JPH06500395A - inspection of clothing - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] inspection of clothing The present invention relates to the inspection of garments, particularly Y Concerning the inspection of garments such as underwear with glyph fronts.

Inspecting mass-produced clothing such as socks and underwear is a particularly labor-intensive process. It is a business. Currently, garments are inspected for dimensional defects by eye. , inspected for correct sized labels or defects in workmanship. , such as bad stitching or missing seams, defects on the label, discoloration, holes, smudges, etc. Inspected for fabric defects, print defects, color defects, etc. like this The number of workers engaged in inspection may significantly exceed the number of workers engaged in tailoring. There are quite a lot of them.

It is clear that automatic mechanical inspection is desirable. However, it is not so expensive. High detection efficiency and rejection of defect-free garments using non-defective equipment automatically generate images quickly enough to keep pace with the production process while keeping the rate of It is difficult to develop a method for forming and inspecting these materials.

forming an image from the garment; identifying seams on the image; a step of inspecting the panel areas individually; It includes a method consisting of:

The present invention is particularly useful for garments in which the seam itself has a width and the seam can also be inspected. It can be applied to

Seams that are raised higher than adjacent panels, more specifically x, y on the surface of the garment A seam with a Z coordinate for Therefore, it can be detected.

The method of the invention has a waistband structure with wide, raised seams. Applicable for inspecting Y-shaped fronted undergarments containing edged panels and laid flat. a step of forming an image of the front of the garment, and a step of placing an outline of the garment on the image. Image the edges of the waistband structure with wide, raised seams. The step of placing Waistband with over-image panels and wide, raised seams and inspecting the code structure for defects.

A garment can be measured by taking measurements from its front image, and this image It is irradiated with front lighting to form a wide area, and then irradiated with diagonal lighting to create a wide area. Position the edge of the waistband structure with the raised seams and place the image in front. It can be superimposed on an image formed with surface illumination.

The invention also automatically inspects garments with panels separated by seams. A device for an image forming means for forming an image of the garment; seam identification means for identifying a seam on the image formed by the image forming means; and panel inspection means for individually inspecting panel areas on the image. do.

The image forming means is an area scanning video camera aimed at a background on which the garment can lie flat. and frontal and oblique lighting that illuminates clothing placed in the background.

The seam identification means identifies highlights and shadows on the obliquely illuminated image; Image processing means may be included to superimpose the image onto the front-lit image of the garment.

The panel inspection means identifies one panel on the image and processes it to make a predetermined decision. Detect defects according to standards and create warning devices or garment rejection devices when defects are detected. It may include image processing means adapted to be moved.

The panel inspection means may be configured to sequentially process each of a plurality of panels of images. Ru. Defect recording means can be provided to record each of multiple panels during batch inspection of similar garments. It is possible to record the occurrence of defects associated with each defect. This equipment is also capable of batch inspection processing Arrange multiple panels in an inspection order according to the defect rate for each panel recorded as Heuristic controls may be included.

An embodiment of the apparatus and method for automatically inspecting clothes according to the present invention is attached below. This will be explained with reference to the drawings. In the drawings: FIG. 1 is a schematic diagram of the apparatus.

FIG. 2 is a first image produced by the apparatus of FIG.

FIG. 3 shows the processing steps in assessing garment dimensions from the first image.

FIG. 4 is a second image produced by the device.

FIG. 5 shows the processing steps in locating the garment seam from the second image.

FIG. 6 shows the step of superimposing the seam located from the second image onto the first image.

FIG. 7 shows the stages of the defect detection operation.

The drawing is laid flat on a conveyor 12 passing under an area scanning video camera 13. Automatic garment inspection device adapted to inspect Y-shaped front underwear 11 It shows. This camera has 5 images of each garment to be processed by the image processing device 14. form.

Y-shaped front underwear consists of panels lla with seams llb sandwiched between them. , has special problems with automatic mechanical inspection. Treat the garment as a single panel To do this, we need to somehow ignore the signal caused by the seam. the image Fitting to a standard template is important for garments of uniform size or shape. It does not take into account the fact that no tiling occurs.

The drawings show how seam llb is identified and panel area 11a is inspected individually. It shows.

In underwear with a Y-shaped front, the seam llb itself has a width, but this method It is also possible to test for mullb.

If the seam llb has a different color or a different texture than the panel lla If it is certain, this difference can be used to identify the seam. but However, generally speaking, the seam is the same or very similar in appearance to the panel.

Accordingly, in the preferred method of the invention, the seams of the Y-fronted underwear are The fact that it is raised from the panel, that is, the x, y coordinates of the clothing surface Use the fact that it has a Z coordinate.

As can be seen, the garment 14 has a front side and a back side, both of which need to be inspected. This does not necessarily mean that the preliminary inspection of textiles will reduce the single defect occurrence rate to a negligible level. This is not to say that the defect you are trying to detect is a tailoring defect.

