JP5267807B2 - Image processing apparatus, image processing method, program, and recording medium - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, a program, and a recording medium for converting a color that cannot be reproduced when the color reproduction range of an image output apparatus is narrower than the color reproduction range of an output image.

  The color information is basically expressed from three components of hue (H *), lightness (L *), and saturation (C *). FIG. 13 shows a color plane of lightness (L *) and saturation (C *) in a certain hue (H *). In FIG. 13, a portion indicated by a triangle is an output color reproduction range (color gamut), and an outer portion indicates a color range that cannot be reproduced (represented) by the output system. In the output image data (color information), there may be a color inside the above-described triangle, or a color outside the triangle (outside the color reproduction range of the output system) as shown by an ellipse. Colors outside this range are converted to colors in the output system reproduction range, but are usually converted to triangle side colors.

  For example, there is a method of compressing and mapping a color outside the color reproduction range of the output system within the color reproduction range of the output system with a constant hue around the white point on the chromaticity diagram (see Patent Document 1).

  However, since the above-described conventional conversion processing method performs a single color conversion process on one output image, “color change”, “gradation loss”, etc. in a part of the output image (plane) There is a problem that occurs. Specifically, even though the color is originally the same color, an image that does not feel uncomfortable in one part of the output image may have noticeable color change or gradation collapse in another part. The part appears as a phenomenon in which pseudo gradation occurs.

  Here, the color change is a phenomenon in which the color of the output image data has changed to such an extent that the color of the output image is clearly different from the color of the output image. This is a phenomenon in which an image with a gray level is reduced in the output image, and the level of the gray level is reduced in the output image.Further, the pseudo gradation is a smooth gray level image in the output image data, This is a phenomenon in which the shading is lost.

  A conventional processing method performed for colors outside the color reproduction range of the output system in the output image data performs a single color conversion process for one output image. For this reason, even though the color is the same, even if appropriate conversion is performed in one part of the image, the gradation is lost in another part, or pseudo gradation occurs in another part. There is a problem.

  This is because the colors outside the color reproduction range to be converted are uniformly color-converted ignoring the size and distribution in the entire output image. . That is, for example, whether the color of the pixel of interest is one color in a continuous tone, a solid color, or a color in a region where the change in brightness direction is large, or the change in saturation direction is The optimum color conversion method varies depending on peripheral information such as whether the color is in a large area.

  As a method for performing the optimum color conversion according to the color distribution information in the output image as described above, in Patent Document 2, pixels that are out of the color reproduction range and that have the same hue are present. Color information conversion processing of different output systems is performed for pixels that are present and those that are not.

  However, if the mapping direction is determined based only on the color distribution without considering the influence of noise contained in the pixels that are present in a solid state, mapping may be performed in a direction that increases noise. Also in Patent Document 3, color conversion processing is performed by changing the direction of converting the color gamut of the image data in accordance with the color distribution of the image data. Therefore, noise may be increased.

The present invention has been made in view of the above problems,
An object of the present invention is to cluster a particularly dark pixel outside the color reproduction range based on the color difference, and then analyze the noise contained in the pixel in the cluster to determine the mapping direction for reducing the noise in the cluster. Another object of the present invention is to provide an image processing apparatus, an image processing method, a program, and a recording medium that perform color conversion processing.

The present invention provides an image processing apparatus for converting color information that cannot be reproduced when part of the color information of the output image information is wider than the color reproduction range of the image output system. One or more pixels that are outside the color reproduction range and are in the dark area, and a color reproduction range determination means that determines whether the image is a dark area in the output image; Cluster classification means for classifying into clusters, noise determination means for determining whether or not each pixel of the output image is noise, and features for the pixels determined by the noise determination means as noise in the classified clusters extracting, and determining the noise analyzing means whether noise is present in the clusters, the color conversion of the cluster it is determined that noise is present by the noise analysis unit To select a first color conversion method for mapping a particular point direction, as the color conversion of the cluster it is determined that there is no noise, the first of the second color conversion method that is different from the color conversion method The main feature is to have a color conversion method determining means to be selected and a color conversion processing means for performing color conversion processing by the color conversion method determined by the color conversion method determining means.

  According to the present invention, particularly dark pixels outside the color reproduction range are clustered based on the color difference, and then the noise included in the pixels in the cluster is analyzed to determine the mapping direction for reducing the noise in the cluster. Since the optimum color conversion process is performed, noise can be made inconspicuous.

The structure of Example 1 of this invention is shown. The processing flowchart of Example 1 is shown. The structure of a clustering part is shown. The processing flowchart of a clustering part is shown. The structure of a noise determination part is shown. The processing flowchart of a noise determination part is shown. An example of a filter for extracting high-frequency components and edges is shown. The processing flowchart of a noise analysis part is shown. An example of a color conversion method will be shown. 9 shows a flowchart of noise analysis processing in the second embodiment. 10 is a processing flowchart for selecting a mapping direction in the second embodiment. It is a figure explaining the mapping direction of Example 2. FIG. Indicates the color reproduction range of the output system.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows the configuration of Embodiment 1 of the present invention. In FIG. 1, 101 is output image data, 102 is a reproduction color determination unit, 103 is a dark region extraction unit, 104 is a clustering unit, 105 is a noise determination unit, 106 is a noise analysis unit, 107 is a conversion method determination unit, 108 Is the color conversion processing unit, and 109 is the color-corrected output image data.

  In the following description, it is assumed that the color information of the output image is basically composed of RGB data, and each processing unit appropriately converts the RGB value into a CIEL * a * b * value or an L * C as necessary. * H * value (converted from coordinates (a *, b *) on the a * b * plane in the CIE L * a * b * color system to polar coordinate display (C *, H *)) Processing shall be performed.

  FIG. 2 shows a processing flowchart of the present embodiment. In step 101, the reproduction color determination unit 102 determines whether the color information (L * a * b *) of the output image data 101 is an outside color or an inside color of the output system color reproduction range. judge. As a result of the determination, if the color is out of the output color reproduction range (YES in step 101), the process proceeds to step 102. If the color is not out of the output color reproduction range (NO in step 101), step 110 is performed. The color conversion process is not performed.

  In step 102, the dark area extraction unit 103 extracts a dark area based on the color information in the output image data. Hereinafter, the extraction method will be described in detail.

  The dark area extraction unit 103 extracts the dark area of the image based on the color information of the output image data 101 (at least pixel position information and color information (L * a * b *) correspond).

  That is, the brightness histogram of the image is taken, the distribution of the brightness histogram is analyzed, and a threshold th_dark for adaptively determining whether the area is a dark area is determined. Alternatively, a threshold value th_dark that is common to all images is simply set in advance, and if the lightness L * of the target pixel is smaller than the preset threshold value, it is determined that the pixel is in the dark area, and the lightness L * of the target pixel is If the threshold value is equal to or greater than a preset threshold value, it may be determined that the pixel is in the non-dark area, and the dark area is extracted by sequentially performing the process on each pixel. It is appropriate to set the threshold th_dark common to all images to L * = 20 empirically.

  Next, in step 103, it is determined whether or not the target pixel is a pixel in the dark area. Compared with the threshold th_dark determined in step 102, and if it is greater than or equal to th_dark, it is determined that it is not a dark area (NO in step 102), and the routine proceeds to step 108. If it is a dark area (YES in step 103), the routine proceeds to step 104.

  In step 104, the clustering unit 104 classifies the input pixel data into one or more clusters based on the color difference. By this processing, pixels that are out of the color reproduction range of the output system and are dark areas are classified into clusters that are considered to be substantially the same. “Substantially the same” means that the brightness, saturation, and hue are substantially the same. Each pixel is given a cluster number after classification together with pixel position information and color information. The cluster classification method will be described later.

