr (i, j), Rg (i, j), Rb (i, j)) extracted. Is converted into VRPoij (Rro (i, j), Rgo (i, j), Rbo (i, j)), and the maximum value Vmax (rx, gx, bx) of each component of the pixel Pij in the input image and Processing is performed to keep the value within the minimum value Vmin (rn, gn, bn). VRPoij (Rro (i, j), Rgo (i, j), Rbo (i, j)) obtained here is generated as a pixel value in the pixel Pij of the processed image 3.

  As described above, the image processing apparatus according to the present embodiment (embodiment) seems to be more effective than the contrast adjustment amount obtained by comparison with the surrounding pixels in the pixel value conversion unit 32 of the embodiment (embodiment) 1. After extracting the region to be detected, the sum of the edge intensities obtained from the luminance of the input image is calculated. Next, prepare a plurality of conversion curves for converting the extracted contrast adjustment amount, calculate the sum of the edge intensity of the luminance of the final output candidate image obtained from each conversion curve, and the edge of the input image of that value A conversion curve estimation means 81 for estimating an optimal conversion curve by a genetic algorithm using the ratio to the intensity sum as an evaluation value is provided. In this way, when estimating the conversion function for converting the extracted contrast adjustment amount, the contrast can be further enhanced by providing the evaluation value with the degree of improvement of the edge strength related to the clarity of detail. At the same time, it leads to sharp edges. Further, by using a genetic algorithm for estimating such an optimal conversion function, an appropriate conversion function can be automatically estimated without depending on the input image.

  In the present embodiment (embodiment), the contrast adjusting means 20 has been described in the case of the configuration of FIG. 3 described in the embodiment (embodiment) 1. However, as shown in FIG. It is also possible to combine the pixel value converting means 32 in the present embodiment (embodiment) with the contrast adjusting means 20 used in the embodiment (embodiment) 2. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the present embodiment (embodiment). Can be realized.

  The basic configuration according to the embodiment (embodiment) 8 of the present invention is the configuration shown in FIG. The contrast adjusting means 20 as the image processing means 11 has a configuration as shown in FIG. The present embodiment (embodiment) is different from the embodiment (embodiment) 3 in that the target signal converting means 52 in the contrast adjusting means in FIG. 5 is configured as shown in FIG. is there.

  FIG. 8B is a block diagram showing the target signal conversion means 52 of FIG.

  As shown in FIG. 8B, the target signal conversion unit 52 includes a reference intensity calculation unit 80 that calculates the edge strength of the target signal representing the contrast information in the input image vi, and the target obtained by the extraction unit 31. A target signal conversion curve estimation unit 82 for estimating a target signal conversion curve when converting a signal contrast adjustment amount into an actual target signal value by a genetic algorithm, and a target signal estimated by the target signal conversion curve estimation unit 82 Based on the conversion curve, a target signal estimation unit 74 that converts the contrast adjustment amount of the target signal obtained by the extraction unit 31 into an actual target signal value is included as a component. The target signal conversion curve estimation means 82 is configured to execute a genetic algorithm in the same manner as in Example (Embodiment) 7, and the configuration diagram thereof is shown in FIG. The target signal conversion curve estimation means 82 is obtained by the initial candidate setting means 90, the target signal conversion candidate calculation means 96 for obtaining the actual target signal obtained from each conversion curve candidate, and the target signal conversion candidate calculation means 96. Evaluation value calculation means 92 for calculating an evaluation value of each conversion curve from the edge strength of each candidate obtained from the target signal and the edge strength obtained from the target signal of the input image, and each of the evaluation value calculation means 92 obtained by the evaluation value calculation means 92 The fitness calculation means 93 for calculating the fitness from the evaluation value, the selective selection of each candidate set obtained by the fitness calculation means 93, the recombination operation means 94 for performing crossover and mutation recombination operations, the conversion curve It is determined whether or not the optimization estimation has been completed, and has an estimation end determination means 95 that passes an optimal target signal conversion curve to the next target signal estimation means 74 as a constituent element when the optimization estimation has been completed.

  The feature of this embodiment (embodiment) is that the value obtained by relative comparison with the subject signal in the peripheral visual field obtained in embodiment (embodiment) 3 is the actual subject signal value (lightness value). When converting to, the edge part of the image obtained with the contrast adjustment amount of the brightness value using a genetic algorithm is further sharpened, and in Example (Embodiment) 6 Thus, it is considered that the versatility is higher than selecting a conversion method (conversion function) prepared in advance with average brightness.

  The flow of processing in the present embodiment (embodiment) is the same as that in embodiment (embodiment) 3 for the image input means 10, the image composition means 12, and the image output means 13. The contrast adjusting means 20 as the image processing means 11 is configured as shown in FIG. 5 and FIG. 8B, and the processing is the same as the processing of the embodiment (embodiment) 7 shown in FIG. Changed to perform contrast adjustment. The processing in the target signal conversion curve estimation means 82 by the genetic algorithm used at that time is also used to calculate an evaluation value indicating the superiority or inferiority of each target signal conversion curve candidate in the embodiment (embodiment) 7 shown in FIG. The signal is changed so as to use the edge strength of the target signal obtained from each conversion curve.

  The image input means 10, the image composition means 12, and the image output means 13 perform the same operations as in the embodiment (embodiment) 1. In the contrast adjustment process, first, the input image vi is converted into a signal in the La * b * space, the hue is maintained, and the lightness is normalized from 0.0 to 1.0. The target correction information deriving unit 51 calculates the maximum brightness value Lx and the minimum value Ln of the input image, and calculates the weighted average brightness AL (i, j) of the brightness L (i, j) in the peripheral visual field of the target pixel Pij. Then, the ratio between L (i, j) and AL (i, j) is calculated as the brightness contrast adjustment amount RL (i, j). The extraction unit 31 obtains an average value and a standard deviation amount of the lightness contrast adjustment amount RL (i, j), and extracts an effective region from the lightness contrast adjustment amount RL (i, j) determined by the value. The brightness contrast adjustment amount RL (i, j) is converted to a value in the range of 0.0 to 1.0 using the minimum value emin and the maximum value emax. Thereafter, the processing shifts to the target signal converting means 52 shown in FIG. First, the edge intensity of the brightness representing the contrast information in the input image 1 is calculated by the reference intensity calculating means 80, and the target signal conversion at the time of converting the brightness contrast adjustment amount obtained by the extracting means 31 into the actual brightness value A genetic algorithm performs curve estimation. The target signal conversion curve estimation means for performing this estimation is configured as shown in FIG. 9B, and sets conversion curve candidates, calculates target signals obtained from the respective conversion curve candidates, and calculates evaluation values of the respective conversion curve candidates. The optimal target signal conversion curve is estimated by sequentially calculating the fitness of each candidate, the candidate group recombination operation, and the estimation end determination. Then, using the obtained conversion curve, the brightness contrast adjustment amount obtained by the extraction means 31 is converted into an actual brightness value, and the input brightness value Lx and the minimum value Ln are suppressed to within the input image. The contrast-adjusted image v is generated by calculating the actual pixel value from the hue of each pixel and the adjusted brightness by inverse transformation. Processing is performed in this way.

  As described above, the image processing apparatus according to the present embodiment (embodiment) has the effect of improving the color stability of the pixel in a dark part such as a shadow in the embodiments (embodiments) 3 and 4, and the embodiment. (Embodiment) This has the effect of increasing the contrast in 7 and promoting the sharpening of the edge portion. In the present embodiment (embodiment), the contrast adjusting means has been described in the case of the configuration of FIG. 5 described in the embodiment (embodiment) 3. However, as shown in FIG. It is also possible to combine the target signal conversion means 52 in this example (embodiment) with the contrast adjustment means used in Example (embodiment) 4. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the eighth embodiment (embodiment). Can be realized.

  The basic configuration according to Example (Embodiment) 9 of the present invention is the same as that of Example (Embodiment) 1 as shown in FIG. Unlike the examples (embodiments) 1 to 8, the image processing means 11 is composed of a color adjustment means 21.

  FIG. 24 is a block diagram showing the color adjusting means 21 constituting the image processing means.

  As shown in FIG. 24, the color adjusting unit 21 divides an input image into a plurality of small regions, and representative colors representing color data in each small region obtained by the image dividing unit 240. Representative color extraction means 241 for obtaining the color, a color conversion table 243 that collects color data after color conversion prepared in advance, and a color closest to the representative color obtained by the representative color extraction means 241 is selected from the color conversion table 243 Representative color conversion color selection means 242 and the representative color obtained by the representative color conversion color selection means 242 and the distance between the conversion colors are finely adjusted, and the representative color is adjusted to the representative color. A color conversion unit 244 for converting to a converted color after fine adjustment is included as a component.

  FIG. 26 is a block diagram showing the representative color extracting means 241. As shown in FIG.

  As shown in FIG. 26, the representative color extraction unit 241 includes an initialization unit 260, a division axis determination unit 261, a cluster division unit 262, a cluster representative determination unit 263, a convergence determination unit 264, and a cluster division end determination unit. 265 and representative color output means 266 are included as constituent elements.

  The operation of the image processing apparatus configured as described above will be described with reference to FIGS. FIG. 25 is a flowchart showing color adjustment processing in the color adjustment means 21, and FIG. 27 is a flowchart showing processing in the representative color extraction means 241.

  The color image vi is digitally input to the image processing means 11 via the image input means 10, and each component value of the pixel value is normalized to a value from 0.0 to 1.0. Next, color adjustment processing is started by the color adjusting means 21 for the digital input image. As shown in FIG. 25, the input image is first divided into a plurality of small regions CGs (s = 0,..., NCnum-1) (other than the rectangular region) by the image dividing unit 240 (S91). . Here, NCnum represents the number of small areas.

  Thereafter, the representative color extracting means 241 obtains m representative colors representing each area for each small area (S101, FIG. 27). In this case, m is described as being fixed. However, the number of representative colors extracted from the target area based on the color distribution in the target area can be made variable. It is possible to further reduce shakiness at the color balance and edge boundary.

  The processing of the representative color extracting unit 241 is as shown in FIG. As a process in the representative color extraction unit 241, a clustering method is applied to the space of the pixel value VPij (r (i, j), g (i, j), b (i, j)) in the pixel Pij. In particular, this time, a vector belonging to a vector quantization (VQ) method that divides a vector space having a plurality of elements into a plurality of clusters based on the distribution state of the sample population is used to connect adjacent quantized representative vectors. The configuration is based on the k-mean method in which the cluster area frame is set as a perpendicular bisector of the line segment. It is also possible to use other clustering methods. First, the initialization unit 260 assumes that all color data VPij in all target regions belong to one cluster, and one representative color vector VC1 = (Cr1, Cg1, Cb1) has all color data in CGs. Is set (S102). Here, nnn represents the number of color data in the entire target color area, Cr1 represents the r component of the representative color vector VC1, Cg1 represents the g component, and Cb1 represents the b component. Thereafter, the pixel value VPij (r (i, j), g (i, j), b (i, j)) in the pixel Pij in the target area is a color vector using the serial number k in the area. It is assumed that VPk (rk, gk, bk) is used.

  Next, the split axis determining unit 261 represents the representative color vector VCj representing the cluster j in the current nn clusters and the nn−m color vectors uVPm = (urm, ugm, ubm) ( m = 0,..., nn−m−1) are calculated as absolute values (| urm−Crj |, | ugm−Cgj |, | ubm−Cbj |) of the differences of the elements constituting the sum ( tr = Σ | urm-Crj |, tg = Σ | ugm-Cgj |, tb = Σ | ubm-Cbj |) is calculated for each element. Then, it is considered that the current cluster can be divided in the direction of the axis having the largest value among tr, tg, and tb (S103). The cluster dividing unit 262 sets two temporary representative color vectors dVCj and dVCj + 1 in the axial direction obtained by the dividing axis determining unit 261 with the current representative color vector VCj as the center (S104). For example, when the r axis is selected by the split axis determining means 261, dVCj = (Crj−gamma × tr, Cgj, Cbj) and dVCj + 1 = (Crj + gamma × tr, Cgj, Cbj) using a certain small positive constant gamma. Is set by the cluster dividing means 262. Two temporary representative color vectors dVCj and dVCj + 1 are generated from one representative color vector VCj, and the processing in the split axis determining means 261 and the cluster splitting means 262 is performed for all nn clusters at the present time. The number of clusters generated is 2nn. Note that this method is not unique, and other methods are possible. Next, in the cluster representative determining means of the cluster representative determining means 263, the Euclidean distance disk [k, i] (i = i) between the 2nn provisional representative color vectors dVCi newly obtained by the cluster dividing means 262 and the input color vector VPk. 1,..., 2nn) is calculated, and processing is performed so that VPk belongs to i = i_min where disk [k, i] is minimum with respect to i (S105). Then, after each input color vector VPk can be assigned to a cluster, the center of gravity of the input color vector in cluster i is set as the representative color vector VCi of cluster i (S106). Based on the Euclidean distance between the representative color vector VCi of the cluster i obtained by the cluster representative determining unit 263 and the temporary representative color vector dVC obtained by the cluster dividing unit 262, the convergence determining unit 264 It is determined whether the representative color vector VCi has converged (S107). If not converged, the process moves to the cluster dividing means 262, and the temporary representative color vectors dVCj and dVCj + 1 are reset in the axial direction obtained in 261 around the current representative color vector VCj. is there. Specifically, dVCj = (Crj−2 × gamma × tr, Cgj, Cbj) and dVCj + 1 = (Crj + 2 × gamma × tr, Cgj, Cbj) are set by the cluster dividing unit 262 and obtained again by the cluster representative determining unit 263. The process returns to the process of distributing the input color vector VPk to 2nn clusters using the 2nn provisional representative color vectors dVCi. When it is determined that the convergence is determined by the convergence determination unit 264, the cluster division end determination unit 265 determines whether or not the number of clusters satisfies a predetermined cluster division number m (S108). The representative of the cluster i finally obtained by the representative color output means 266 is output as the representative color VCi (Cri, Cgi, Cbi) (S109, S92).

  As described above, the representative color extracting unit 241 and the process there perform the process of dividing the color data in the target color space into a plurality of clusters based on the distribution using VQ. However, this method is not unique, and instead of VQ, it is possible to simply divide the clusters so that the histograms in each cluster are the same from the maximum value and the minimum value in each color. A VQ method having a feature capable of clustering with high accuracy according to the statistical distribution is used. As other clustering, it is also possible to use a technique represented by a self-organizing neural network (for example, Kohonen's self-organizing network). Further, here, as the representative color vector representing each cluster, the split axis determining means 261 and the cluster representative determining means 263 set the barycentric vector of the color vector VPk belonging to each cluster, but the color vector belonging to each cluster itself is optimal. It is also possible to divide the cluster by selecting the correct one. Further, when the color representative VPk is allocated to each cluster by the cluster representative determining unit 263, VPk belongs to the cluster i having the smallest Euclidean distance disk [k, i] between the representative color vector VCi and the color vector VPk at the present time. However, in addition to the Euclidean distance, it is also possible to use the sum of absolute differences of the elements of the representative color vector VCi and the color vector VPk.

  The representative colors VCj [S] (Crj [S], Cgj [S], Cbj [S]) of the region CGs (s = 0,..., NCnum-1) obtained in this way (j = 0,. .., M−1) is compared with the color Vtu (rtu, gtu, btu) (u = 0,..., Tnum−1) in the color conversion table of the color conversion means 244 (S93). Here, Tnum indicates the number of color data in the color conversion table. Then, color data u = u_b [j, s] in the table having the smallest Euclidean distance len [s, u, j] with the representative color VCj [S] is found, and this is used as temporary conversion target color data. Further, the color Vtu_b (rtu_b, gtu_b, btu_b) is corrected based on the len [s, u_b, j] len [s, u # b, j]. As an example of the correction method, considering that the color does not change suddenly, the temporary target color is corrected as shown in (Equation 12) and the conversion target color VT [s, j] with respect to the representative color VCj [S] is obtained. (Rt [s, j], gt [s, j], bt [s, j]) (S94). Here, δ is a minute positive constant.

  The color conversion means 244 converts the representative color VCj [S] (j = 0,..., M−1) of the target region CGs (s = 0,..., NCnum−1) to the conversion target color VT [s. , J] (rt [s, j], gt [s, j], bt [s, j]), the processed image v is generated.

  The image synthesizing unit 12 generates a weighted average image of the processed image v and the input image vi obtained by the color adjusting unit 21, and this is an image vo output from the image output unit 13. Note that the processing in the image synthesizing unit 12 uses the same processing as that in the case of contrast adjustment in the first embodiment (embodiment), and is omitted here.

  By constructing the configuration as described above, the image processing apparatus according to the present embodiment (embodiment) has an advantage that it is not necessary to define both pre-conversion and post-conversion colors in advance. In addition, by extracting representative colors with high accuracy and generating a weighted average composite image with the original input image, anxiety about post-conversion color balance that occurs when using a color histogram simply for color adjustment. It becomes possible to suppress qualitativeity to some extent. These processes are also performed by software processing using a central processing unit (CPU) and a digital signal processor (DSP) used in a computer or the like according to the image processing method according to the present embodiment (embodiment). Can be realized.

  As described above, according to the image processing apparatus of the first aspect of the present invention, the image input means for converting the input analog image signal into digital image data, and desired image processing are performed on the converted digital image data. Image processing means for generating processed image data, image combining means for combining processed image data and digital image data to generate composite image data, and an image for converting the composite image data into an analog composite image signal Output means, and the image processing means is a contrast adjusting means for adjusting contrast with respect to digital image data or a color adjusting means for adjusting color with respect to digital image data. Since contrast adjustment or color adjustment can be performed on the input image, it has a captured light and dark part. Only it is possible to adjust the contrast of the captured image easily by using an image, also advantageous effect is obtained that it is possible to eliminate collapse of the color balance after converted dependence on the input image.

  According to the second image processing apparatus, in the image processing apparatus according to the first aspect, the contrast adjustment unit includes a correction information deriving unit that obtains the contrast adjustment amount of the target pixel, and the obtained contrast adjustment amount of the target pixel. An extraction unit that extracts a limited effective range, and a pixel value conversion unit that converts the pixel value of the target pixel into an adjustment pixel value in a limited range based on the contrast adjustment amount of the target pixel. Since the pixel contrast adjustment amount is obtained and the pixel value of the target pixel is converted into the adjustment pixel value within a limited range based on the contrast adjustment amount of the target pixel, the contrast with respect to the input image is not required without experience. An advantageous effect is obtained that adjustment or color adjustment can be performed automatically and reliably.

  According to the image processing apparatus described in the third aspect, in the image processing apparatus described in the first aspect, the contrast adjustment unit sets an initial condition and a pixel comparison range when calculating the contrast adjustment amount of the target pixel. Means for determining the contrast adjustment amount of the target pixel based on the pixel comparison range, and determining whether the contrast adjustment processing based on the calculated contrast adjustment amount of the target pixel has been completed for all the pixel comparison ranges An end determination unit that performs the process, a correction range changing unit that changes the pixel comparison range when the end determination unit does not determine the end, and a process that determines the end by the end determination unit. Is based on extraction means for extracting an effective range from contrast adjustment amounts obtained from a plurality of pixel comparison ranges and the contrast adjustment amount of the target pixel. A pixel value conversion unit that converts the pixel value of the target pixel into an adjustment pixel value in a limited range, so that a plurality of contrast adjustment amounts calculated by comparing the target pixel value with the average pixel value in the surrounding area Since it can be expanded to the weighted average value of the contrast adjustment amount in the peripheral area, there is an advantageous effect that the influence of the input image and the influence of the setting of a constant indicating the size of the peripheral visual field area can be reduced. It is done.

  According to the fourth image processing apparatus, in the first image processing apparatus, the contrast adjustment unit includes a signal conversion unit that converts a pixel value in the input image into a target signal that is a target at the time of contrast adjustment. The target correction information deriving means for obtaining the contrast adjustment amount of the target pixel with respect to the target signal obtained by the signal conversion means, and the effective range from the contrast adjustment amount of the target pixel obtained by the target correction information deriving means is limited. And extraction means for extracting and target signal conversion means for converting the target signal obtained by the signal conversion means into an adjustment target signal within a limited range based on the contrast adjustment amount of the target pixel and generating an adjusted target signal A signal reverse conversion means for performing a reverse conversion process on the pixel value in the input image to the adjusted pixel value based on the adjusted target signal and the target signal obtained by the signal conversion means. By converting the input image into information such as brightness and hue, and using this brightness as the target signal, the brightness adjustment amount of brightness is obtained by comparing the brightness of the target pixel with the average brightness in the surrounding area. Therefore, the pixel value of the image after adjustment is obtained from the extracted brightness and the hue obtained from the input image, and the obtained contrast adjusted image and the input are obtained. An advantageous effect is obtained that the contrast of the input image can be adjusted by performing weighted average composition with the image based on an appropriate coupling coefficient.

  According to the image processing apparatus described in the fifth aspect, in the image processing apparatus described in the first aspect, the contrast adjustment unit includes a signal conversion unit that converts a pixel value in the input image into a target signal that is a target at the time of contrast adjustment. Initial setting means for setting an initial condition and a pixel comparison range when calculating a contrast adjustment amount for the converted target signal, and an object for obtaining a contrast adjustment amount of each pixel with respect to the converted target signal based on the pixel comparison range A correction information deriving unit, an end determination unit that determines whether or not the contrast adjustment process based on the calculated contrast adjustment amount of each pixel has been completed in all pixel comparison ranges, and when the end determination unit does not determine the end Correction range changing means for changing the pixel comparison range and passing the processing to the target correction information deriving means; Extraction means for extracting the limited effective range from the contrast adjustment amount, and the target signal converted in the limited range based on the contrast adjustment amount of each pixel is converted into the adjustment target signal to obtain the adjusted target signal By having a target signal conversion means to generate, and a signal reverse conversion means for performing a reverse conversion process on the pixel value in the input image to the adjustment pixel value by the adjusted target signal and the target signal converted by the signal conversion means, Since the contrast adjustment amount calculated by comparing the target pixel value and the average pixel value in the surrounding area can be expanded to the weighted average value of the contrast adjustment amounts in a plurality of surrounding areas, high-precision contrast adjustment is performed. The advantageous effect that it can be obtained is obtained.

  According to the image processing apparatus described in item 6, in the image processing apparatus described in item 2 or 3, the pixel value conversion unit includes an average luminance calculation unit that calculates average luminance in the input image, and the calculated average luminance. Based on the conversion method obtained by the conversion method classifying means and the conversion method classification means for selecting the conversion method for converting the pixel value in the input image into the adjustment pixel value based on the obtained contrast adjustment amount. After extracting a region that seems to be effective from the contrast adjustment amount obtained by comparison with the surrounding pixels, by having a pixel value estimation means that converts the obtained contrast adjustment amount into a pixel value of an actual pixel Since a conversion method for converting the extracted contrast adjustment amount can be selected and applied based on the average luminance in the input image, the contrast adjustment can be easily and reliably performed. Advantageous effect of wear can be obtained.

  According to the image processing device described in item 7, in the image processing device described in item 4 or 5, the target signal conversion unit calculates an average value by calculating an average value in the input image of the target signal obtained by the signal conversion unit. Average target signal calculation means for generating the target signal, and a conversion method for converting the target signal obtained based on the contrast adjustment amount of the obtained target signal into the adjustment target signal based on the generated average target signal Conversion means classifying means for selecting, and target signal estimation means for converting the contrast adjustment amount of the obtained target signal into the target signal value of the actual pixel in accordance with the conversion method obtained by the conversion method classification means. After extracting a region that seems to be effective from the contrast adjustment amount of the target signal (for example, brightness) obtained by comparison with surrounding pixels, the average target signal value (average brightness) in the input image is also obtained. , Since a conversion method for converting the contrast adjustment amount of the extracted target signal can be selected and applied, advantageous effect can be performed easily and reliably contrast adjustment can be obtained.

  According to the image processing apparatus described in item 8, in the image processing apparatus described in item 2 or 3, the pixel value conversion unit includes a reference intensity calculation unit that calculates a reference intensity value indicating the contrast intensity in the input image; Based on the calculated reference intensity value, based on the obtained contrast adjustment amount, a conversion curve estimation means for estimating a conversion curve when converting a pixel value in the input image into an adjustment pixel value, and an estimated conversion curve By using the pixel value estimation means that converts the obtained contrast adjustment amount into the pixel value of the actual pixel, the region that seems to be effective is extracted from the contrast adjustment amount obtained by comparison with the surrounding pixels After that, the sum of the edge strengths obtained from the brightness of the input image can be calculated. Therefore, a plurality of conversion curves for converting the extracted contrast adjustment amount are prepared, and each conversion curve is prepared. It is advantageous that the sum of the edge strengths of the luminance of the final output candidate image obtained can be calculated, and the optimal conversion curve can be estimated by a genetic algorithm using the ratio of the value to the sum of the edge strengths of the input image as an evaluation value. An effect is obtained.

  According to the ninth aspect of the image processing device, in the image processing device according to the fourth or fifth aspect, the target signal conversion means indicates the contrast intensity in the input image with respect to the target signal obtained by the signal conversion means. A reference intensity calculating means for calculating a reference intensity value, and a conversion curve for converting the target signal obtained based on the contrast adjustment amount of the obtained target signal to the adjustment target signal based on the calculated reference intensity value Obtained by comparison with surrounding pixels by having a target signal conversion curve estimating means for estimating the target signal and a target signal estimating means for converting the obtained target signal into an adjustment target signal using the estimated conversion curve Target signal (brightness) After extracting a region that seems to be effective from the contrast adjustment amount, the sum of the edge intensity of the input image can be calculated from the target signal of the input image. Prepare multiple conversion curves for converting the contrast adjustment amount of the image, calculate the total edge strength of the target signal of the final output candidate image obtained from each conversion curve, and calculate the value for the total edge strength of the input image An advantageous effect is obtained that an optimal conversion curve can be estimated by a genetic algorithm using the ratio as an evaluation value.

  According to the tenth image processing apparatus, in the image processing apparatus according to the eighth aspect, the conversion curve estimation means sets an initial candidate group of adjustment vectors composed of a predetermined number of adjustment parameters. Initial candidate setting means for performing, pixel value conversion candidate calculation means for converting the pixel value in the input image into an adjustment pixel value based on the obtained contrast adjustment amount using each vector in the adjustment vector group at the present time, Obtained by the evaluation value calculation means for evaluating the contrast intensity by each conversion curve candidate using the reference intensity value, each vector in the adjustment vector group at the present time, and the converted pixel value obtained from the conversion vector candidates, and the previous evaluation value deriving means A new adjustment vector set by performing the recombination operation of the current candidate based on the fitness calculation means for calculating the fitness of each candidate and the calculated fitness of each candidate It seems to be more effective than the contrast adjustment amount obtained by comparison with the surrounding pixels by having the recombination operation means for generating and the estimation end determination means for determining whether or not the optimization vector optimization has been completed. Since the sum of the edge strengths obtained from the brightness of the input image can be calculated after extracting the region, multiple conversion curves for converting the extracted contrast adjustment amount are prepared and obtained from each conversion curve. An advantageous effect is obtained that the sum of the edge intensity of the final output candidate image is calculated, and the optimal conversion curve can be estimated by a genetic algorithm using the ratio of the value to the sum of the edge intensity of the input image as an evaluation value. It is done.

  According to the eleventh image processing apparatus, in the image processing apparatus according to the ninth aspect, the target signal conversion curve estimation means includes an initial candidate group of adjustment vectors configured by a predetermined number of adjustment parameters. Target signal conversion for converting the obtained target signal into the adjustment target signal based on the contrast adjustment amount of the obtained target signal using the initial candidate setting means for setting and each vector in the adjustment vector group at the present time Candidate calculation means, evaluation value calculation means for evaluating the contrast intensity of the target signal by each conversion curve candidate using the reference intensity value and the target signal value after conversion obtained from each vector in the adjustment vector group at the present time And a fitness calculation means for calculating the fitness of each candidate obtained by the evaluation value deriving means, and a recombination operation of the current candidate based on the calculated fitness of each candidate It was obtained by comparison with surrounding pixels by having recombination operation means for generating a new adjustment vector set by performing and estimation end determination means for determining whether or not adjustment vector optimization was completed After extracting a region that seems to be effective from the target signal (brightness) contrast adjustment amount, the sum of the edge intensity of the input image can be calculated from the target signal of the input image. Prepare multiple conversion curves at the time of conversion, calculate the sum of edge strengths of the target signal of the final output candidate image obtained from each conversion curve, and use the ratio of that value to the sum of edge strengths of the input image as the evaluation value An advantageous effect is obtained that an optimal conversion curve can be estimated by a genetic algorithm.

  According to the image processing apparatus of the twelfth aspect, in the image processing apparatus of the first aspect, the color adjustment unit includes an image dividing unit that divides the input image into a plurality of small regions, and each of the image obtained by the image dividing unit. Representative color extraction means for obtaining a representative color representing the color data in the small area, a color conversion table for collecting color data after color conversion prepared in advance, and a color closest to the representative color obtained by the representative color extraction means Based on the distance between the representative color obtained by the representative color conversion color selecting means and the representative color obtained by the representative color conversion color selecting means, the selected conversion color is selected as a conversion color from the color conversion table. Fine adjustment and color conversion means to convert the representative color to the converted color after fine adjustment, the input image is divided into multiple detail areas, and the representative color is selected by the statistical distribution of colors in each area Color conversion table prepared in advance Compare the representative color with the representative color, extract the converted color closest to each representative color, fine-tune the converted color data selected based on the distance between the two colors, The color of the input image can be converted into the finely converted conversion color by weighted average synthesis of the color-adjusted image thus obtained and the input image based on an appropriate coupling coefficient. An advantageous effect that adjustment can be performed is obtained.

  According to the image processing apparatus of the thirteenth aspect, in the image processing apparatus of the twelfth aspect, the representative color extracting means is a start state when performing sequential division processing on the color data in the divided small area Focus on initializing means to classify all color data into each group, obtain the representative color of each group, and distribution of color data belonging to the group to be divided, when dividing the target group A division axis determining means for determining the component, a cluster dividing means for dividing the target group into a plurality of groups according to the obtained target component, and distributing the color data belonging to the target group to the group obtained after the division, and each obtained The cluster representative color determining means for obtaining the representative color of the color data belonging to the group and whether or not the representative color has converged. When it is determined that the convergence determination unit and the convergence determination unit have converged, the representative color from the target region obtained so far is determined. If it is determined that the predetermined number has been obtained, if it is determined that the process has been completed by the cluster division end determination unit and the cluster division end determination unit in which the process to the division axis determination unit moves if not, By having representative color output means for outputting the predetermined number of representative colors obtained, the input image can be divided into a plurality of detailed areas, and the representative colors can be selected based on the statistical distribution of colors in each area. Therefore, the color in the color conversion table prepared in advance is compared with the representative color, and the converted color closest to each representative color is extracted, and after the conversion selected based on the distance between the two colors Fine-tune the color data of The color of the input image can be converted into the finely converted conversion color by weighted average synthesis of the color-adjusted image thus obtained and the input image based on an appropriate coupling coefficient. An advantageous effect that adjustment can be performed is obtained.

  According to the fourteenth image processing apparatus, in the image processing apparatus according to any one of the first to thirteenth aspects, the image synthesizing means is either an input image or an adjusted image obtained by the image processing means. Selection criterion value determining means for determining whether to give priority to the image, and a coupling coefficient derivation for determining a coupling coefficient for the input image and the adjusted image obtained by the image processing means based on the determination result of the selection criterion value determination means And a weighted average combining means for generating a weighted average image of the input image and the adjusted image obtained by the image processing means using the coupling coefficient of each image determined by the coupling coefficient deriving means. The contrast adjustment amount can be obtained by comparing the pixel value of the target pixel with the average pixel value in the surrounding area, and the contrast adjustment amount that is considered to be effective can be extracted. Advantageous effect that it is possible to perform contrast adjustment of the input image by weighted-averaging-synthesis based contrast adjustment image and the input image and the appropriate coupling coefficient is obtained.

  According to the image processing method of the fifteenth aspect, an image input step for converting the input analog image signal into digital image data, and an image for generating processed image data by performing desired image processing on the converted digital image data A processing step, an image synthesizing step for synthesizing the processed image data and the digital image data to generate synthesized image data, and an image output step for converting the synthesized image data into an analog synthesized image signal. The processing step is a contrast adjustment step for performing contrast adjustment for digital image data or a color adjustment step for performing color adjustment for digital image data. Since color adjustment can be performed, One image only can adjust the contrast of the captured image easily with, also is advantageous effect that it is possible to eliminate collapse of the color balance of the converted and dependence on an input image obtained.

  According to the sixteenth image processing method, in the image processing method according to the fifteenth aspect, the contrast adjustment step includes a correction information deriving step for obtaining the contrast adjustment amount of the target pixel, and the obtained contrast adjustment amount of the target pixel. An extraction step for extracting a limited effective range, and a pixel value conversion step for converting the pixel value of the target pixel into an adjustment pixel value in a limited range based on the contrast adjustment amount of the target pixel. Since the pixel contrast adjustment amount is obtained and the pixel value of the target pixel is converted into the adjustment pixel value within a limited range based on the contrast adjustment amount of the target pixel, the contrast with respect to the input image is not required without experience. An advantageous effect is obtained that the adjustment can be performed automatically and reliably.

  According to the seventeenth image processing method, in the image processing method according to the fifteenth aspect, the contrast adjustment step is an initial setting for setting an initial condition and a pixel comparison range when calculating the contrast adjustment amount of the target pixel. Step, a correction information derivation step for obtaining the contrast adjustment amount of the target pixel based on the pixel comparison range, and whether or not the contrast adjustment processing based on the obtained contrast adjustment amount of the target pixel has been completed for all the pixel comparison ranges An end determination step in which the pixel comparison range is changed and the process is passed to the correction information deriving step if the end determination step is not determined in the end determination step, and the end determination step is performed in the end determination step. Is an extraction step that extracts an effective range from the contrast adjustment amount obtained from a plurality of pixel comparison ranges. And a pixel value conversion step for converting the pixel value of the target pixel into an adjustment pixel value in a limited range based on the contrast adjustment amount of the target pixel, thereby obtaining the target pixel value and the average pixel value in the surrounding area. The contrast adjustment amount calculated by comparison can be expanded to the weighted average value of contrast adjustment amounts in multiple surrounding areas, reducing the influence of the input image and the influence of setting a constant indicating the size of the peripheral visual field area. The advantageous effect that it can be made is obtained.

  According to the image processing method of the eighteenth aspect, in the image processing method of the fifteenth aspect, the contrast adjustment step includes a signal conversion step of converting a pixel value in the input image into a target signal that is a target at the time of contrast adjustment; The target correction information derivation step for obtaining the contrast adjustment amount of the target pixel with respect to the target signal obtained in the signal conversion step, and the effective range from the contrast adjustment amount of the target pixel obtained in the target correction information derivation step is limited. An extraction step for extracting, and a target signal conversion step for generating an adjusted target signal by converting the target signal obtained in the signal conversion step into an adjustment target signal in a limited range based on the contrast adjustment amount of the target pixel; The pixel value in the input image is adjusted using the adjusted target signal and the target signal obtained in the signal conversion step. A signal inverse conversion step for performing an inverse conversion process to a value, thereby converting the input image into information such as brightness and hue, and using the brightness as a target signal, the brightness of the target pixel and the average brightness in the surrounding area Since the contrast adjustment amount of lightness can be obtained from the comparison, and the contrast adjustment amount of lightness that seems to be effective among them can be extracted, the image of the image after adjustment is determined based on the extracted lightness and the hue obtained from the input image. An advantageous effect is obtained that the contrast of the input image can be adjusted by obtaining the pixel value and performing the weighted average synthesis of the obtained contrast adjusted image and the input image based on an appropriate coupling coefficient.

  According to the nineteenth image processing method, in the image processing method according to the fifteenth aspect, the contrast adjustment step includes: a signal conversion step of converting a pixel value in the input image into a target signal that is a target at the time of contrast adjustment; An initial setting step for setting an initial condition and a pixel comparison range when calculating a contrast adjustment amount for the converted target signal, and a target for obtaining a contrast adjustment amount of each pixel for the converted target signal based on the pixel comparison range When the correction information deriving step, the end determination step for determining whether or not the contrast adjustment processing based on the calculated contrast adjustment amount of each pixel is completed in all pixel comparison ranges, and the end determination step are not completed. Ends the correction range change step that changes the pixel comparison range and passes the processing to the target correction information derivation step. When it is determined that the process is completed in the fixed step, an extraction step for extracting the limited effective range from the contrast adjustment amount of each pixel and an adjustment of the target signal converted in the limited range based on the contrast adjustment amount of each pixel A target signal conversion step that converts the target signal to generate an adjusted target signal, and a pixel value in the input image is inversely converted into an adjusted pixel value by the adjusted target signal and the target signal converted in the signal conversion step. And a signal inverse conversion step for performing the contrast expansion of the contrast adjustment amount calculated by comparing the target pixel value and the average pixel value in the surrounding area to the weighted average value of the contrast adjustment amounts in the plurality of surrounding areas. Therefore, an advantageous effect that high-precision contrast adjustment can be performed is obtained.

  According to the twentieth image processing method, in the image processing method according to the sixteenth or seventeenth aspect, the pixel value conversion step includes an average luminance calculation step for calculating an average luminance in the input image, and the calculated average luminance. Based on the conversion method obtained in the conversion method classification step and the conversion method classification step for selecting the conversion method for converting the pixel value in the input image into the adjustment pixel value based on the obtained contrast adjustment amount. A pixel value estimation step for converting the obtained contrast adjustment amount into a pixel value of an actual pixel, and after extracting a region that seems to be effective from the contrast adjustment amount obtained by comparison with surrounding pixels Since a conversion method for converting the extracted contrast adjustment amount can be selected and applied based on the average luminance in the input image, it can be easily and reliably copied. Advantageous effect that it is possible to perform trust adjustment is obtained.

  According to the image processing method of the twenty-first aspect, in the image processing method of the eighteenth or nineteenth aspect, the target signal conversion step calculates an average value by calculating an average value in the input image of the target signal obtained in the signal conversion step. Average target signal calculation step for generating the target signal, and a conversion method for converting the target signal obtained based on the contrast adjustment amount of the obtained target signal into the adjustment target signal based on the generated average target signal And a target signal estimation step for converting the contrast adjustment amount of the obtained target signal into a target signal value of an actual pixel according to the conversion method obtained in the conversion method classification step. After extracting a region that seems to be effective from the contrast adjustment amount of the target signal (for example, brightness) obtained by comparison with surrounding pixels, Since it is possible to select and apply a conversion method for converting the contrast adjustment amount of the extracted target signal based on the average target signal value (average brightness), it is possible to easily and reliably perform the contrast adjustment. An advantageous effect is obtained.

  According to the image processing method described in Item 22, in the image processing method described in Item 16 or 17, the pixel value conversion step includes a reference intensity calculation step of calculating a reference intensity value indicating the contrast intensity in the input image; Based on the calculated reference intensity value, a conversion curve estimation step for estimating a conversion curve for converting a pixel value in the input image into an adjustment pixel value based on the obtained contrast adjustment amount, and an estimated conversion curve And a pixel value estimation step that converts the obtained contrast adjustment amount into the pixel value of the actual pixel, thereby extracting a region that seems to be effective from the contrast adjustment amount obtained by comparison with surrounding pixels After that, the sum of the edge strengths obtained from the brightness of the input image can be calculated, so there are multiple conversion curves when converting the extracted contrast adjustment amount. Complementary calculation, the sum of the edge intensity of the final output candidate image obtained from each conversion curve is calculated, and the ratio of the value to the sum of the edge intensity of the input image is used as an evaluation value to determine the optimal conversion curve using a genetic algorithm. The advantageous effect that it can be estimated is obtained.

  According to the 23rd image processing method, in the 18th or 19th image processing method, the target signal conversion step indicates the contrast intensity in the input image with respect to the target signal obtained in the signal conversion step. A reference intensity calculation step for calculating a reference intensity value, and a conversion curve for converting the target signal obtained based on the contrast adjustment amount of the obtained target signal to the adjustment target signal based on the calculated reference intensity value Obtained by comparison with surrounding pixels by having a target signal conversion curve estimation step for estimating the target signal and a target signal estimation step for converting the obtained target signal into an adjustment target signal using the estimated conversion curve After extracting a region that seems to be effective from the target signal (brightness) contrast adjustment amount, the sum of the edge strength of the input image can be calculated from the target signal of the input image. Therefore, a plurality of conversion curves for converting the contrast adjustment amount of the extracted target signal are prepared, and the sum of edge strengths by the target signal of the final output candidate image obtained from each conversion curve is calculated, and the value An advantageous effect is obtained that an optimal conversion curve can be estimated by a genetic algorithm using the ratio of the input image to the sum of edge strengths as an evaluation value.

  According to the image processing method described in Item 24, in the image processing method described in Item 22, the conversion curve estimation step sets an initial candidate group of adjustment vectors composed of a predetermined number of adjustment parameters. An initial candidate setting step, and a pixel value conversion candidate calculation step for converting a pixel value in the input image into an adjustment pixel value based on the obtained contrast adjustment amount using each vector in the adjustment vector group at the present time; It is obtained in the evaluation value calculation step that evaluates the contrast strength by each conversion curve candidate using the reference intensity value, each vector in the adjustment vector group at the present time, and the converted pixel value obtained from the conversion curve candidate, and the previous evaluation value derivation step. The fitness calculation step for calculating the fitness of each candidate, and the current candidate recombination operation based on the calculated fitness of each candidate. Contrast adjustment obtained by comparison with surrounding pixels by having a recombination operation step for generating a new adjustment vector set in step 1 and an estimation end determination step for determining whether or not adjustment vector optimization has ended Since it is possible to calculate the sum of the edge strengths obtained from the brightness of the input image after extracting a region that seems to be more effective than the amount, prepare multiple conversion curves when converting the extracted contrast adjustment amount, The sum of the edge intensity of the final output candidate image obtained from each conversion curve is calculated, and the optimal conversion curve can be estimated by a genetic algorithm using the ratio of the value to the total edge intensity of the input image as an evaluation value. The advantageous effect is obtained.