Such defects occur almost always on the complex front side of the garment, and on the fairly simple back side. It is known that this rarely occurs. While doing so, check the front and back of your clothes. If both sides must be inspected, turn the garment over as necessary. You will have to make changes and repeat the process. This specification describes front inspection. However, in the case of underwear with a Y-shaped front, place the garment with the front side up and Do you want to expose the open seam across the front, but inspecting the other side is unnecessary? I don't know.

First, an image of the garment 1 is formed using the front illumination setup 15. This image ( May be averaged over two or more frames for noise reduction purposes. ) in the frame memory of device 14 and using an appropriate gray scale threshold. segment. Therefore, the image has a bright background as can be seen in Figure 2. In contrast, it becomes uniformly dark. Of course, other methods may be used as desired.

Next, outline this image using the Robot Edge Follower program. and then decide. In this case, any robot moves across the image until it encounters a dark vixel. scans the pixels of a set group, then moves and rotates the group to configure its configuration. It maintains both light and dark pixels and plots the edges of the image. Position of each step is saved, and at the end of the edge plotting process, a collection of image pixels (X %y+) constitutes the edge.

The centroid (X, Y) of the image (Fig. 3) is determined as follows.

where N is a point on the periphery, which is a reasonable approximation from a simple and quick calculation. is the number of

The distance of each edge vixel (X+, y +) from the centroid (X, Y) is , is calculated as follows.

D+=J [(X-x l) 2+ (Y V +)”Next, this image is Divide the toroid into six regions a, b, c, d, e, and f. Each area is , contains only one corner A, B, C, D, E, F. The corner corner is Centroi It may be the point within the farthest region. Each corner A, B, C, D, E, F is The highest edge points encountered in the course of step work for all of the areas of It is found by finding point D1.

Corners A, B, C, D, E, F (7) Coordinates (xA, YA), (x,, y, ), (xc, yc), (x+y, yo), (xi, ya), (XF, y Once F) is found, calculate the next dimension.

Side legs including waistband A C=f [(XA-XC) 2+ (yA yc)”] Leg opening C-E=J [(xc xi)”+(yc 3’ri”) Also width E F = J [(XE XF)”+ (yt-yp)”] West B A=J [(xi xA)”+(, Yl-yA)”] These dimensions are Check whether a garment of a given size is within the tolerance range or check whether all measurements are within the tolerance range for the size determined in this way. Used for checking.

The conveyor 12 stops for image formation and once the first image is front illuminated Once captured using the device 15 and the images sent for the above processing, With the front illumination turned off and the diagonal illumination 16 turned on, a second diagonally illuminated image is formed. It is convenient to do so. Since the seam llb is raised, this diagonal lighting Highlight any seam edge closest to the light source and remove shadows from other edges. to project. The position of the light source is such that the garment 11 has highlights and shadows for each seam. Choose to be on the conveyor 12 as always. In this way, the second image Captured under this diagonal illumination, information about the peripheral structure of this power dependence, Includes raised wide seams and waistband construction. Figure 4 shows the sea This has 41 highlights and 42 shades showing the edges and waistband of the mulch. A second image like this is shown.

Highlights 41 and shadows 42 are processed separately. To process shading, image (or rather a copy of it) into grayscale segments using a suitable threshold. image and leave only the shading information. The shading is not always clear and other parts of the clothing Since there may be false shadows from clarify and eliminate false information. One way to do this is, for example, vertically Directional characteristics (for Y-shaped frontal underwear, one of the lines, i.e. the vertical direction There is a way to isolate multiple features by starting with the shading on one edge of the procentric seam.

Define a box containing the top, bottom, and side points, and Test the straight line through the box (the straight line automatically generated by the -luck formula) This characteristic can be determined by selecting one straight line that passes through the majority of data points. Adapt the lines. Isolate the shading information using an appropriate segmentation threshold level. Therefore, this information is stored in the isolated area (rather than in the extended area), as shown in Figure 5. data point. Figure 5 shows a set of shadings on slanted seam edges. A trial line 51 adapted to point 52 is shown.

This method eliminates all other shadow areas until the sharpened shaded area can be incorporated into the image. Repeat for minutes. Next, perform the process on the highlight part as well to capture it first. A diagram similar to FIG. 6 is generated showing the image captured.

Next, divide the image into sections S separated by edges, shading, or highlights. 3. Divide into 32 etc. Each section S1%82 etc. (panel lla or wide seam llb) is then isolated and subjected to defect detection processing as appropriate. segment using appropriate threshold levels, filtering or any other suitable technique. Different extension regions from the threshold background (this allows for defects in the weir or in the weir or both ). Figure 7 shows the isolation as described above for defect detection work. The waistband region S1 is shown, and the remaining sections 82-S are sequentially examined. Inspect.

Of course, if a defect is detected, further inspection may be futile. If the warning or rejection mechanism 19 (Fig. 1) is activated, the inspection process It may be a good idea to close the program.

Here, one area may have more defects than others, and the equipment power is limited. store statistical values that can determine the defect incidence of panels lla, 1 of garment 11; It would be advantageous to be able to use it to grade 1b. then , select the section with the highest number of defects, first investigate the defect detection stage, and then operates heuristically. Of course, the statistical values collected in this way are from the perspective of defect inspection. It can also be used to decide which manufacturing area to tackle first. .

