Image processing apparatus and method for evaluating the dyeing quality of knitted fabrics

Abstract

To enable securing consistency with evaluations by visual inspection of inspectors when evaluating dyeing quality of knit fabrics by performing analytical processing to images in which dyed knit fabrics are photographed.SOLUTION: Provided is an image processing device 7 for extracting unevenness areas corresponding to dyeing unevenness from an image in which a dyed knit fabric S is photographed, executing: edge detection processing for detecting edges included in the image; first extraction processing for extracting high contrast areas where gradients of the edges are equal to or higher than a prescribed value; and second extraction processing for, out of the high contrast areas, extracting areas where an aspect ratio representing a ratio of a size in a wale direction to a size of a course direction is equal to or greater than the prescribed value, as the unevenness areas.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an image processing apparatus that extracts a non-uniform region corresponding to dyeing unevenness from an image of a dyed knitted fabric, and a method for evaluating the dyeing quality of a knitted fabric based on the non-uniform region extracted by the image processing apparatus.

Background Art

[0002] When dyeing synthetic fiber yarns made of polyester or the like, dyeing unevenness may occur. Such dyeing unevenness is considered to be caused by variations in the molecular orientation and crystallization during the production or processing of the yarn, resulting in differences in the ease of penetration of the dyeing solution depending on the location. Since dyeing unevenness has a significant impact on the commercial value of the yarn, tests for pre-evaluating the dyeing quality of the yarn have been conventionally conducted. Specifically, a knitted fabric sample obtained by circular knitting the yarn to be evaluated is dyed, and the dyeing quality of the yarn is evaluated based on the dyeing unevenness occurring in this sample.

[0003] Such tests are visually performed by inspectors, but there may be differences in evaluation by inspectors, and even the same inspector may have different evaluations when there is a time gap. Therefore, in order to minimize the variation in evaluation, attempts have been made to mechanically evaluate the dyeing quality by photographing a dyed knitted fabric sample with a camera and analyzing the image. For example, as part of such attempts, in Patent Document 1, a technique for removing the stitch pattern of a knitted fabric sample from an image has been proposed in order to easily detect dyeing unevenness from the photographed image.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] When knitting a fabric, loops are formed by a single thread in the course direction, and the thread is hooked on the loops to form another loop in the wale direction. Since dyeing unevenness often appears continuously in the length direction of the thread, when a fabric sample is dyed, the dyeing unevenness becomes long and thin streak-like unevenness extending in the course direction. When evaluating the dyeing quality based on such dyeing unevenness, the inspector comprehensively and intuitively evaluates it considering various factors such as the length of the unevenness, the number of unevenness, and the density of the unevenness. Therefore, when evaluating the dyeing quality by analyzing the image of the fabric sample, there has been a problem that it is difficult to achieve consistency with the visual evaluation by the inspector.

[0006] In view of the above problems, an object of the present invention is to achieve consistency with the visual evaluation by the inspector when evaluating the dyeing quality of a fabric by analyzing an image obtained by photographing the dyed fabric.

Means for Solving the Problems

[0007] The image processing apparatus according to the present invention is an image processing apparatus that extracts a non-uniformity region corresponding to dyeing non-uniformity from an image obtained by photographing a dyed fabric, and includes an edge detection process for detecting an edge included in the image, a first extraction process for extracting a high-contrast region where the gradient of the edge is equal to or greater than a predetermined value, and a second extraction process for extracting, as the non-uniformity region, a region where an aspect ratio, which is a ratio of the size in the course direction to the size in the wale direction among the high-contrast regions, is equal to or greater than a predetermined value.

[0008] The dye unevenness can be recognized as a part with high contrast in the captured image. However, simply extracting high contrast areas would extract unevenness other than the dye unevenness caused by the variation in the oriented crystals of the yarn (for example, unevenness caused by scratches when knitting the fabric), and the dye quality could not be evaluated correctly. Therefore, in the present invention, following the first extraction process for extracting the high contrast areas, a second extraction process is performed for extracting only those with high aspect ratios from the extracted high contrast areas as uneven areas. By performing the second extraction process, it is possible to accurately extract uneven areas corresponding to the actual dye unevenness that extends in a long and thin manner in the course direction. As a result, it is possible to achieve consistency with the visual evaluation by an inspector.

[0009] In the present invention, in the second extraction process, a region of the high contrast region, in which the aspect ratio is equal to or greater than the predetermined value and in addition, the size in the wale direction is within a predetermined range, may be extracted as the uneven region.

[0010] By adding the size in the wale direction to the extraction conditions for the uneven region, the fineness of the actual dye unevenness can also be taken into consideration, thereby improving the extraction accuracy of the uneven region.

[0011] In the present invention, after the second extraction process is performed, two or more of the uneven regions whose separation distance in the wale direction is equal to or less than a predetermined value and whose separation distance in the course direction is equal to or less than a predetermined value may be connected.

[0012] During the image processing process, a single uneven area may be broken into several parts. By connecting the uneven areas under certain conditions as described above, the divided uneven areas can be joined together.

[0013] In the present invention, before the second extraction process is performed, it is preferable to delete, from the high contrast region, a region whose size in the X-axis direction of the image is equal to or smaller than a predetermined value.

[0014] By doing so, it is possible to remove minute noise generated during image processing.

[0015] In the present invention, before executing the second extraction process, it is preferable to correct the image so that the course direction of the knitted fabric shown in the image is parallel to the X-axis of the image.

[0016] Generally, when photographing a knitted fabric, it is adjusted so that the course direction of the knitted fabric and the camera are horizontal, so the course direction of the knitted fabric shown in the image is parallel to the X-axis of the image. However, due to errors during placement or the like, the course direction may be inclined with respect to the X-axis. Therefore, by performing the above-described correction, such inclination can be eliminated.

[0017] In the present invention, before executing the edge detection process, it is preferable to convert the image to grayscale.

[0018] By converting to grayscale, it is possible to avoid the edges becoming unclear due to the influence of hue in the edge detection process. Also, the computational load of subsequent image processing can be reduced.

[0019] In the present invention, before executing the edge detection process, it is preferable to remove the stitch pattern in the image.

[0020] By doing so, the influence of the stitch pattern can be reduced, and the uneven area can be extracted well.

[0021] In the present invention, before executing the edge detection process, it is preferable to smooth the image.

[0022] By smoothing the image, it is possible to remove the noise included in the photographed image. In particular, if smoothing is executed after removing the stitch pattern, the remnants of the stitch pattern that could not be completely removed can be reduced.

[0023] The method for evaluating the dyeing quality of knitted fabrics according to the present invention is characterized by evaluating the dyeing quality of the knitted fabric based on the uneven area extracted by any of the above image processing apparatuses.

[0024] The uneven area extracted by any of the above image processing apparatuses is extracted in consideration of the characteristics of actual dyeing unevenness. Therefore, if the dyeing quality of the knitted fabric is evaluated based on the extracted uneven area, it is possible to achieve consistency with the conventional visual evaluation by inspectors.

[0025] In the present invention, it is preferable to evaluate the dyeing quality of the knitted fabric based on the width ratio, which is the ratio of the size of the uneven area in the course direction of the knitted fabric to the size of the knitted fabric in the course direction.

[0026] When evaluating the dyeing quality of a knitted fabric, by considering the length of the uneven area with respect to the entire knitted fabric as described above, an evaluation closer to the inspector's perception can be performed.

[0027] Specifically, it is preferable to evaluate the dyeing quality of the knitted fabric based on the average value of the width ratio of the uneven area.

[0028] Alternatively, it is preferable to evaluate the dyeing quality of the knitted fabric based on the standard deviation of the width ratio of the uneven area.

[0029] Alternatively, it is preferable to evaluate the dyeing quality of the knitted fabric based on the number of those in the uneven area whose width ratio is equal to or greater than a predetermined value.

Brief Description of Drawings

[0030]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Mode for Carrying Out the Invention

[0031] (Outline of Test Equipment) Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic diagram of a test facility for evaluating the dyeing quality of a knitted fabric sample. This test facility 1 photographs a knitted fabric sample S (the knitted fabric of the present invention) with a camera 3, and by subjecting the photographed image to image processing by an image processing apparatus 7, an uneven region corresponding to actual dyeing unevenness is extracted from the image, and the dyeing quality of the knitted fabric sample S is evaluated.

[0032] Here, the knitted fabric sample S will be described first. FIG. 2 is an enlarged schematic diagram of a part of the knitted fabric sample S. The knitted fabric sample S is a knitted fabric obtained by performing dyeing on a knitted fabric obtained by circular knitting a synthetic fiber yarn Y made of, for example, polyester. The knitted fabric sample S has a structure in which loops are formed by one yarn Y in the course direction (horizontal direction), and another loop is formed in the wale direction (vertical direction orthogonal to the course direction) by hooking the yarn Y on the loop. Dyeing unevenness in the yarn Y generally occurs continuously in the length direction of the yarn Y. Therefore, the dyeing unevenness that occurs when the knitted fabric sample S is dyed becomes streak-like unevenness extending in the course direction as shown by the thick line in FIG. 2.

[0033] Returning to FIG. 1, the test equipment 1 will be described. The test equipment 1 includes a camera 3, a support stand 4, two illuminations 5, a reflector 6, an image processing device 7, and a monitor 8. The camera 3, the support stand 4, the two illuminations 5, and the reflector 6 are arranged in the darkroom 2. The knitted fabric sample S is supported by a support stand 4 arranged at the center of the darkroom 2 such that the opening of the tubular knitted fabric faces up and down, that is, the course direction is horizontal.

[0034] The camera 3 is horizontally arranged in front of the support stand 4. With such an arrangement, the camera 3 can photograph the front surface of the knitted fabric sample S such that the course direction of the knitted fabric sample S is substantially parallel to the X-axis of the photographed image. By setting the magnification of the camera 3 and the distance between the camera 3 and the knitted fabric sample S to prescribed values and then taking a photograph, the size of the knitted fabric sample S reflected in the image becomes known. In this embodiment, it is set such that the size of the knitted fabric sample S in the X-axis direction in the image is about 2056 pixels.

[0035] The two illuminations 5 are arranged behind the support stand 4 and are arranged symmetrically with respect to the support stand 4 on the left and right. The two illuminations 5 are inclined by about 45 degrees from the straight line connecting the camera 3 and the support stand 4 and are located on the straight line passing through the camera 3. The reflector 6 is arranged near the camera 3 and reflects the light from the two illuminations 5 toward the knitted fabric sample S. Thereby, the front surface of the knitted fabric sample S has an appropriate brightness for photographing.

[0036] The image processing device 7 is composed of a CPU, a memory, a hard disk, etc. The image processing device 7 is electrically connected to the camera 3 and can capture the image data of the knitted fabric sample S photographed by the camera 3. The image processing device 7 extracts a non-uniformity region corresponding to the actual dyeing non-uniformity from the photographed image by executing predetermined image processing on the image of the knitted fabric sample S photographed by the camera 3. Further, the image processing device 7 evaluates the dyeing quality of the knitted fabric sample S based on the extracted non-uniformity region. The monitor 8 is connected to the image processing device 7. The monitor 8 can display the image of the knitted fabric sample S, the evaluation result of the dyeing quality of the knitted fabric sample S, etc.

[0037] (Extraction of uneven areas and evaluation of dyeing quality) Next, the extraction of uneven areas and the evaluation of dyeing quality by the image processing apparatus 7 will be described. FIGS. 3 and 4 are flowcharts showing a series of processes by the image processing apparatus 7. The image processing apparatus 7 captures the image data of the knitted fabric sample S photographed by the camera 3 and grayscales the image (step S11). By performing image processing using the grayscaled image, the arithmetic processing can be speeded up. Also, when performing edge detection processing in step S14, it is possible to avoid the edges becoming unclear due to the influence of hue. However, grayscaling of the image is not essential, and it is also possible to proceed with image processing while maintaining the color image.

[0038] Subsequently, the image processing apparatus 7 removes the knitting pattern from the grayscaled image (step S12). To remove the knitting pattern, for example, the technique described in Japanese Patent Application Laid-Open No. 2015-212924 may be adopted. Specifically, periodic information having a repetition number of aN or more obtained by multiplying the repetition number N (known number) of loops per unit length by a predetermined coefficient a is removed by a low-pass filter. The coefficient a is an arbitrary value of, for example, 0.7 to 0.9, and is set to 0.85 in the present embodiment. However, the method for removing the knitting pattern is not limited to that described here, and other methods may be used.

[0039] Subsequently, the image processing apparatus 7 smooths the image from which the knitting pattern has been removed (step S13). Smoothing is performed to remove minute noise caused by remnants of the knitting pattern that could not be completely removed in step S12. For smoothing, for example, a known Wiener filter can be applied. The filter size is, for example, about 5 to 25 pixels × 10 to 60 pixels, and is set to 12 pixels × 18 pixels or 15 pixels × 40 pixels in the present embodiment. However, the type of filter and the filter size are not limited to these.

[0040] Subsequently, the image processing apparatus 7 executes edge detection processing on the smoothed image (step S14). The edge detection processing is executed to detect locations (edges) in the image where the brightness changes abruptly. For example, a known Prewitt filter can be applied to the edge detection processing. The filter size is, for example, 3 pixels × 3 pixels. However, the type of filter and the filter size are not limited to these. FIG. 5a is an example of an image 10 in which an edge 11 is detected by the edge detection processing. As shown in each figure of FIG. 5, the horizontal direction of the image 10 is defined as the X-axis direction, and the vertical direction is defined as the Y-axis direction.

[0041] Subsequently, the image processing apparatus 7 executes first extraction processing on the image 10 on which the edge detection processing has been executed (step S15). In the first extraction processing, a high-contrast region 12 where the gradient (brightness slope) of the edge 11 is equal to or greater than a predetermined value is extracted. Specifically, binarization processing is executed on the image 10 in which the edge 11 is detected, using the predetermined value as a threshold. Pixels with a gradient of the edge 11 that is equal to or greater than the threshold are set to white, and pixels with a gradient of the edge 11 that is less than the threshold are set to black. FIG. 5b is an image in which the high-contrast region 12 is extracted in white in the image 10 shown in FIG. 5a.

[0042] The method of determining the predetermined value in the first extraction processing will be described with reference to FIG. 6. First, as shown in FIG. 6a, for the image 10 on which the edge detection processing has been executed, a histogram is created with the gradient of the edge 11 on the horizontal axis and the number of pixels on the vertical axis. Then, the upper limit value U of the gradient of the edge 11 is obtained from this histogram, and bU obtained by multiplying the upper limit value U by a predetermined coefficient b is set as the predetermined value, that is, the threshold for the binarization processing. The coefficient b is an arbitrary value, for example, in the range of 0.15 to 0.5, and is set to 0.25 in this embodiment.

[0043] When the tail of the histogram extends long to the right side, the upper limit value U may be approximately obtained as follows. Figure 6(b) is an enlarged view of the vicinity of the tail in Figure 6(a). As shown by the dashed line in Figure 6(b), a linear approximation formula in a certain section of the tail of the histogram is obtained, and the angle θ formed by the straight line of this linear approximation formula and the horizontal axis of the histogram is obtained. Then, when this angle θ becomes a predetermined angle, the intersection of the straight line of the linear approximation formula and the horizontal axis may be used as the upper limit value U. The above-mentioned predetermined angle is, for example, an arbitrary angle of 5 to 30 degrees, and in this embodiment, it is set to 25 degrees. However, the method of approximately obtaining the upper limit value U is not limited to this.

[0044] Returning to Figure 3, the continuation of the flowchart will be described. When the first extraction process is completed, the image processing apparatus 7 removes noise by deleting regions in the high-contrast region 12 extracted in the first extraction process where the size in the X-axis direction is equal to or less than a predetermined value (step S16). The above-mentioned predetermined value is, for example, 100 pixels, but it can be changed as appropriate. By such processing, minute noise generated in the binarization process can be removed. Figure 5(c) is an image obtained by removing minute noise from the image 10 shown in Figure 5(b).

[0045] Subsequently, when the course direction of the knitting sample S shown in the image 10 is inclined with respect to the X-axis of the image 10, the image processing apparatus 7 corrects the inclination of the image 10 so that the course direction becomes parallel to the X-axis (step S17). Specifically, the inclinations of several high-contrast regions 12 may be calculated by the least squares method, and the image 10 may be rotated according to the average value of the calculated inclinations. As described above, the knitting sample S is photographed so that the course direction is substantially parallel to the X-axis, but there may be cases where the course direction is inclined with respect to the X-axis due to errors, so such correction is being performed. Figure 5(d) is an image obtained by correcting the inclination of the image 10 shown in Figure 5(c).

[0046] Next, the image processing device 7 executes a second extraction process on the tilt-corrected image 10 (step S18). In the second extraction process, a region of the high contrast region 12 that has the characteristic of actual dyeing unevenness, that is, that it is elongated in the course direction, is extracted as an uneven region 13. Specifically, a high contrast region 12 that satisfies both a first condition that "the aspect ratio, which is the ratio of the size in the course direction to the size in the wale direction, is equal to or greater than a predetermined value" and a second condition that "the size in the wale direction is within a predetermined range" is extracted as an uneven region 13.

[0047] In this embodiment, the inclination of the image 10 is corrected in step S17 so that the course direction is parallel to the X axis, so that the aspect ratio of the first condition can be calculated by (size in the X axis direction) / (size in the Y axis direction). The above-mentioned predetermined value of the aspect ratio can be any value, for example, 4 or more, and is set to 7 in this embodiment. The size in the wale direction of the second condition coincides with the size in the Y axis direction. The above-mentioned predetermined range of the size in the wale direction is, for example, any range between 10 and 60 pixels, and is set to 17 to 47 pixels in this embodiment. The specific values of the first and second conditions can be changed as appropriate. By extracting the uneven area 13 from the high contrast area 12 using the first and second conditions, it is possible to accurately extract elongated streak-like unevenness in the course direction corresponding to the actual dye unevenness.

[0048] Fig. 7 shows images in which high contrast regions 12 have been extracted from the photographed images of three knitted samples S (samples S1 to S3) and inclination correction has been performed. Fig. 8 shows images in which uneven regions 13 have been extracted from the photographed images of samples S1 to S3, that is, images in which the second extraction process has been performed on each image 10 in Fig. 7. As can be seen by comparing Fig. 7 with Fig. 8, by performing the second extraction process, short or thick high contrast regions 12 that do not match the characteristics of actual dyeing unevenness were eliminated, and elongated high contrast regions 12 that extend linearly in the course direction (X-axis direction) were extracted as uneven regions 13.

[0049] Subsequently, the image processing apparatus 7 connects two or more uneven areas 13 where the separation distance in the wale direction is equal to or less than a predetermined value and the separation distance in the course direction is equal to or less than a predetermined value (step S19). Specifically, regarding the separation distance in the wale direction, it is determined whether the absolute value of the difference in the Y coordinates of the centroids of two uneven areas 13 is, for example, equal to or less than an arbitrary value of 5 to 25 pixels (10 pixels in this embodiment). Regarding the separation distance in the course direction, it is determined whether the absolute value of the difference in the X coordinates of the adjacent ends of two uneven areas 13 is, for example, equal to or less than an arbitrary value of 3 to 20% of the size of the knitting sample S in the X-axis direction (100 pixels in this embodiment).

[0050] Note that whether to connect two or more uneven areas 13 into one uneven area 13 is only determined by calculation, and an image with the uneven areas 13 actually connected is not created. In the case of this embodiment, in the areas circled in FIG. 9, the adjacent uneven areas 13 are regarded as being connected into one. In the evaluation of the dyeing quality of samples S1 to S3 described below, two or more connected uneven areas 13 are treated as one long uneven area 13.

[0051] Next, in order to evaluate the dyeing quality of samples S1 to S3, the image processing apparatus 7 creates a histogram regarding the width ratio of the extracted uneven areas 13 (step S20). FIG. 10 is a histogram showing the relationship between the width ratio of the uneven areas 13 and the number of areas. As shown in FIG. 9(c), the width ratio means the ratio L2 / L1 of the size L2 in the course direction of the uneven area 13 to the size L1 in the course direction of the knitting sample S shown in image 10. Regarding the uneven areas 13 connected in step S19, the size of the uneven area 13 after connection is set as L2. The image processing apparatus 7 evaluates the dyeing quality of samples S1 to S3 based on the appearance frequency for each class of the width ratio (step S21). Note that the image processing apparatus 7 does not actually need to create a graph as shown in FIG. 10, and it is sufficient to store information regarding the width ratio in the memory.

[0052] FIG. 11 is a table comparing each numerical value related to the visual evaluation score and the width ratio of the uneven area 13. When the dyeing quality of samples S1 to S3 was visually evaluated by an inspector on a scale of 0 to 4.5 points (the higher the score, the higher the quality), the score for sample S1 was 4.0 points, the score for sample S2 was 3.75 points, and the score for sample S3 was 3.5 points. On the other hand, regarding the uneven area 13 extracted by the image processing device 7 from the photographed images of samples S1 to S3, the average value of the width ratio, the standard deviation of the width ratio, and the number of areas with a width ratio of 0.25 or more were determined. Note that the threshold value of 0.25 can be changed as appropriate.

[0053] A larger average value of the width ratio means that the length of the extracted uneven area 13 in the course direction is, on average, longer. Therefore, it is reasonable to think that the lower the visual evaluation score, the larger the average value of the width ratio, and indeed such results were obtained. Next, a larger standard deviation of the width ratio means that the distribution of the width ratio spreads to larger values, indicating the presence of uneven areas 13 with a large width ratio. Therefore, it is reasonable to think that the lower the visual evaluation score, the larger the standard deviation of the width ratio, and indeed such results were obtained. Finally, a larger number of areas with a width ratio of 0.25 or more means that there are many uneven areas 13 that extend long in the course direction. Therefore, it is reasonable to think that the lower the visual evaluation score, the larger the number of areas with a width ratio of 0.25 or more, and indeed such results were obtained.

[0054] As described above, when the average value of the width ratio, the standard deviation of the width ratio, and the number of areas with a width ratio of 0.25 or more of the extracted uneven area 13 were used as evaluation items, good correspondence with the visual evaluation score was obtained in all cases. Therefore, if the dyeing quality of the knitted fabric sample S is evaluated by combining any one or two or more of these evaluation items, consistency with the visual evaluation score can be achieved. The image processing device 7, for example, evaluates the dyeing quality of the knitted fabric sample S by combining any one or two or more of the above evaluation items and displays the evaluation result on the monitor 8 to end a series of processes.

[0055] (Effect) The image processing device 7 according to the present embodiment executes an edge detection process for detecting an edge 11 included in an image 10 obtained by photographing a knitted fabric sample S, a first extraction process for extracting a high contrast region 12 in which the gradient of the edge 11 is equal to or greater than a predetermined value, and a second extraction process for extracting a region in the high contrast region 12 in which the aspect ratio is equal to or greater than a predetermined value as an uneven region 13. Dye unevenness can be recognized as a part with high contrast in the photographed image. However, simply extracting the high contrast region 12 would extract unevenness other than dye unevenness caused by variations in the oriented crystals of the yarn Y (for example, unevenness caused by scratches when knitting the knitted fabric sample S), making it impossible to correctly evaluate the dyeing quality. Therefore, following the first extraction process for extracting the high contrast region 12, a second extraction process is executed for extracting only those with a high aspect ratio from the extracted high contrast region 12 as an uneven region 13. By executing the second extraction process, it is possible to accurately extract the uneven region 13 corresponding to the actual dye unevenness extending in an elongated manner in the course direction. As a result, it became possible to achieve consistency with the visual evaluations made by inspectors.

[0056] In this embodiment, in the second extraction process, among the high contrast regions 12, regions that have an aspect ratio equal to or greater than a predetermined value and also have a size in the wale direction within a predetermined range are extracted as the uneven regions 13. By adding the size in the wale direction to the extraction conditions for the uneven regions 13, the fineness of the actual dyeing unevenness can also be taken into consideration, and therefore the extraction accuracy of the uneven regions 13 can be improved.

[0057] In this embodiment, after the second extraction process is performed, two or more uneven regions 13 whose separation distance in the wale direction is equal to or less than a predetermined value and whose separation distance in the course direction is equal to or less than a predetermined value are connected. During the image processing process, one uneven region 13 may be disconnected and divided into multiple regions. Therefore, by connecting the uneven regions 13 under the predetermined conditions as described above, the divided uneven regions 13 can be joined.

[0058] In this embodiment, before executing the second extraction process, an area of the high-contrast area 12 where the size of the image 10 in the X-axis direction is equal to or less than a predetermined value is deleted. By doing so, minute noise generated during image processing can be removed.

[0059] In this embodiment, before executing the second extraction process, the image 10 is corrected so that the course direction of the knitted fabric sample S shown in the image 10 is parallel to the X-axis of the image 10. Generally, when photographing the knitted fabric sample S, the course direction of the knitted fabric sample S and the camera 3 are adjusted to be horizontal, so the course direction of the knitted fabric sample S shown in the image 10 and the X-axis of the image 10 are parallel. However, due to errors during placement or the like, the course direction may be inclined with respect to the X-axis. Therefore, by performing the above-described correction, such inclination can be eliminated.

[0060] In this embodiment, before executing the edge detection process, the image 10 is grayscale-converted. By grayscale-converting, it is possible to avoid the edges becoming unclear due to the influence of hue in the edge detection process. Also, the computational load of subsequent image processing can be reduced.

[0061] In this embodiment, before executing the edge detection process, the stitch pattern in the image 10 is removed. By doing so, the influence of the stitch pattern is reduced, and the uneven area 13 can be extracted well.

[0062] In this embodiment, before executing the edge detection process, the image 10 is smoothed. By smoothing the image 10, the noise included in the photographed image 10 can be removed. In particular, if smoothing is executed after removing the stitch pattern, the remains of the stitch pattern that could not be completely removed can be reduced.

[0063] In this embodiment, the dyeing quality of the knitted fabric sample S is evaluated based on the uneven area 13 extracted by the image processing apparatus 7. The uneven area 13 extracted by the image processing apparatus 7 is extracted in consideration of the characteristics of actual dyeing unevenness. Therefore, if the dyeing quality of the knitted fabric sample S is evaluated based on the extracted uneven area 13, it is possible to achieve consistency with the evaluation by the visual inspection of conventional inspectors.

[0064] In this embodiment, the dyeing quality of the knitted fabric sample S is evaluated based on the width ratio L2 / L1, which is the ratio of the size L2 in the course direction of the uneven area 13 to the size L1 in the course direction of the knitted fabric sample S. When evaluating the dyeing quality of the knitted fabric sample S, by considering the length of the uneven area 13 with respect to the whole of the knitted fabric sample S as described above, an evaluation closer to the sense of the inspector can be performed.

[0065] (Other embodiments) A modified example in which various modifications are made to the above embodiment will be described.

[0066] In the above embodiment, as a method for evaluating the dyeing quality of the knitted fabric sample S, the aspect ratio of the uneven area extracted by the image processing apparatus 7 is obtained, and at least one of the average value of the width ratio, the standard deviation of the width ratio, and the number of uneven areas with a width ratio of 0.25 or more is adopted as an evaluation item. However, the evaluation items are not limited to these, and indicators other than the width ratio may be used.

[0067] In the above embodiment, in the second extraction process, the first condition regarding the aspect ratio and the second condition regarding the size in the wale direction are imposed. However, if the uneven area corresponding to the actual dyeing unevenness can be extracted well only by the aspect ratio, it is also possible to eliminate the second condition. Alternatively, conditions other than the size in the wale direction can be combined with the aspect ratio condition.

[0068] In the above-described embodiment, the image processing of steps S11 to S19 is performed on the captured image of the knitted fabric sample S. Among these, the processes essential to the present invention are the edge detection process (step S14), the first extraction process (step S15), and the second extraction process (step S18). Regarding the other processes, it is also possible to omit them according to the conditions or substitute them with other processes.

[0069] In the above-described embodiment, after the image processing apparatus 7 performs a series of image processing, it is assumed that the evaluation of the dyeing quality in step S21 is automatically performed. However, the final evaluation of the dyeing quality may be performed by an inspector. For example, the image processing apparatus 7 may finish the process by displaying the histogram shown in FIG. 10 on the monitor 8, and the final evaluation of the dyeing quality may be performed by the inspector based on the histogram.

Explanation of Reference Numerals

[0070] 1: Image processing apparatus 10: Image 11: Edge 12: High-contrast region 13: Uneven region S (S1 to S3): Knitted fabric sample (knitted fabric)

Claims

1. An image processing apparatus that extracts a non-uniformity region corresponding to dyeing non-uniformity from an image of a knitted fabric taken such that the course direction of the dyed knitted fabric is substantially parallel to the X-axis of the captured image, comprising: an edge detection process for detecting edges included in the image; a first extraction process for extracting a high contrast region where the gradient of the edge is equal to or greater than a first predetermined value; a second extraction process for extracting, as the non-uniformity region, a region where an aspect ratio, which is a ratio of the size in the course direction to the size in the wale direction of the knitted fabric, among the high contrast regions is equal to or greater than a second predetermined value; executing; after execution of the second extraction process, connecting two or more of the non-uniformity regions among the plurality of non-uniformity regions where the distance between the centroids in the wale direction is equal to or less than a third predetermined value and the distance between the ends close to each other in the course direction in the course direction is equal to or less than a fourth predetermined value. An image processing apparatus characterized by the above.

2. An image processing apparatus that extracts a non-uniformity region corresponding to dyeing non-uniformity from an image of a knitted fabric taken such that the course direction of the dyed knitted fabric is substantially parallel to the X-axis of the captured image, comprising: an edge detection process for detecting edges included in the image; a first extraction process for extracting a high contrast region where the gradient of the edge is equal to or greater than a first predetermined value; a second extraction process for extracting, as the non-uniformity region, a region where an aspect ratio, which is a ratio of the size in the course direction to the size in the wale direction of the knitted fabric, among the high contrast regions is equal to or greater than a second predetermined value; executing; before execution of the second extraction process and after execution of the first extraction process, correcting the image such that the course direction of the knitted fabric shown in the image is parallel to the X-axis of the image. An image processing apparatus characterized by the above.

3. The image processing apparatus according to claim 2, characterized in that after execution of the second extraction process, two or more of the non-uniformity regions among the plurality of non-uniformity regions where the distance between the centroids in the wale direction is equal to or less than a third predetermined value and the distance between the ends close to each other in the course direction in the course direction is equal to or less than a fourth predetermined value are connected.

4. The image processing apparatus according to any one of claims 1 to 3, characterized in that in the second extraction process, in addition to the aspect ratio of the high contrast region being equal to or greater than the second predetermined value, a region where the size in the wale direction is within a predetermined range is extracted as the non-uniformity region.

5. The image processing apparatus according to any one of claims 1 to 4, characterized in that, after the execution of the first extraction process and before the execution of the second extraction process, an area of the high-contrast area where the size of the image in the X-axis direction is equal to or less than a fifth predetermined value is deleted.

6. The image processing apparatus according to any one of claims 1 to 5, characterized in that the image is converted to grayscale before the execution of the edge detection process.

7. The image processing apparatus according to any one of claims 1 to 6, characterized in that moire patterns in the image are removed before the execution of the edge detection process.

8. The image processing apparatus according to any one of claims 1 to 7, characterized in that the image is smoothed before the execution of the edge detection process.

9. A method for evaluating the dyeing quality of a knitted fabric, characterized in that the dyeing quality of the knitted fabric is evaluated based on the width ratio, which is the ratio of the size of the uneven area extracted by the image processing apparatus according to any one of claims 1 to 8 in the course direction of the knitted fabric to the size of the knitted fabric in the course direction.

10. The method for evaluating the dyeing quality of a knitted fabric according to claim 9, characterized in that the dyeing quality of the knitted fabric is evaluated based on the average value of the width ratios of the uneven areas.

11. The method for evaluating the dyeing quality of a knitted fabric according to claim 9 or 10, characterized in that the dyeing quality of the knitted fabric is evaluated based on the standard deviation of the width ratios of the uneven areas.

12. The method for evaluating the dyeing quality of a knitted fabric according to any one of claims 9 to 11, characterized in that the dyeing quality of the knitted fabric is evaluated based on the number of those having a width ratio equal to or greater than a sixth predetermined value among the uneven areas.

Images