  According to the image processing method described in Item 25, in the image processing method described in Item 23, the target signal conversion curve estimation step includes an initial candidate group of adjustment vectors composed of a preset number of adjustment parameters. Target signal conversion for converting the obtained target signal to the adjustment target signal based on the contrast adjustment amount of the obtained target signal using each vector in the adjustment vector group at the current time Candidate calculation step and evaluation value calculation step for evaluating the contrast intensity of the target signal by each conversion curve candidate using the reference intensity value and the target signal value after conversion obtained from each vector in the adjustment vector group at the present time And a fitness calculation step for calculating the fitness of each candidate obtained in the evaluation value deriving step, and the fitness of each candidate calculated. A recombination operation step of generating a new adjustment vector set by performing a recombination operation of the current candidate, and an estimation end determination step of determining whether or not the adjustment vector optimization has been completed. After extracting a region that seems to be effective from the target signal (brightness) contrast adjustment amount obtained by comparison with the pixel, the sum of the edge intensity of the input image can be calculated from the target signal of the input image. Prepare multiple candidate conversion curves for converting the contrast adjustment amount of the target signal, calculate the sum of the edge strengths of the target signal of the final output candidate image obtained from each conversion curve, and the edge of the input image of that value An advantageous effect is obtained that an optimal conversion curve can be estimated by a genetic algorithm using the ratio to the sum of the strengths as an evaluation value.

  According to the twenty-sixth image processing method, in the fifteenth image processing method, the color adjustment step includes an image division step for dividing the input image into a plurality of small regions, and each of the image division steps obtained in the image division step. A representative color extraction step for obtaining a representative color representing the color data in the small area, a color conversion table for collecting color data after color conversion prepared in advance, and a color closest to the representative color obtained in the representative color extraction step Each conversion color selected based on the distance between the representative color and the conversion color obtained in the representative color conversion color selection step and the representative color conversion color selection step. Fine adjustment and color conversion step to convert the representative color to the converted color after fine adjustment, the input image is divided into multiple detail areas, and the representative color is selected by the statistical distribution of colors in each area can do Then, after comparing the color in the color conversion table prepared in advance with the representative color and extracting the converted color closest to each representative color, and after the conversion selected based on the distance between the two colors The color data can be fine-tuned and the representative colors in each area can be converted to the fine-tuned conversion color. The color-adjusted image thus obtained and the input image are used based on an appropriate coupling coefficient. An advantageous effect that the color adjustment of the input image can be performed by performing the weighted average composition is obtained.

  According to the image processing method described in Item 27, in the image processing method described in Item 26, the representative color extraction step is a start state when performing sequential division processing on the color data in the divided small region. Focus on the division of the target group based on the initialization step of classifying all the color data into each group, obtaining the representative color of each group, and the distribution of the color data belonging to the division target group A division axis determination step for determining components, a cluster division step for dividing the target group into a plurality of groups according to the obtained component of interest, and distributing the color data belonging to the target group to the groups obtained after the division, and each obtained The cluster representative color determination step for obtaining the representative color of the color data belonging to the group and whether the representative color has converged. In the convergence determination step in which the process moves to the cluster division step in order to perform group division again based on the current representative color, and when it is determined that the convergence is determined in the convergence determination step, the values obtained so far are obtained. It is determined whether or not a predetermined number of representative colors from the target area have been obtained. If not, the process is terminated at the cluster division end determination step and the cluster division end determination step. The input image is divided into a plurality of detail areas, and a statistical distribution of colors in each area is provided. Since the representative color can be selected, the color in the color conversion table prepared in advance is compared with the representative color, and the converted color closest to each representative color is extracted, and the distance between the two colors is set. The color data after conversion selected in (1) and (2) can be finely adjusted, and the representative color in each area can be converted into this finely adjusted conversion color. An advantageous effect is obtained that color adjustment of an input image can be performed by performing weighted average synthesis based on a simple coupling coefficient.

  According to the twenty-eighth image processing method, in the image processing method according to any one of the fifteenth to twenty-seventh aspects, the image synthesizing step includes either the input image or the adjusted image obtained in the image processing step. Selection criterion value determining step for determining whether to give priority to the image, and a coupling coefficient derivation for determining a coupling coefficient for the input image and the adjusted image obtained in the image processing step based on the determination result of the selection criterion value determination step And a weighted average synthesis step for generating a weighted average image of the input image and the adjusted image obtained in the image processing step using the coupling coefficient of each image determined in the coupling coefficient derivation step. Find the contrast adjustment amount by comparing the pixel value of the target pixel with the average pixel value in the surrounding area, and extract the contrast adjustment amount that seems to be effective Therefore, it is possible to adjust the contrast of the input image by performing the weighted average synthesis of the contrast adjustment image and the input image obtained as described above based on an appropriate coupling coefficient. It is done.

The block diagram which shows the basic composition of the image processing apparatus by the Example (embodiment) 1, 2, 3, 4, 5, 6, 7, 8, 9 of this invention (A) Block diagram showing the image processing means of FIG. 1 (b) Block diagram showing the image processing means of FIG. FIG. 2A is a block diagram showing the contrast adjusting means of FIG. Block diagram showing contrast adjustment means Block diagram showing contrast adjustment means Block diagram showing contrast adjustment means (A) Block diagram showing pixel value conversion means (b) Block diagram showing object signal conversion means (A) Block diagram showing pixel value conversion means (b) Block diagram showing object signal conversion means (A) Block diagram showing transformation curve estimation means prepared for genetic algorithm estimation (b) Block diagram showing target signal transformation curve estimation means prepared for genetic algorithm estimation The block diagram which shows the image composition means of FIG. Explanatory diagram schematically showing human vision Flow chart showing operation in contrast adjusting means 7 is a flowchart showing the operation of the image synthesizing means of the image processing apparatus according to Embodiment 1 of the present invention. Explanatory drawing schematically showing the peripheral visual field area Flow chart showing operation in contrast adjusting means Flow chart showing operation in contrast adjusting means Flow chart showing operation in contrast adjusting means Flow chart showing operation of pixel value changing means A flowchart showing the flow of processing in the contrast adjusting means Flowchart showing the flow of processing of the conversion curve estimation means in contrast adjustment processing Schematic diagram schematically showing the chromosome structure used in the genetic algorithm in the conversion curve estimation means Schematic diagram schematically showing the outline of the roulette selection method in the conversion curve estimation means Schematic diagram schematically showing the recombination operation processing of the genetic algorithm used in the conversion curve estimation means Block diagram showing color adjustment means Flow chart showing color adjustment processing in color adjustment means Block diagram showing representative color extraction means Flow chart showing processing in representative color extraction means Block diagram showing an image processing apparatus in Document 1 The block diagram which shows the imaging device in literature 2 Explanatory drawing of the digital image improvement method in Reference 3

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image input means 11 Image processing means 12 Image composition means 13 Image output means 20 Contrast adjustment means 21 Color adjustment means 30 Correction information derivation means 31 Extraction means 32 Pixel value conversion means 40 Initial setting means 41 End determination means 42 Correction range change means DESCRIPTION OF SYMBOLS 50 Signal conversion means 51 Target correction information derivation means 52 Target signal conversion means 53 Signal reverse conversion means 70 Average brightness calculation means 71 Conversion method classification means 72 Pixel value estimation means 73 Average target signal calculation means 74 Target signal estimation means 80 Reference intensity calculation Means 81 Conversion curve estimation means 82 Target signal conversion curve estimation means 90 Initial candidate setting means 91 Pixel value conversion candidate calculation means 92 Evaluation value derivation means 93 Fitness calculation means 94 Recombination operation means 95 Estimation end determination means 100 Selection reference value determination means 101 Coupling coefficient deriving means 10 2 Weighted average combining means 240 Image dividing means 241 Representative color extracting means 242 Representative color conversion color selecting means 243 Color conversion table 244 Color converting means 260 Initializing means 261 Division axis determining means 262 Cluster dividing means 263 Cluster representative determining means 264 Convergence determination Means 265 Cluster division end judgment means 266 Representative color output means

Claims (13)

In an image processing apparatus that performs contrast adjustment on image data,
Means for obtaining a pixel value of a processing target pixel and pixel values of a plurality of pixels included in a peripheral region of the processing target pixel;
Contrast adjusting means for changing and outputting the pixel value of the processing target pixel based on a comparison between a pixel value of the processing target pixel and a pixel value of a plurality of pixels included in the peripheral region or a converted value thereof;
Means for generating image data from the output individual pixel values,
The contrast adjusting unit outputs the pixel value of the processing target pixel without performing the contrast adjustment on the processing target pixel when a plurality of pixels included in the peripheral region uniformly show the same pixel value. thing,
An image processing apparatus. The case where the same pixel value is uniformly shown is a case where the pixels included in the peripheral region are uniformly white or black.
The image processing apparatus according to claim 1. In the case of showing the white color, the pixel value in the peripheral region has a pixel value higher than a predetermined threshold value,
The image processing apparatus according to claim 2, wherein the black color indicates a case where a pixel value in the peripheral region has a pixel value lower than a predetermined threshold value. In the case of indicating the white color, the pixel value in the peripheral region has a pixel value higher than a threshold value that is decreased by a predetermined value from the maximum value that can be taken,
3. The image processing device according to claim 2, wherein the black color indicates a case where a pixel value in the peripheral region has a pixel value lower than a threshold value that is increased by a predetermined value from a minimum value that can be taken. . In an image processing apparatus that performs contrast adjustment of image data,
Obtaining a pixel to be processed and pixel values of a plurality of pixels included in a peripheral region located around the pixel to be processed;
Determining a contrast adjustment amount of the processing target pixel based on a comparison between the acquired processing target pixel and a pixel value of the plurality of pixels included in the peripheral region or an information amount obtained by converting the pixel value;
Adjust the pixel value to be processed based on the contrast adjustment amount,
Contrast adjustment means for keeping the contrast adjustment amount within a predetermined range when the pixels included in the peripheral area have the same uniform color,
An image processing apparatus comprising: image output means for outputting the obtained adjusted image data in a predetermined format. In the contrast adjusting means, when the pixels included in the peripheral area have the same uniform color, the color information obtained from the pixel values in the peripheral area is centered on a preset reference color. Represents a case that falls within the prescribed range.
The image processing apparatus according to claim 5, wherein the reference color is a value of color information prepared according to a preset processing target. In the contrast adjusting means, when the pixels included in the peripheral area have the same uniform color, the color information obtained from the pixel values in the peripheral area is centered on a preset reference color. Represents a case that falls within the prescribed range.
The image processing apparatus according to claim 5, wherein the reference color is a value determined from a distribution of color information obtained from pixel values in a preset processing region. The contrast adjustment unit is configured to adjust a contrast adjustment amount of the target pixel based on a difference / ratio between a pixel value of the target pixel and an information amount representing pixel characteristics in the peripheral region, or a value obtained by predetermined conversion to the value. Correction information deriving means for obtaining
Extraction means for extracting a limited effective range from the contrast adjustment amount distribution of the target pixel included in the predetermined region;
Pixel value conversion means for converting the contrast adjustment amount of each target pixel into an adjusted pixel value so that the contrast adjustment amount included in the extracted range falls within a predetermined pixel value range; and the pixel value conversion means 5. The image processing apparatus according to claim 1, wherein the image processing apparatus is composed of composite image data of the image data obtained in step 1 and the input image data. The pixel value conversion means includes a reference intensity calculation means for calculating a reference intensity value obtained from a contrast adjustment amount distribution, a contrast intensity distribution, or a pixel value distribution within a predetermined region including the target pixel,
Conversion curve estimation means for estimating a curve shape of a pixel value conversion function for converting a contrast adjustment amount obtained for each target pixel so as to fall within a predetermined pixel value range based on the calculated reference intensity value When,
The image processing apparatus according to claim 7, further comprising: a pixel value estimating unit that converts the contrast adjustment amount of each target pixel into an actual pixel value using the estimated conversion curve. The contrast adjusting means is a signal converting means for converting a pixel value in the input image into a target signal to be subjected to contrast adjustment;
Based on the difference / ratio between the signal for contrast adjustment of the target pixel and the amount of information representing the pixel feature in the predetermined area perceived by the target signal, or the value obtained by predetermined conversion to the value, Correction information deriving means for obtaining a contrast adjustment amount;
Extraction means for extracting a limited effective range from the contrast adjustment amount distribution of the target pixel included in the predetermined region;
Target signal conversion means for converting the contrast adjustment amount of each target pixel into an adjusted target signal value so that the contrast adjustment amount included in the extracted range falls within a predetermined target signal range;
Image synthesis means for generating composite image data of the target signal data obtained by the target signal conversion means and the input target signal data;
5. The apparatus according to claim 1, further comprising: a signal reverse conversion unit that performs reverse conversion to an adjusted pixel value using the target signal that has undergone image synthesis and the target signal obtained by the signal conversion unit. An image processing apparatus according to any one of the above. The target signal conversion means includes a reference intensity calculation means for calculating a reference intensity value obtained from a contrast adjustment amount distribution, a contrast intensity distribution, or a target signal distribution in a predetermined area including the target pixel, and
A target signal for estimating a curve shape of a target signal value conversion function for converting the contrast adjustment amount obtained for each target pixel so as to fall within a predetermined target signal value range based on the calculated reference intensity value A conversion curve estimation means;
The image processing apparatus according to claim 10, further comprising target signal value estimation means for converting a contrast adjustment amount of each target pixel into an adjusted target signal value using the estimated conversion curve. The image composition means includes a selection reference value determination means for determining which of the pixel value of the input image and the pixel value of the adjusted image obtained by the contrast adjustment means is prioritized;
A coupling coefficient deriving unit that determines a coupling coefficient according to the input pixel value and a coupling coefficient according to the pixel value of the adjusted image based on the result of the selection reference value determination unit;
Using the two coupling coefficients determined by the coupling coefficient deriving unit, a weighted average value of the pixel value of the input image and the pixel value of the adjusted image obtained by the image processing unit is obtained, and the target pixel of the composite image is calculated. The image processing apparatus according to claim 8, wherein the image processing apparatus includes weighted average combining means that regards a pixel value. The coupling coefficient derivation means for determining the coupling coefficient concerning the input pixel value and the coupling coefficient concerning the pixel value of the adjusted image has a pixel value of the input pixel and a difference / ratio between the input pixel value and the pixel value of the adjusted image. The image processing apparatus according to claim 11, wherein a coupling coefficient relating to a pixel value of the input image and a pixel value of the image after adjustment is determined based on a value obtained by predetermined conversion.

Images