Copy and translation of written amendment submitted (Article 184-8 of the Patent Law) February 19, 1993 2. Name of the invention tiM ID 3. Patent applicant Address: United Kingdom, L.L. 19, B.H. Leicester, B.O. Box 143 people Name: De Montfort University Nationality: United Kingdom 4. Agent Address: 3-4-11-5 Shiba, Minato-ku, Tokyo Date of submission of amendment: July 1992 June 6th, 6th, List of attached documents Copy and translation of written amendment) The present invention provides a method for automatically inspecting garments having panels separated by seams. The law is Forming a machine vision image from clothing and automatically identifying seams on the image and the steps of automatically inspecting the panel areas individually.

The invention is particularly useful for garments in which the seams themselves have a width and the seams can also be inspected. Can be applied.

Seams that are raised above adjacent panels, more commonly x, y on the garment surface Seams with Z coordinates are affected by shadows and highlights in oblique lighting. can be detected.

The method of the invention has a waistband structure with wide, raised seams. Applicable for inspecting Y-shaped fronted undergarments containing edged panels and laid flat. The stage of forming the image of the front of the clothes l. A method for automatically inspecting garments with panels separated by seams. So, forming a machine vision image from the garment; and automatically detecting the seams on this image. 11; and automatically inspecting panel areas individually.

2. In the method according to claim 1, the seam itself has a width; A method characterized in that these seams are also inspected.

3. In the method according to claim 1 or 2, A seam with a Z coordinate relative to a mark is shaded and highlighted from an oblique light. A method characterized by detecting

4. The method according to any one of claims 1 to 3. wide raised seams, waistband construction and hemmed panels In a method for inspecting underwear with a Y-shaped front, the front of the garment is placed flat. Image formation stages and international search report international search report PCT/GB 91101414 Continuation of front page (81) Designated countries EP (AT, BE, CH, DE.

DK, ES, FR, GB, GR, IT, LU, NL, SE), 0A ( BF, BJ, CF, CG, CI, CM, GA, GN, ML, MR, SN, TD, TG), AU, BB, BG, BR, CA, FI, HU, JP, KP, K.R.

LK, MC, MG, MW, No, PL, RO, SD, SU, US

Claims (14)

[Claims] 1. A method for automatically inspecting garments with panels separated by seams. hand, forming an image from the garment; identifying seams on this image and inspecting panel areas individually; How to include. 2. The method according to claim 1, wherein the seam itself has a width; A method characterized in that the seams of the seams are also inspected. 3. In the method according to claim 1 or 2, the x, y coordinates of the clothing surface Seams with z coordinates by shadows, highlights from oblique lighting A method characterized by detecting. 4. A method according to any one of claims 1 to 3, , from wide raised seams and hemmed panels with waistband construction In the method of inspecting underwear with a Y-shaped front, an image of the front of the garment is placed flat. The steps of forming the image, placing the outline of the garment on the image, and adding width to the image. placing the raised seam, the edge of the waistband structure on the image; Above-imaged panels and wide raised seams, waistband construction and inspecting for defects. 5. In the method according to claim 4, the size of the garment is determined from the front image of the garment. A method characterized by measuring. 6. The method according to claim 4 or 5, wherein the garment is irradiated with front illumination. A method characterized by forming an image. 7. In the method according to claim 6, the clothes are irradiated with oblique lighting to make the clothes have a width. Position the raised seam, edge of the waistband structure, and place that image in front. A method characterized by superimposing images on an image formed by partial illumination. 8. 1. A method for automatically inspecting garments having seam-separated panels, the method comprising: an image forming means for forming an image of the garment; seam identification means for identifying a seam on the image formed by the image forming means; and panel inspection means for individually inspecting panel areas on the image. 9. 9. The apparatus of claim 8, wherein the imaging means is arranged on a back on which the garment lies flat. An area-scanning video camera aiming at the scene and front lighting illuminating clothing placed in the background, diagonal A device characterized in that it includes: a lighting device; 10. The apparatus according to claim 9, wherein the seam identifying means receives oblique illumination. Image processing that identifies highlights and shadows on the illuminated image and overlays them on the front-lit image of the garment. A device characterized in that it includes a means for controlling. 11. The apparatus according to any one of claims 8 to 10 The panel inspection means identifies the panel on the image and the image according to predetermined criteria. detects defects there and activates a warning device or garment rejection device when a defect is detected. An apparatus characterized in that it includes image processing means adapted to perform. 12. The apparatus according to any one of claims 8 to 11 and the panel inspection means is adapted to sequentially process multiple panels of the image. A device characterized by: 13. In the apparatus according to claim 12, during batch inspection of similar garments, A defect record designed to record the incidence of defects associated with each of a plurality of panels. A device characterized in that it includes means. 14. In the apparatus according to claim 13, the batch inspection progress is recorded. The plurality of panels are sequentially inspected according to the defect rate for each panel. An apparatus characterized in that it includes a heuristic control means for disposing: