Image processing apparatus

Abstract

An uneven region is extracted with high precision from a captured image obtained by capturing an image of a dyed knitted fabric, and the dyeing quality of the knitted fabric is evaluated. An edge in a captured image of a captured knitted fabric is detected, a high-contrast region in which the gradient of the edge (11) is equal to or greater than a first predetermined value is extracted, and a non-uniform candidate region serving as a candidate for a non-uniform region is extracted from the high-contrast region (steps S105-S107). On the other hand, in the captured image, a plurality of highly similar regions having high similarities with the sample image of the trunk are extracted, information on the positions of representative points of the trunk corresponding to each of the highly similar regions is acquired, and information on the positions of the trunk is acquired on the basis of the information on the positions of the representative points (steps S108-S110). Afterwards, among the non-uniform candidate regions, regions that extend continuously in the course direction across a predetermined number or more of trunks are extracted as non-uniform regions (step S111), and the dyeing quality of the knitted fabric is evaluated on the basis of the extracted non-uniform regions (step S112).

Description

Technical Field

[0001] The present invention relates to an image processing apparatus for extracting uneven regions corresponding to uneven dyeing from photographed images obtained by photographing dyed knitted fabrics. Background Technology

[0002] Patent Document 1 describes an image processing apparatus for extracting uneven regions corresponding to uneven dyeing from an image obtained by photographing a dyed knitted fabric. After converting the photographed image to grayscale, removing the knitted fabric's stitch patterns, and smoothing it, the image processing apparatus of Patent Document 1 detects edges within the photographed image where brightness changes abruptly. The apparatus then extracts high-contrast regions where the gradient (slope of brightness) of the edges is above a predetermined value. Next, the apparatus removes noise from the processed image and corrects the slope. Specifically, it calculates the slope of several high-contrast regions using the least squares method and rotates the image based on the average of the calculated slopes, thereby correcting the image's slope. The apparatus then extracts regions within the high-contrast regions that exhibit actual uneven dyeing characteristics, such as elongated extension in the horizontal direction, as uneven regions. Specifically, high-contrast regions satisfying both the condition that "the ratio of the dimension in the horizontal direction to the dimension in the vertical direction, i.e., the aspect ratio, is above a predetermined value" and the condition that "the dimension in the vertical direction is within a predetermined range" are extracted as uneven regions. Furthermore, in Patent Document 1, the dyeing quality of knitted fabrics is evaluated based on the uneven regions extracted by the image processing device. Existing technical documents Patent documents

[0003] Patent Document 1: Japanese Patent Application Publication No. 2021-140373 Summary of the Invention The problem that the invention aims to solve

[0004] In Patent Document 1, after correcting the slope of the image as described above, a region extending elongated in the horizontal direction is extracted from the high-contrast region as an uneven region. However, due to the slight deviation in the stitch pattern, even after correcting the slope of the image, there is still a possibility that the horizontal direction in the corrected image is inconsistent with the actual horizontal direction of the stitch pattern, resulting in a decrease in the accuracy of uneven region extraction.

[0005] The purpose of this invention is to provide an image processing apparatus capable of extracting uneven regions from images obtained by photographing dyed knitted fabrics with higher accuracy. Methods for solving problems

[0006] The image processing apparatus of the first embodiment is an image processing apparatus for extracting uneven regions corresponding to uneven dyeing from an image obtained by photographing a dyed knitted fabric. It performs the following processes: brightness information acquisition processing, acquiring brightness information, which is brightness information at various locations in the photographed image; unevenness candidate region information acquisition processing, acquiring unevenness candidate region information based on the brightness information, which is information about the location of an unevenness candidate region that could become a candidate for the uneven region; stitch information acquisition processing, acquiring stitch information based on the brightness information, which is information about the location of stitches in the photographed image; and unevenness region extraction processing, extracting the uneven region based on the unevenness candidate region information and the stitch information.

[0007] The direction of uneven dyeing extension is determined based on the direction of the stitches in the knitted fabric. Furthermore, due to factors such as stitch deviations, the stitch direction may sometimes deviate depending on the area of ​​the knitted fabric. In this method, information about the location of candidate uneven regions (i.e., uneven candidate region information) is obtained based on brightness information. Additionally, information about the location of stitches in the captured image (i.e., stitch information) is obtained based on brightness information. Then, the uneven region is extracted based on the uneven candidate region information and the stitch information. Therefore, even with deviations in stitch direction, uneven regions can be extracted with high accuracy.

[0008] The second embodiment of the image processing apparatus is that, in the first embodiment of the image processing apparatus, during the stitch information acquisition process, information on the positions of multiple loops constituting the stitch is acquired as the stitch information.

[0009] The direction of the stitches in a knitted fabric is determined by the direction in which multiple loops are arranged. However, due to factors such as stitch deviations, the direction of the multiple loops can sometimes deviate depending on the area of ​​the knitted fabric. According to this solution, uneven regions are extracted with high precision based on information about candidate uneven regions and the positions of multiple loops, which serve as stitch information. Therefore, even if the direction of the multiple loops is deviated, uneven regions can still be extracted with high precision.

[0010] The third embodiment of the image processing apparatus is that, in the image processing apparatus of the second embodiment, there is a storage unit for storing sample image information, which is the information of the sample image of the stitch. In the stitch information acquisition processing, based on the brightness information and the sample image information, multiple high similarity regions in the captured image that have a high similarity to the sample image are extracted, and the position information of the multiple stitches is obtained using the position information of the multiple high similarity regions.

[0011] The region containing the stem in the captured image has a high similarity to the sample image. In this scheme, multiple highly similar regions with high similarity to the sample image can be extracted from the captured image, and the position information of the stem can be obtained using the position information of the highly similar regions.

[0012] The image processing apparatus of the fourth embodiment is, in the image processing apparatus of the third embodiment, in the stitch information acquisition processing, based on the brightness information, for each of the plurality of highly similar regions, information on the position of the representative point of the loop is obtained, and based on the information on the position of the multiple representative points related to the plurality of highly similar regions, information on the position of the plurality of loops is obtained.

[0013] According to this scheme, the position information of the representative points of the loop trunk in each highly similar region is obtained, and the position information of the loop trunk is obtained based on the position information of the representative points of multiple loop trunks. Thus, the position information of the loop trunk can be obtained with high precision.

[0014] The fifth embodiment of the image processing apparatus is as follows: In the image processing apparatus of any one of the second to fourth embodiments, in the uneven candidate region information acquisition process, based on the brightness information, edges contained in the captured image are detected, and high-contrast regions with an edge gradient of a first predetermined value or higher are extracted. Regions in the high-contrast regions whose aspect ratio (length-to-width ratio) in the horizontal direction is a second predetermined value or higher are extracted as the uneven candidate regions. In the uneven region extraction process, continuous regions in the uneven candidate regions that overlap with a predetermined number or more loops arranged continuously along the horizontal direction are extracted as the uneven regions.

[0015] According to this scheme, high-contrast regions with an aspect ratio of at least a second predetermined value relative to the dimension in the longitudinal direction are extracted as candidate regions for unevenness. Furthermore, continuous regions within these candidate regions that overlap with a predetermined number or more loops arranged consecutively along the transverse direction are also extracted as uneven regions. Thus, high-contrast regions with high aspect ratios—that is, regions exhibiting actual unevenness in dyeing, such as elongation in the transverse direction—are extracted as candidate regions for unevenness. Continuous regions within these candidate regions that are consecutively arranged along the loop direction are then extracted as uneven regions. Therefore, uneven regions can be extracted with high precision.

[0016] The image processing apparatus of the sixth embodiment is, in the image processing apparatus of the fifth embodiment, in the uneven candidate region information acquisition process, extracting a region in the high contrast region whose aspect ratio is greater than or equal to the second predetermined value and whose dimension in the longitudinal direction is within a predetermined range as the uneven candidate region.

[0017] According to this method, regions within the high-contrast area that have an aspect ratio of 2 or higher and a length within a predetermined range in the longitudinal direction are extracted as candidate regions for unevenness. Thus, it is possible to extract high-contrast regions with a high aspect ratio and short length in the longitudinal direction—that is, high-contrast regions exhibiting uneven coloring characteristics such as elongation in the transverse direction—as candidate regions for unevenness.

[0018] The image processing apparatus of the seventh embodiment is as follows: In the image processing apparatus of any of the second to fourth embodiments, in the stitch information acquisition process, as information on the positions of the plurality of loops, information on the positions of the plurality of loops in the captured image is acquired, which respectively contains one of the loops and is arranged in the longitudinal and transverse directions. In the unevenness candidate region information acquisition process, based on the brightness information and the stitch information, a representative value of the brightness of the thread region where the thread is located is calculated for each of the plurality of loops, i.e., the representative brightness, and the loops whose representative brightness meets a predetermined condition are extracted as the unevenness candidate regions. In the unevenness region extraction process, the region is extracted as the unevenness region, that is, the region spans more than a predetermined number of the unevenness candidate regions arranged continuously along the transverse direction.

[0019] In this scheme, for each of the multiple loop regions, the representative brightness of the thread region is calculated, and the loop regions whose representative brightness meets predetermined conditions are extracted as candidate unevenness regions. Furthermore, regions are extracted as unevenness regions if they span a predetermined number or more consecutive candidate unevenness regions arranged along the horizontal direction. This allows for the extraction of unevenness regions with high accuracy. Invention Effects

[0020] According to the present invention, uneven regions can be extracted with higher precision. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of a testing device used to evaluate the dyeing quality of knitted fabric samples. Figure 2 This is a magnified schematic diagram of a portion of a knitted fabric sample. Figure 3 This is a flowchart illustrating a series of processes performed by the image processing apparatus in the first embodiment. Figure 4 middle, Figure 4 (a) is a diagram representing an example of an edge-detected image. Figure 4 Figure (b) is an example of an image extracted from a high-contrast region. Figure 4 (c) is a diagram representing an example of an image with uneven candidate region extraction. Figure 5 middle, Figure 5 (a) is a diagram representing an example of a sample image. Figure 5 (b) is a diagram used to illustrate the representative points of the circle stem. Figure 5 (c) is a diagram used to illustrate the location of the stem. Figure 6 middle, Figure 6 (a) is a diagram showing an example of a detected ring-shaped stem. Figure 6 (b) is to Figure 4 A magnified portion of (c) Figure 6 (c) is to Figure 6 (a) and Figure 6 The image obtained by overlapping (b). Figure 7 This is a flowchart illustrating a series of processes performed by the image processing apparatus in the second embodiment. Figure 8 middle, Figure 8 (a) is a diagram used to illustrate the circled area. Figure 8 (b) is a diagram showing an example of the distribution of average brightness of the silk thread areas in each loop area. Detailed Implementation

[0022] [First Embodiment] The following is a reference to the appendix. Figure 1 A suitable first embodiment of the present invention will be described below.

[0023] <Overview of the test equipment> Figure 1 This is a schematic diagram of the test device 1 used to evaluate the dyeing quality of knitted fabric samples according to the first embodiment. The test device 1 takes pictures of the knitted fabric sample S (the knitted fabric of the present invention) using a camera 3, performs image processing on the captured images using an image processing device 7, extracts the uneven areas corresponding to uneven dyeing from the captured images, and evaluates the dyeing quality of the knitted fabric sample S.

[0024] Here, we will first explain the knitted fabric sample S. Figure 2This is a magnified schematic diagram of a portion of a knitted fabric sample S. The knitted fabric sample S is, for example, a dyed fabric obtained by knitting synthetic fiber yarn Y from polyester using a single yarn. The knitted fabric sample S has the following structure: multiple needle loops N are formed along the transverse direction (lateral direction) through a single yarn Y; yarn Y is hooked onto each of the multiple needle loops N, and multiple other needle loops N are formed in the longitudinal direction (the longitudinal direction orthogonal to the transverse direction). That is, the stitches of the knitted fabric sample S consist of multiple needle loops N. Uneven dyeing on the yarn Y generally occurs continuously along the length of the yarn Y. Therefore, the uneven dyeing that occurs when dyeing the knitted fabric sample S is as follows: Figure 2 The medium-thick lines indicate unevenness that extends in stripes along the horizontal direction.

[0025] It should be noted that, for convenience, the horizontal direction is shown here as... Figure 2 The horizontal parallel direction is used to represent the vertical direction as... Figure 2 The longitudinal parallel direction. However, in actual knitted fabric samples S, due to the influence of stitch deviations, the direction of multiple loops N sometimes deviates, resulting in deviations in both the transverse and longitudinal directions according to the area of ​​the knitted fabric sample S.

[0026] return Figure 1 The following describes the experimental equipment 1. Experimental equipment 1 includes a camera 3, a support platform 4, two illuminations 5, a reflector 6, an image processing device 7, and a monitor 8. The camera 3, support platform 4, two illuminations 5, and reflector 6 are arranged inside a darkroom 2. The knitted fabric sample S is supported by the support platform 4 located in the center of the darkroom 2, with the opening of the knitted fabric tube facing upwards and downwards, i.e., horizontally in the transverse direction.

[0027] Camera 3 is horizontally positioned on the front of support platform 4. This configuration allows camera 3 to capture images of the front surface of the knitted fabric sample S with the transverse direction of the knitted fabric sample S approximately parallel to the transverse axis of the captured image. By setting the magnification of camera 3 and the distance between camera 3 and the knitted fabric sample S to predetermined values ​​and taking the image, the size of the knitted fabric sample S reflected in the captured image becomes known. In this embodiment, the transverse (horizontal) dimension of the knitted fabric sample S within the captured image is set to approximately 2056 pixels.

[0028] Two illumination lamps 5 are positioned rearward of the support platform 4 and are arranged symmetrically on either side of the support platform 4. The two illumination lamps 5 are located on a line inclined at approximately 45 degrees from the line connecting the camera 3 and the support platform 4 and passing through the camera 3. A reflector 6 is positioned near the camera 3 to reflect the light from the two illumination lamps 5 toward the knitted fabric sample S. Thus, the front surface of the knitted fabric sample S achieves a suitable brightness for photographing.

[0029] The image processing device 7 comprises a CPU, memory, hard disk, etc. Electrically connected to the camera 3, the image processing device 7 can acquire image data of the knitted fabric sample S captured by the camera 3. By performing predetermined image processing on the captured image of the knitted fabric sample S, the image processing device 7 extracts the uneven areas corresponding to the actual dyeing unevenness from the captured image. Furthermore, based on the extracted uneven areas, the image processing device 7 evaluates the dyeing quality of the knitted fabric sample S. A monitor 8 is connected to the image processing device 7. The monitor 8 can display the captured image of the knitted fabric sample S and the evaluation results of the dyeing quality of the knitted fabric sample S.

[0030] <Extraction of uneven regions and evaluation of staining quality> Next, the extraction of uneven regions and the evaluation of staining quality performed by the image processing device 7 will be explained. Figure 3 This is a flowchart illustrating a series of processes performed by the image processing device 7. The image processing device 7 first performs grayscale processing (step S101). In the grayscale processing, the image processing device 7 acquires image data of an image of the knitted fabric sample S captured by the camera 3 and converts the captured image to grayscale. By using the grayscale captured image for image processing, the computational processing speed can be increased. Furthermore, when edge detection processing is performed in step S105, it is possible to avoid edges becoming unclear due to hue influence. Additionally, when extracting the position of the loop N in step S110, it is possible to avoid the edges of the loop N becoming unclear due to hue influence. However, subsequent processing can proceed while maintaining the color image without performing grayscale processing.

[0031] Next, the image processing device 7 performs brightness information acquisition processing (step S102). In brightness information acquisition processing, the image processing device 7 acquires brightness information, i.e., brightness information, at each location of the grayscale captured image. Next, the image processing device 7 performs the processing steps S103 to S107 and the processing steps S108 to S110, which will be described below, in parallel.

[0032] In step S103, the image processing apparatus 7 performs a stitch removal process. In this process, the image processing apparatus 7 removes stitch patterns from the grayscale captured image based on the brightness information obtained in step S102. For removing stitch patterns, any technique described, for example, in Japanese Patent Application Publication No. 2015-212924, can be used. Specifically, a low-pass filter is used to remove periodic information having a repetition number of a×N or more, obtained by multiplying the number of repetitions (a known number) of loops N per unit length by a predetermined coefficient a. The coefficient a is, for example, any value from 0.7 to 0.9. However, the method for removing stitch patterns is not limited to the method described herein, and other methods may also be used.

[0033] Next, the image processing apparatus 7 performs a smoothing process (step S104). In the smoothing process, the image processing apparatus 7 smooths the captured image after removing the stitch patterns. Smoothing is performed to remove minor noise caused by residual stitch patterns not completely removed in step S103. For smoothing, a known Wiener filter can be applied, for example. The filter size is, for example, about 5-25 pixels × 10-60 pixels. However, the type and size of the filter are not limited to these. Alternatively, subsequent processing can be performed without performing a smoothing process.

[0034] Next, the image processing device 7 performs edge detection processing on the smoothed captured image (step S105). Edge detection processing is performed to detect edges that are regions of abrupt changes in brightness within the captured image. For edge detection processing, a known Prewitt filter can be applied. The filter size is, for example, 3 pixels × 3 pixels. However, the type and size of the filter are not limited to these. Figure 4 (a) is an example of an edge detection image 10A of an captured image in which edge 11 was detected through edge detection processing. Figure 4 As shown in the figures, the horizontal direction of each image corresponds to the horizontal column direction, and the vertical direction corresponds to the vertical column direction.

[0035] Next, the image processing apparatus 7 performs high-contrast region extraction processing (step S106). In the high-contrast region extraction processing, the image processing apparatus 7 extracts high-contrast regions 12 whose gradient (slope of brightness) of edge 11 is a first predetermined value or higher. Specifically, the image processing apparatus 7 performs binarization processing on the edge detection image 10A, setting the first predetermined value as a threshold, setting pixels whose gradient of edge 11 is a threshold or higher as white, and setting pixels whose gradient of edge 11 is less than the threshold as black. Figure 4 (b) is in Figure 4In the edge detection image 10A shown in (a), a high-contrast region extraction image 10B is extracted from the high-contrast region 12 in white. The aforementioned first predetermined value can be determined in the same way as described in Patent Document 1. Specifically, regarding the edge detection image 10A, a histogram is created with the gradient of the edge 11 as the horizontal axis and the number of pixels as the vertical axis. The upper limit value U of the gradient of the edge 11 obtained from the histogram is multiplied by a predetermined coefficient b, and the result b×U is set as the threshold for binarization processing. The coefficient b is, for example, any value from 0.15 to 0.5. However, the upper limit value can also be determined by a different method.

[0036] Next, the image processing apparatus 7 performs uneven candidate region extraction processing (step S107). In the uneven candidate region extraction processing, regions within the high-contrast regions 12 that have actual uneven coloring characteristics, such as extending elongatedly in the horizontal direction, are extracted as candidate uneven candidate regions 13 to become uneven regions. Specifically, high-contrast regions 12 that satisfy both the condition that "the ratio of the size in the horizontal direction to the size in the vertical direction, i.e., the aspect ratio, is a second predetermined value or higher" and the condition that "the size in the vertical direction is within a predetermined range" are extracted as uneven candidate regions 13. Thus, information about the position of the uneven candidate regions 13 is obtained.

[0037] Figure 4 (c) is from Figure 4 Image 10C, which is the high-contrast region extraction image 10B, extracted the uneven candidate region 13. That is, it is the... Figure 4 Image (b) is obtained by performing uneven candidate region extraction processing on the high-contrast region extraction image 10B. Figure 4 (b) and Figure 4 As can be seen from (c), by performing the uneven candidate region extraction process, short or thick high-contrast regions 12 that do not have the characteristics of actual staining unevenness are excluded, and long, thin high-contrast regions 12 that extend linearly in the transverse direction can be extracted as uneven candidate regions 13.

[0038] Furthermore, in the uneven candidate region extraction process, the horizontal direction of the high-contrast region extraction image 10B is set as the horizontal column direction, and the vertical direction of the high-contrast region extraction image 10B is set as the vertical column direction. Regions that satisfy the above two conditions are extracted as uneven candidate regions 13. That is, the uneven candidate regions 13 extracted in the uneven candidate region extraction process are extracted without considering the deviations of the knitted fabric sample S in the horizontal and vertical directions of the region.

[0039] It should be noted that, in the first embodiment, the processing of steps S105 to S107 together is equivalent to the uneven candidate region information acquisition processing of the present invention.

[0040] In step S108, the image processing device 7 performs high similarity region extraction processing. Here, as... Figure 1 As shown, the image processing apparatus 7 includes a storage unit 9. The storage unit 9 stores, for example... Figure 5 The information of a sample image P of a circle stem N is shown in (a). It should be noted that in this embodiment, the information of the sample image P is equivalent to the "sample image information" of this invention. Furthermore, for easier viewing of the accompanying drawings, in... Figure 5 In (a), the sample image P is magnified. In the high similarity region extraction process, the image processing device 7 compares each region of the captured image with the sample image P, and extracts multiple high similarity regions in the captured image that have a high similarity to the sample image P. Thus, it is possible to extract the approximate region where each stem N is located in the captured image. In the high similarity region extraction process, for example, a correlation matrix is ​​calculated based on the brightness data of each region of the captured image and the data of the sample image P, and high similarity regions are extracted based on the values ​​of the correlation coefficients in the correlation matrix. However, the method for extracting high similarity regions is not limited to this. Furthermore, although the illustration is omitted, the extracted high similarity regions are those with stem regions R (see reference 1) described later. Figure 5 (c) Similar regions. However, if a highly similar region is compared with the stem region R, the accuracy of the position relative to the actual stem N is low.

[0041] Next, the image processing apparatus 7 performs representative point acquisition processing (step S109). In the representative point acquisition processing, the image processing apparatus 7, based on the distribution of brightness in each of the multiple high-similarity regions extracted in step 108, such as... Figure 5 As shown in (b), information on the location of the representative point M of the stem N corresponding to the highly similar region is obtained. Figure 5 In (b), the case where the position of the lower right end of each loop N in the diagram is obtained as the position information of the representative point M of loop N is shown. However, for example, it is also possible to obtain the position information of each loop N in the diagram. Figure 5 Information about the lower left position in (b), and the position of each ring N. Figure 5 The information of the upper position in (b) and the information of other positions of each ring N are used as the information of the position of the representative point M.

[0042] Next, the image processing device 7 performs the stem position extraction process (step S110). In the stem position extraction process, the image processing device 7 obtains the position information of multiple stem regions R in the captured image based on the position information of multiple representative points M obtained in the representative point acquisition process, wherein each stem region R is a region where a stem N is located. Figure 5(c) shows an example of the loop region R obtained in the loop position extraction process. In the loop position extraction process, since the position information of the loop region R is obtained based on the position information of multiple representative points M, the obtained loop region R shows the position of the loop N with higher accuracy than the aforementioned high similarity region. Furthermore, the multiple loop regions R obtained in the loop position extraction process are arranged along the horizontal and vertical directions. However, due to the influence of the aforementioned stitch deviations, sometimes the arrangement direction of the multiple loop regions R is relative to... Figure 2 The horizontal direction shown (the horizontal direction of the captured image) and Figure 2 The vertical direction shown (the longitudinal direction of the image) is slightly tilted.

[0043] Furthermore, in the process of extracting the loop positions, the image processing device 7 divides each loop region R into a loop region V containing the thread Y and a background region W where the thread Y is not located, based on the relationship between the brightness at each position of each loop region R and a predetermined threshold, and generates... Figure 6 The image shown in (a) is a region-distinguishing image 15, which distinguishes the silk thread region V and the background region W by giving them a difference in brightness of more than a predetermined value.

[0044] It should be noted that in the first embodiment, the processes S108 to S110 together constitute the stitch information acquisition process of the present invention. Furthermore, in the first embodiment, Figure 6 The information on the position of the thread region V shown in the data of the region differentiation image 15 shown in (a) is the information related to the position of the loop N of the present invention (stitch information of the present invention).

[0045] After the processing in steps S103 to S107 and steps S108 to S110 is completed, the image processing device 7 then performs uneven region extraction processing (step S111). In the uneven region extraction processing, the image processing device 7 extracts the continuous region portion of the uneven candidate region 13 that overlaps with a predetermined number or more of the loop stems N arranged continuously along the horizontal direction as the uneven region 14, based on the position of the uneven candidate region 13 extracted in step S107 and the position of the loop stem N extracted in step S110.

[0046] For example, Figure 6 (b) is to Figure 4 Uneven candidate region extraction in image 10C of (c) Figure 6 (a) is an enlarged image of a portion of the same size. Furthermore, in the uneven region extraction process, the image processing device 7, for example, Figure 6 As shown in (c), the following region portion is extracted from the uneven candidate region 13 as the uneven region 14, which is the region portion in which... Figure 6The region differentiation image 15 of (a) and Figure 6 When the uneven candidate region extraction image 10C of (b) overlaps, it is a continuous region that overlaps with a predetermined number or more loops N arranged continuously along the transverse direction. Figure 6 In (c), different concentrations are shown for easier understanding of the accompanying figures. Figure 6 The position of the stem N in (a) and Figure 6 The location of the uneven candidate region 13 in (b). Figure 6 In the case of example (c), for instance, if the predetermined number is set to two, the regions in the uneven candidate regions 13 that are located within the multiple boxes shown in the figure with double-dotted lines are extracted as uneven regions 14.

[0047] Next, the image processing device 7 evaluates the dyeing quality of the knitted fabric based on the extracted uneven regions 14 (step 112). The method for evaluating the dyeing quality of the knitted fabric is the same as that described, for example, in Patent Document 1. Specifically, the image processing device 7 connects two or more uneven regions 14 whose separation distance in the longitudinal direction is less than a first distance and whose separation distance in the transverse direction is less than a second distance, and considers them as one uneven region. Furthermore, the image processing device 7 creates a histogram related to the aspect ratio of each of the connected uneven regions 14, and evaluates the dyeing quality of the knitted fabric based on the frequency of occurrence of each aspect ratio level. However, other methods, different from this, can also be used to evaluate the dyeing quality of the knitted fabric based on the extracted uneven regions 14.

[0048] <Effect> The direction of uneven dyeing extension is determined based on the direction of the knitted fabric's stitches, i.e., the direction in which the multiple loops N are arranged. Furthermore, due to factors such as stitch deviations, the direction of the loop N arrangement may sometimes deviate according to the area of ​​the knitted fabric. In the first embodiment, unevenness candidate region information is obtained based on brightness information; this unevenness candidate region information is information about the unevenness candidate region 13 that will become the unevenness region 14. Additionally, information about the positions of the multiple loops N is obtained as stitch information, i.e., stitch information, in the captured image. Then, the unevenness region 14 is extracted based on the unevenness candidate region information and the stitch information. Therefore, even if the direction of the multiple loops N arrangement deviates due to factors such as stitch deviations, the unevenness region 14 can be extracted with high accuracy.

[0049] Furthermore, in the first embodiment, multiple highly similar regions in the captured image that have a high similarity to the sample image P of the stem N are extracted. Then, based on brightness information, information about the position of the representative point M of the stem N is obtained for each highly similar region. Then, based on the information about the position of the representative point M related to the multiple stems N, information about the position of the stem N is obtained. Thus, information about the position of the multiple stems N can be obtained with high accuracy.

[0050] Furthermore, in the first embodiment, edges contained in the captured image are detected, and high-contrast regions 12 with an edge gradient of a first predetermined value or higher are extracted. High-contrast regions 12 with an aspect ratio (length-to-width ratio) of a horizontal direction relative to a vertical direction of a second predetermined value or higher, and a vertical length within a predetermined range, are extracted as unevenness candidate regions 13. Then, continuous regions within the unevenness candidate regions 13 that overlap with a predetermined number or more of the loop stems N arranged continuously along the horizontal direction are extracted as unevenness regions 14. Thus, high-contrast regions 12 with a high aspect ratio and short vertical length, i.e., high-contrast regions 12 with uneven coloring characteristics such as elongation in the horizontal direction, are extracted as unevenness candidate regions 13. Then, continuous regions within the extracted unevenness candidate regions 13 arranged in the direction of the loop stems N are extracted as unevenness regions 14. Therefore, unevenness regions 14 can be extracted with high precision.

[0051] [Second Embodiment] Next, refer to the appendix Figure 1 A suitable second embodiment of the present invention will now be described. The second embodiment involves the same experimental apparatus 1 as the first embodiment. In the second embodiment, the image processing apparatus 7, according to... Figure 7 The flowchart shown is used to process and extract uneven regions, and the dyeing quality of the knitted fabric is evaluated based on the extracted uneven regions.

[0052] In a detailed explanation, the image processing device 7 performs the same grayscale processing as step S101 of the first embodiment (step S201), and then performs the same brightness information acquisition processing as step S102 of the first embodiment (step S202). However, in the second embodiment, it is also possible to proceed with the processing after step S202 without performing grayscale processing while maintaining the color image.

[0053] Next, the image processing apparatus 7 performs in parallel the same stitch removal process (step S203) and smoothing process (step S204) as steps S103 and S104 of the first embodiment, and the same high similarity region extraction process (step S205), representative point acquisition process (step S206), and loop position extraction process (step S207) as steps S108 to S110 of the first embodiment. Additionally, as described in the first embodiment, the smoothing process in step S204 may be omitted.

[0054] In addition, in the second embodiment, unlike the first embodiment, in the loop position extraction process of step 207, information on the positions of multiple loop regions R is obtained, and information on the positions of the thread regions V in each loop region R is obtained.

[0055] It should be noted that, in the second embodiment, the processing of steps S205 to S207 together is equivalent to the stitch information acquisition processing of the present invention. Furthermore, in the second embodiment, the information regarding the position of the loop region R and the position of the thread region V within each loop region R is the information regarding the position of the loop N (the stitch information of the present invention).

[0056] After the processing in steps S203 and S204 and steps S205 to S207 is completed, the image processing device 7 then performs the uneven candidate region information acquisition process (step S208). In the uneven candidate region information acquisition process, the image processing device 7 calculates the average value of the brightness of the thread region V for each of the multiple loop regions R, i.e., the average brightness, and extracts the loop regions R whose average brightness is less than the threshold TH as uneven candidate regions, thereby obtaining the location information of the uneven candidate regions.

[0057] As an example, regarding such Figure 8 As shown in (a), the 33 loop regions R with loop numbers (1) to (33) are labeled. Figure 8 (b) shows the average brightness of the thread region V. In this example, the loop region R with loop numbers marked (16), (17), (19), (21), (22), and (23) is extracted as the unevenness candidate region. It should be noted that in the second embodiment, the average brightness of the thread region V is the representative brightness of the present invention. In addition, the condition that "the average brightness of the thread region V is less than the threshold TH" is a predetermined condition of the present invention.

[0058] Next, the image processing apparatus 7 performs uneven region extraction processing (step S209). In the uneven region extraction processing, the image processing apparatus 7 extracts a region as an uneven region, that is, the region spans a predetermined number or more uneven candidate regions arranged continuously along the horizontal direction.

[0059] For example, in Figure 8 In examples (a) and (b), when the predetermined number is two, the regions spanning the two stem regions R labeled with stem numbers (16) and (17), and the regions spanning the three stem regions R labeled with stem numbers (21) to (23), are extracted as uneven regions. Additionally, in Figure 8 In examples (a) and (b), when the predetermined number is three, the region spanning the three loop regions R marked with loop numbers (21) to (23) is extracted as an uneven region.

[0060] Next, the image processing device 7, similar to step S111 of the first embodiment, evaluates the dyeing quality of the knitted fabric based on the extracted uneven regions (step 210).

[0061] <Effect> In the second embodiment, uneven regions are extracted based on uneven candidate region information and the position information of multiple loops N as stitch information. Thus, even if the direction of the multiple loops N is deviated due to the influence of stitch deviation, the uneven regions can be extracted with high accuracy.

[0062] Furthermore, in the second embodiment, multiple highly similar regions are extracted from the captured image. For each highly similar region, information about the position of a representative point M of the loop stem N is obtained. Based on the information about the positions of the representative points M related to the multiple loop stems N, information about the position of the loop stem N is obtained. Thus, information about the position of the loop stem N can be obtained with high precision.

[0063] Furthermore, in the second embodiment, the average brightness of the thread regions V in each of the multiple loop regions R is calculated, and loop regions R whose average brightness of the thread regions V is less than a threshold TH are extracted as unevenness candidate regions. Then, regions that span a predetermined number or more unevenness candidate regions arranged continuously along the horizontal direction are extracted as unevenness regions. This allows for high-precision extraction of unevenness regions.

[0064] [Variation Example] The above describes suitable embodiments of the present invention, but the present invention is not limited to the above embodiments and various modifications can be made.

[0065] In the first embodiment, a high-contrast region 12 with an aspect ratio of 2 or higher and a length in the longitudinal direction within a predetermined range is extracted as an unevenness candidate region 13, but this is not a limitation. In the first embodiment, a high-contrast region 12 with an aspect ratio of 2 or higher may also be extracted as an unevenness candidate region 13 regardless of the length in the longitudinal direction.

[0066] Furthermore, in the second embodiment, the average brightness of the thread regions V of the multiple loop regions R is calculated as a representative value of the brightness of the thread regions V. Loop regions R whose average brightness of the thread regions V is less than a threshold TH are extracted as candidate regions for unevenness, but this is not limited to this. For example, in the second embodiment, other values ​​such as the median, maximum, and minimum brightness of the thread regions V can also be calculated as representative values ​​of the brightness of the thread regions V. Alternatively, loop regions R that satisfy other predetermined conditions different from the condition that the representative brightness of the thread regions V is less than the threshold TH can be extracted as candidate regions for unevenness.

[0067] Furthermore, in the first and second embodiments, multiple highly similar regions are extracted from the captured image, and information about the position of a representative point M of the stem N is obtained for each highly similar region. The position information of the stem N is obtained based on this information about the position of the representative point M, but the method is not limited to this. For example, if highly similar regions can be extracted as regions that accurately indicate the position of the stem N, the position information of the highly similar regions can be obtained as the position information of the stem N. Alternatively, the position information of multiple stems N can also be obtained using other methods.

[0068] Furthermore, the method for extracting the uneven region 14 based on the uneven candidate region and stitch information is not limited to the methods described in the first and second embodiments. The uneven region 14 can also be extracted based on the uneven candidate region information and stitch information using other methods different from those described in the first and second embodiments.

[0069] In this case, the uneven candidate region information can be the same as the uneven candidate region information described in the first and second embodiments, or it can be information about the location of other uneven candidate regions that are different from the uneven candidate region information described in the first and second embodiments. Furthermore, regardless of whether the uneven candidate region information is the same as the uneven candidate region information described in the first and second embodiments, the uneven candidate region information can be obtained using methods different from those described above.

[0070] Furthermore, in this case, the stitch information can be information about the positions of multiple loops N that are the same as the stitch information described in the first and second embodiments, or it can be information about the positions of other stitches that are different from them. Additionally, regardless of whether the stitch information is information about the positions of multiple loops N that are the same as the stitch information described in the first and second embodiments, the stitch information can be obtained using methods different from those described above.

[0071] Alternatively, the dyeing quality of the knitted fabric can be evaluated based on the extracted uneven region 14 using methods different from the first and second embodiments. Furthermore, in the first and second embodiments, the image processing device 7 evaluates the dyeing quality of the knitted fabric based on the extracted uneven region 14, but is not limited to this. For example, the image processing device 7 could output histogram data related to the aspect ratio of the extracted uneven region 14, and the operator could evaluate the dyeing quality of the knitted fabric based on the output data. Explanation of reference numerals in the attached figures

[0072] 7: Image processing device 9: Storage Department 12: High contrast area 13: Uneven candidate regions 14: Uneven regions M: Representation point N: Circle stem P: Sample image R: Enclosed area S: Knitted fabric sample V: Silk thread area.

Claims

1. An image processing apparatus for extracting uneven regions corresponding to uneven dyeing from an image obtained by photographing a dyed knitted fabric, characterized in that, Perform the following processing: Brightness information acquisition and processing: Brightness information is obtained, which is the brightness information at various locations in the captured image; The uneven candidate region information acquisition process obtains uneven candidate region information based on the brightness information. This uneven candidate region information is the information on the location of the uneven candidate region that becomes a candidate for the uneven region. The stitch information acquisition and processing involves obtaining stitch information based on the brightness information. This stitch information refers to the location of the stitches in the captured image. Uneven region extraction processing is performed to extract the uneven region based on the uneven candidate region information and the stitch information.

2. The image processing apparatus according to claim 1, characterized in that, In the stitch information acquisition process, information on the positions of multiple loops constituting the stitch is obtained as the stitch information.

3. The image processing apparatus according to claim 2, characterized in that, The system includes a storage unit for storing sample image information, which is information about the sample image of the loop. In the process of acquiring and processing the stitch information, Based on the brightness information and the sample image information, multiple highly similar regions in the captured image that have a high similarity to the sample image are extracted. Using the location information of the multiple highly similar regions, the location information of the multiple loop stems is obtained.

4. The image processing apparatus according to claim 3, characterized in that, In the process of acquiring and processing the stitch information, Based on the brightness information, for each of the plurality of highly similar regions, the location information of the representative point of the circle stem is obtained. Based on the location information of the representative points associated with the multiple highly similar regions, the location information of the multiple loop stems is obtained.

5. The image processing apparatus according to any one of claims 2 to 4, characterized in that, In the process of obtaining information about uneven candidate regions, Based on the brightness information, edges contained in the captured image are detected. Extract high-contrast regions where the gradient of the edge is above a predetermined value. Regions within the high-contrast region whose aspect ratio (length to width) is greater than or equal to a second predetermined value are selected as candidate regions for unevenness. In the process of extracting the uneven region, The continuous region that overlaps with a predetermined number or more of the loops arranged continuously along the horizontal direction in the candidate uneven region is extracted as the uneven region.

6. The image processing apparatus according to claim 5, characterized in that, In the process of obtaining information about uneven candidate regions, The region in the high-contrast region that has an aspect ratio greater than or equal to the second predetermined value and a longitudinal dimension within a predetermined range is extracted as the unevenness candidate region.

7. The image processing apparatus according to any one of claims 2 to 4, characterized in that, In the process of acquiring and processing the stitch information, As information regarding the positions of the plurality of loop stems, information is obtained regarding the positions of multiple loop stem regions in the captured image, each containing one of the loop stems and arranged in both the longitudinal and transverse directions. In the process of obtaining information about uneven candidate regions, Based on the brightness information and the stitch information, for each of the multiple loop regions, a representative value of the brightness of the thread region where the thread is located is calculated, i.e., the representative brightness. Loop regions whose representative brightness meets predetermined conditions are then extracted as candidate regions for unevenness. In the process of extracting the uneven region, The region is extracted as the uneven region, that is, the region spans a predetermined number or more of the uneven candidate regions arranged continuously along the horizontal direction.

Images