  After clustering, the process proceeds to step 105. In step 105, the noise analysis unit 106 analyzes a noise component in pixels to which the same cluster number is assigned. The noise analysis unit 106 receives the noise determination result from the noise determination unit 105 at the same time as the clustering result from the clustering unit 104, and performs noise analysis based on both pieces of information. Details of the processing in the noise determination unit 105 and the processing in the noise analysis unit 106 will be described later.

  In step 106, the noise analysis unit 106 determines the presence of noise for each cluster. If noise is present (YES in step 106), the process proceeds to step 107, and if noise is not present (NO in step 106). Goes to step 108. The conversion method determination unit 107 executes the processes in steps 107 and 108. Different color conversion methods are selected and determined depending on whether noise is present or not.

  In step 107, after the conversion method determining unit 107 selects the first color conversion method, in step 109, the color conversion processing unit 108 performs color correction by the selected color conversion method for pixels outside the color reproduction range. The process proceeds to step 111. In step 110, since the pixel is within the color reproduction range, the process proceeds to step 111 without performing any particular color conversion processing. In step 111, the result of performing the color conversion process in step 109 (or not performing the color conversion process) is output as a corrected output image 109, and the process ends.

  Next, details of the clustering process in step 104 will be described. In step 104, the clustering unit 104 classifies the cluster into at least one or more clusters based on the input pixel color information.

  FIG. 3 shows the configuration of the clustering unit 104. The clustering unit 104 includes a reproduction color determination result by the reproduction color determination unit 102, a dark part region extraction threshold th_dark for determining whether or not it is a dark region, and L * a * b * which is color information of the input pixel. Is entered.

FIG. 4 is a processing flowchart of the clustering unit 104. In step 202, it is determined whether the target pixel is a dark area and is outside the color reproduction range. To determine whether it is a dark area, the dark area threshold th_dark is compared with the L * value of the target pixel,
If L * ≧ th_dark of the pixel of interest, it is not a dark area, and if L * <th_dark of the pixel of interest, it is determined that it is a dark area.
In addition, the determination result in the reproduction color determination unit 102 is used as it is to determine whether it is out of the color reproduction range.

  As described above, when the target pixel is a dark area and outside the color reproduction range (YES in step 202), the process proceeds to step 203. Otherwise (NO in step 202), the process returns to step 201 to perform processing for the next pixel.

  In step 203 and thereafter, clustering is performed using L * a * b * of the target pixel. When the target pixel is input for the first time, the number of clusters N is in a state initialized to N = 0 (step 200), and therefore the condition branching by color difference (step 206) advances to “NO” and advances to the cluster number. 1 is added as a new cluster, the number of clusters is set to N = 1, and the number of pixels of cluster number 1 is set to n (1) = 0 (step 209). Further, the L * a * b * value of the target pixel is set to the average L * a * b * of cluster number 1 (step 210).

  After that, the color difference with the average L * a * b * of each cluster is obtained for the L * a * b * values of the target pixel that are sequentially input (step 204), and the cluster number j with the smallest color difference among them is obtained. And the color difference dE_min are extracted (step 205).

  If the color difference dE_min is equal to or smaller than a preset threshold value dE_th (YES in step 206) in the conditional branch by color difference (step 206), the process proceeds to step 207 to add input pixels to the cluster j, and the average L * a Recalculate * b *. The number of pixels n (j) in cluster j is incremented by 1, and the average value of L * a * b * is recalculated according to the following equation.

(Average L * after recalculation) = ((Average L * before recalculation) × (n (j) −1) + (L * of input pixels)) / n (j)
(Average a * after recalculation) = ((Average a * before recalculation) × (n (j) −1) + (a * of input pixels)) / n (j)
(Average b * after recalculation) = ((Average b * before recalculation) × (n (j) −1) + (b * of input pixel)) / n (j)
If the color difference dE_min is larger than the preset threshold value dE_th (NO in step 206) in the conditional branch by color difference (step 206), the process proceeds to step 209 to add a new cluster, and the number of clusters N is counted up. The number of pixels of the new cluster is set to 0, and the L * a * b * of the input pixel is set to the average L * a * b * (steps 209 and 210).

  DE_th represents a boundary color difference indicating whether or not a new cluster is to be added, and dE_th is set in advance in consideration of the fact that an area in which a dark portion is not clearly visible is targeted.

  If the pixel of interest belongs to any one of the clusters, a cluster number is given together with color information corresponding to the pixel position (step 211) and used in subsequent noise analysis.

  When the processing is completed for all the pixels (YES in step 201), the clustering is completed. If not finished yet, the processing from step 202 is repeated. When clustering is completed, the number of clusters, which cluster each pixel of interest belongs to (or which cluster does not belong), and an average L * a * b * value in each cluster can be obtained.

  Next, the noise analysis method performed in step 105 will be described. In the noise analysis in step 105, it is analyzed whether or not noise exists in each cluster classified into clusters. As preprocessing for that, first, the noise determination unit 105 performs noise determination on the entire output image. Details of this noise determination will be described.

  FIG. 5 shows a configuration of the noise determination unit 105. The color space conversion unit 501 converts RGB signals of input image data into L * C * H * color space conversion. The high frequency component extraction unit 502 extracts high frequency components for each L *, C *, and H * signal after color space conversion. The edge determination unit 503 performs edge determination for each L *, C *, and H * signal.

  The noise determination units 504 to 506 perform noise determination based on the input high frequency component extraction result and the edge determination result. The overall determination unit 507 makes a final determination as to whether or not the target pixel is noise based on the noise determination result in the L *, C *, and H * signals.

  FIG. 6 shows a process flowchart of the noise determination unit. In step 301, first, the color space conversion unit 501 converts the input RGB data into the L * C * H * space. Hereinafter, for simplification of description, the description will be made as an equivalent processing flow regardless of L *, C *, and H *. However, the processing flow is actually executed for each L *, C *, and H * signal. .

  Next, the signal subjected to the color space conversion process is subjected to a spatial filter process to extract a high frequency component. Extraction is performed for each of L *, C *, and H *. In order to extract a high frequency component, the high frequency component extraction unit 502 performs a high-pass filter processing calculation, and sets the calculation result as an extracted value (step 302). The filter calculation method will be described later.

  Next, the extracted value is compared with a predetermined threshold th_high, and if it is equal to or greater than th_high (YES in step 303), the pixel is determined to be a noise candidate, and the process proceeds to step 304. If it is not greater than th_high (NO in step 303), the process proceeds to step 307.

  Next, the edge determination unit 503 performs a filter processing calculation for determining whether or not the pixel is an edge, and calculates an edge feature amount (step 304). The filter calculation method will be described later.

  In this embodiment, the pixel to be determined as noise is not a pixel like an edge but noise generated in a relatively flat region. For this purpose, two features of a high frequency component and an edge feature amount are used.

  The filter calculation process for high frequency component extraction performed in step 302 is, for example, a 3 × 3 filter calculation (the pixel of interest is a central pixel) as shown in FIG. 7A, and the absolute value of the calculated value is used as the high frequency component. And In addition, the filter calculation processing for edge feature amount extraction performed in step 304 performs, for example, four 3 × 3 (the target pixel is the central pixel) filter calculation as shown in FIGS. The maximum value among the absolute values of the four filter calculation results is calculated as an edge feature amount ((b) is a vertical edge, (c) is a horizontal edge, (d) and (e) are diagonal directions. Extract edges).

  Next, the noise determination units 504 to 506 compare the extracted edge feature amount with a predetermined edge determination threshold th_edge, and if it is equal to or greater than th_edge, the pixel is determined to be an edge (YES in step 305), and step If it is not th_edge or more (NO in step 305), the process proceeds to step 306.

  In step 306, it is determined that the target pixel is a noise pixel, and the process proceeds to step 308. In step 307, it is determined that the target pixel is not a noise pixel, and the process proceeds to step 308.

  In step 308, it is determined whether or not the noise determination has been completed for all the pixels. When the determination is completed for all the pixels, the process is terminated. If not, the process returns to step 301 to continue the process for the next pixel of interest.

  In the present embodiment, an example in which a 3 × 3 high-pass filter is used for high-frequency component extraction and a 3 × 3 edge extraction filter is used for edge feature amount extraction has been described. A simple filter may be used. Further, th_high and th_edge use the same threshold value for the L *, C *, and H * signals, but separate threshold values may be set for each signal.

  Next, processing in the comprehensive determination unit 507 will be described. The overall determination unit 507 finally determines that the target pixel is noise when any one of the L *, C *, and H * signals is determined to be noise in the noise determination flow described above. . The overall determination result is, for example, “1” for a pixel that is finally determined to be noise, and “0” otherwise.

  Next, the noise analysis process in step 105 will be described in detail. Noise analysis is performed in the noise analysis unit 106. The noise analysis unit 106 uses the clustering result from the clustering unit 104 and the noise determination result from the noise determination unit 105 to output a pixel that is out of the color reproduction range and is a dark region in the output image data. Analyzes the presence or absence of noise in pixels classified into the same cluster.

  FIG. 8 shows a processing flowchart in the noise analysis unit 106. In step 401, it is referred to whether or not a cluster number is assigned to the target pixel. As a result of the reference, when the cluster number is assigned to the target pixel (YES in step 402), the noise determination result of the target pixel is referred to (step 403). If no cluster number is assigned to the pixel of interest (NO in step 402), the process proceeds to step 406.

  As a result of referring to the noise determination result of the target pixel, if the target pixel is noise (YES in step 404), the noise count value in the cluster number assigned to the target pixel is counted up and updated (step 405). .

For example, if the noise count value is num_NOISE (n),
num_NOISE (n) = num_NOISE (n) +1
{Where n is the cluster number and num_NOISE (n) is initialized to 0}
In step 406, it is determined whether or not the processing has been performed for all of the output image data. When the processing is completed for all pixels (YES in step 406), the processing proceeds to step 407 and is not completed for all pixels. If so, the process returns to step 401 to repeat the noise counting process. By the processing so far, the number of noise pixels in each cluster is counted.

  Next, in step 407, the target cluster number is set to the first cluster, and the noise count value (num_NOISE (n)) in the set cluster is compared with a predetermined threshold th_num. As a result of the comparison, if the noise count value of the target cluster (eg, num_NOISE (1) for the first cluster) is equal to or greater than the threshold (YES in step 408), it is determined that noise exists in the target cluster (step 409). As a result of the comparison, if the noise count value of the target cluster is not greater than or equal to the threshold value (NO in step 408), it is determined that no noise exists in the target cluster (step 410).

  In step 411, it is determined whether or not the analysis has been completed for all the clusters. If the analysis has been completed for all the clusters (YES in step 411), the process is terminated. If the processing has not been completed for all clusters (NO in step 411), the next cluster number is set as a cluster to be analyzed, and the processing from step 407 is repeated.

  Next, a method for determining a color conversion method will be described. In Step 107 and Step 108, the color conversion method is selected according to the determination result of whether or not noise exists in the cluster. The conversion method determination unit 107 that performs this processing will be described in detail.

  The conversion method determination unit 107 selects a color conversion method for mapping in the direction of a specific point as shown in FIG. 9A as a conversion method when it is determined that noise exists in the cluster. As a conversion method when it is determined that no noise exists in the cluster, a conversion method for mapping in the minimum color difference direction as shown in FIG. 9B is selected. The conversion method is selected for each cluster.

  As described in the processing of the clustering unit 104, information on the average L * a * b * in each cluster is known, and the average hue H * of each cluster can be obtained from this information. For the hue of the cluster that is determined to have noise, select a color conversion method that maps in the direction of the specific point, and in the hue of the cluster that has been determined to have no noise, map in the direction of minimum color difference. The mapping direction is selected and determined according to the average hue.

  As described above, the pixels are out of the color reproduction range of the output system and are classified into clusters for the pixels in the dark area, and noise analysis is performed in each cluster so that noise is not noticeable if noise exists. Since the color conversion method for the average hue of the cluster is selected, the optimal color conversion process is performed partially on the output image, and noise can be made inconspicuous.

  According to the present embodiment, when a part of the color information of the output image information is wider than the color reproduction range of the image output system, in the image processing apparatus for converting the color information that cannot be reproduced, the color reproduction range of the image output system A color reproduction range determination unit that determines whether or not the color is outside, a dark region extraction unit that extracts a dark region in the output image, and one pixel that is outside the color reproduction range and is a dark region. Cluster classification means for classifying the above clusters, noise determination means for determining whether or not each pixel of the output image is noise, and pixels determined by the noise determination means in the classified clusters And a noise analysis unit that determines whether noise exists in the cluster, and a color conversion of the cluster that has been determined that noise exists by the noise analysis unit A color conversion method determining means for making the method different from a color conversion method when it is determined that no noise exists, and a color conversion processing means for performing color conversion processing by the color conversion method determined by the color conversion method determining means. Therefore, based on the characteristics of the pixels that make up the cluster, the color conversion method differs depending on whether noise is present or not, so optimal color conversion according to the partial characteristics of the output image A method can be selected.

  Further, according to the present embodiment, when the noise analysis unit determines that noise exists, the color conversion method determination unit selects a color conversion method for mapping in the direction of a specific point, so that noise exists. In such a cluster, color conversion can be performed so as to make noise inconspicuous.

  Further, according to this embodiment, when there are a plurality of clusters determined to have noise by the noise analysis unit, the color conversion method determination unit performs color conversion based on the cluster representative color for each cluster. Since the method is determined, an optimum color conversion method can be determined in a plurality of clusters.

  Further, according to the present embodiment, the cluster representative color is an average hue value of color information of pixels constituting the cluster, so that an effective color conversion method can be obtained by grasping the hue where the cluster is located in the color space. Can be determined.

  In addition, according to the present embodiment, the noise determination unit determines whether each pixel is a noise pixel with respect to lightness, hue, and saturation, so that the noise pixel can be determined in detail.

  Furthermore, according to the present embodiment, the noise determination means determines by extracting the high frequency component excluding the edge, so that the high frequency component other than the edge can be determined as noise.

  In the first embodiment, the method of determining whether noise exists for each cluster and selecting the mapping direction based on the determination result has been described. However, in the present embodiment, based on the characteristics of the noise distribution for each cluster. A method for selecting the mapping direction will be described.

  The overall configuration of the present embodiment is the same as that of FIG. 1 described in the first embodiment, and the entire processing flow is also the same as that of FIG.

  Since the noise feature for each cluster is analyzed, and the part for selecting the mapping direction is different based on the analyzed noise feature, noise analysis and selection of a color conversion method will be described.

  First, noise analysis in the present embodiment will be described. FIG. 10 shows a process flowchart of noise analysis in the present embodiment. This processing flow shows processing in the noise analysis unit 105.

  In step 501, reference is made to whether or not a cluster number is assigned to the target pixel. As a result of the reference, when the cluster number is assigned to the target pixel (YES in step 502), the noise determination result of the target pixel is referred to (step 503). When the cluster number is not assigned to the target pixel (NO in step 502), the process proceeds to step 507.

  If the pixel of interest is noise as a result of referring to the noise determination result of the pixel of interest (YES in step 504), the standard deviation using the pixel of interest as the noise feature amount in the cluster number assigned to the pixel of interest is calculated. / Update (step 505). The standard deviation is calculated based on a signal obtained by performing L * C * H * space conversion on the RGB data described in the first embodiment. The standard deviation to be calculated / updated is calculated for each L *, C *, and H * signal.

  Next, the noise count value in the cluster number assigned to the target pixel is counted up and updated (step 506). In step 507, it is determined whether or not processing has been performed for all input pixels. If processing has been performed for all pixels (YES in step 507), the process proceeds to step 508; otherwise (NO in step 507). Returning to step 501, the noise analysis process is repeated.

  In step 508, the target cluster is set as the first cluster, and the first noise count value (num_NOISE (1)) is compared with a predetermined threshold th_num. As a result of the comparison, if the noise count value of the target cluster (in this case, num_NOISE (1) because it is the first cluster) is equal to or greater than the threshold (YES in step 509), it is determined that noise exists in the target cluster (step 510). As a result of the comparison, if the noise count value of the target cluster (in this case, num_NOISE (1) because it is the first cluster) is not equal to or greater than the threshold (NO in step 509), it is determined that no noise exists in the target cluster (step 511). .

  In step 512, it is determined whether or not the processing has been completed for all the clusters. If the processing has been completed for all the clusters (YES in step 512), the processing ends. If not (NO in step 512), the process returns to step 508, the next cluster number is set as the target cluster, and the subsequent processing is repeated.

  Next, a method for selecting a color conversion method in this embodiment will be described. In the first embodiment, control is performed so that the mapping direction is different when noise is present. However, in the second embodiment, a method of changing the mapping direction based on the standard deviation that is a feature amount of noise will be described. .

  The conversion method determination unit 107 compares the standard deviation of L * and C * in the noise of each cluster analyzed by the noise analysis unit 106, and when the standard deviation of L * is large, the noise of L * is easily compressed. The mapping direction is determined in the direction of the specific point.

  FIG. 11 is a process flowchart for selecting a mapping direction in the conversion method determination unit 107. In step 601, the standard deviations of L * and C * in the target cluster are referred to. The target cluster here is a cluster that is determined to have noise by analysis in the noise analysis unit 106.

  In step 602, the standard deviations of L * and C * are compared. If the standard deviation of L * is greater than or equal to the standard deviation of C * (YES in step 602), the mapping direction increases the degree of compression of L *. Is determined (step 603). If not (NO in step 602), the mapping direction that increases the degree of compression of L * is not selected (step 604).

  In step 605, it is determined whether or not the processing has been completed for all the target clusters. If the processing has been completed, the processing ends. If not, the process returns to step 601 to repeat the process for determining the mapping direction for the next target cluster.

  The mapping direction in which the degree of compressing L * is high is, for example, as shown in FIG. 12, a direction indicating the degree of compression higher than the case where the specific point P is normal (dotted line) (the direction of the specific point P ′).

  As described above, according to the characteristics of noise, when the L * noise is high, it is possible to increase the degree of lightness compression and select a mapping direction that can effectively reduce the noise. .

  According to the present embodiment, the noise analysis unit further includes a distribution analysis unit that analyzes variations in brightness and saturation of noise pixels belonging to the cluster, and the color conversion method determination unit includes a variation in brightness direction in the cluster. If it is larger than the degree direction, mapping is performed in the direction of a specific point that increases the rate of lightness compression, so select the optimal color conversion method according to the direction of variation of the noise pixels that make up the cluster. Can do.

  In Example 2, the noise of the target cluster was analyzed, and when the standard deviation of L * in the noise was larger than the standard deviation of C *, the mapping direction was selected in the direction of a specific point where the degree of compression of L * was high. However, the direction in which the brightness is preserved as shown in FIG. 9C may be selected as the mapping direction so that at least the distribution of L * is not widened.

  According to this embodiment, the noise analysis means further comprises distribution analysis means for analyzing the distribution of lightness and saturation of the noise pixels belonging to the cluster, and the color conversion method determination means is the distribution of noise in the lightness direction in the cluster. Is larger than the saturation direction, mapping is performed in a direction in which the brightness is preserved, so that color conversion can be performed so that at least noise in the brightness direction is not enlarged.

  According to the present invention, a storage medium in which a program code of software for realizing the functions of the above-described embodiments is recorded is supplied to a system or apparatus, and a computer (CPU or MPU) of the system or apparatus is stored in the storage medium. This is also achieved by reading and executing the code. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments. As a storage medium for supplying the program code, for example, a hard disk, an optical disk, a magneto-optical disk, a nonvolatile memory card, a ROM, or the like can be used. Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) operating on the computer based on an instruction of the program code. A case where part or all of the actual processing is performed and the functions of the above-described embodiments are realized by the processing is also included. Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. This includes the case where the CPU or the like provided in the board or function expansion unit performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing. Further, the program for realizing the functions and the like of the embodiments of the present invention may be provided from a server by communication via a network.

DESCRIPTION OF SYMBOLS 101 Output image 102 Reproduction color determination part 103 Dark area extraction part 104 Clustering part 105 Noise determination part 106 Noise analysis part 107 Conversion method determination part 108 Color conversion process part 109 Corrected output image

JP-A-61-288690 Japanese Patent No. 3502496 Japanese Patent Laid-Open No. 2002-027263

Claims (10)

In the image processing apparatus that converts color information that cannot be reproduced when part of the color information of the output image information is wider than the color reproduction range of the image output system, whether or not the color is outside the color reproduction range of the image output system A color reproduction range determination unit that determines whether or not, a dark region extraction unit that extracts a dark region in the output image, and a pixel that is outside the color reproduction range and that is a dark region is classified into one or more clusters. Cluster classification means, noise determination means for determining whether or not each pixel of the output image is noise, and in the classified cluster, extract features for the pixels determined by the noise determination means as noise, and determining the noise analyzing means whether noise is present in the cluster, as the color conversion of the cluster it is determined that noise is present by the noise analysis unit, especially Selecting a first color conversion method of mapping to the point direction, as the color conversion of the cluster it is determined that there is no noise, the color conversion selecting a different second color conversion method and the first color conversion method An image processing apparatus comprising: method determining means; and color conversion processing means for performing color conversion processing by the color conversion method determined by the color conversion method determining means. The noise analysis unit further includes a distribution analysis unit that analyzes variations in brightness and saturation of noise pixels belonging to the cluster, and the color conversion method determination unit has a variation in brightness direction in the cluster that is greater than the saturation direction. If large, the image processing apparatus according to claim 1, wherein the mapping the specific point direction as the proportion to be compressed lightness increases. The noise analysis means further comprises a distribution analysis means for analyzing the brightness and saturation distribution of the noise pixels belonging to the cluster, and the color conversion method determining means is configured such that the noise distribution in the brightness direction in the cluster is a saturation direction. If more greater, the image processing apparatus according to claim 1, wherein the mapping the direction brightness is saved. When there are a plurality of clusters determined to have noise by the noise analysis unit, the color conversion method determination unit determines a color conversion method based on a cluster representative color for each cluster. the image processing apparatus according to any one of claims 1 to 3. 5. The image processing apparatus according to claim 4 , wherein the cluster representative color is an average hue value of color information of pixels constituting the cluster.   The image processing apparatus according to claim 1, wherein the noise determination unit determines whether each pixel is a noise pixel with respect to brightness, hue, and saturation.   The image processing apparatus according to claim 1, wherein the noise determination unit determines a noise pixel by extracting a high frequency component excluding an edge. In the image processing method for converting color information that cannot be reproduced when part of the color information of the output image information is wider than the color reproduction range of the image output system, whether or not the color is out of the color reproduction range of the image output system A color reproduction range determination step for determining whether or not, a dark region extraction step for extracting a dark region in the output image, and pixels that are outside the color reproduction range and are in the dark region are classified into one or more clusters. A cluster classification step, a noise determination step for determining whether each pixel of the output image is noise, and, in the classified cluster, extract a feature for the pixel determined as noise by the noise determination step, and determining the noise analysis step whether noise is present in the cluster, as the color conversion of the cluster it is determined that noise is present by the noise analysis unit, especially Selecting a first color conversion method of mapping to the point direction, as the color conversion of the cluster it is determined that there is no noise, the color conversion selecting a different second color conversion method and the first color conversion method An image processing method comprising: a method determining step; and a color conversion processing step for performing color conversion processing by the color conversion method determined in the color conversion method determining step. A program for causing a computer to implement the image processing method according to claim 8 . A computer-readable recording medium on which a program for causing a computer to implement the image processing method according to claim 8 is recorded.

Images