JPS6336106A - Defect inspecting method for bearing - Google Patents

Abstract

PURPOSE:To improve the inspection accuracy of rolling bodies by performing a mask processing on the basis of the radius calculated from area, extracting only the array area of the rolling bodies, and counting rolling bodies having specific area in the area. CONSTITUTION:The image of a needle bearing 5 obtained by a camera is inputted first and a light-dark binary coding processing is performed, and then a spider part 8 is masked and erased. Then, the number and area of black part bodies in the image plane are found to recognize the presence of the bearing 5, and the center coordinates and radius R of the inside part of a ball part 6 are found on the basis of the area. Circular masking is performed to the radius obtained by adding an offset quantity to the radius R inside and outside to extract only the array; area G of a needle part 7, thereby detecting the number of needles 7a whose area is within control limits from an area G.

Description

DETAILED DESCRIPTION OF THE INVENTION The invention relates to a bearing defect inspection method, also known as a method for inspecting the number of rolling elements such as dividing needles by pattern recognition.

BACKGROUND OF THE INVENTION As a conventional technique, for example, there is a method shown in the WCA diagram and FIG. 4 as a method for inspecting needle bearings of constant velocity joints for automobiles. This is a needle bearing (hereinafter simply referred to as a bearing) set in jig 1 as shown in Figure 3.
5 is illuminated by a ring light 2, while the image of the art stage 1 captured by a camera 3 such as a CCD is captured by an image processing device 4 and processed.

Bearing 5Vi As shown in FIG. 4, a ball portion 6. as an outer ring. Needle portion 7 as a rolling element. The image taken into the image processing device 4 is binarized, and then a preset t-fixed circular mask is used to form a ball part 6. Part of the spider section 8 and needle section 7 is masked. As a result of step 7, only the central portion (shaded portion) of the needle portion 7 is extracted as shown in step ① in FIG.

Then, the total area of the shaded portion is calculated, and if the area is within a predetermined control limit, it is determined that the needle has no defects.

Problems to be Solved by the Invention However, in the conventional method as described above, the total area of the needle portion is determined and a check is made to determine whether or not the total area of the needle is within the allowable control limits. If there is a difference in the gloss of each needle, or if thin noise occurs on the screen, the total area will vary, and the needle may be mistakenly recognized as missing.

Means for Solving the Problems The present invention attempts to improve inspection accuracy by checking the area of each rolling element such as a needle and increasing the number 'a'ff.

(JSpecifically, as shown in the examples, a method of inspecting the number of rolling elements by pattern recognition for a bearing in which a large number of rolling elements are interposed between an inner ring and an outer ring, Of the digitized image, the inner ring part l is masked and erased, and the area l is obtained by subtracting the area of the rolling element from the area of the inner part from the inner diameter of the outer ring. A royal right to determine the center coordinates and radius of the inner part based on the above-mentioned radius value, a royal right to extract only the array area by masking the area other than the arrangement area of the rolling elements based on the value of the radius, It includes the royal power to count the number of rolling bodies with area l.

Embodiment The basic system configuration for carrying out the present invention is the same as that shown in FIG. 3, and an embodiment of the image processing method is shown in FIGS. 1 and 2.

The image inputted into the image processing device 14 at step (2) in FIG. 1 is first subjected to a light/dark binarization process (step (2)). In the binarized state, as shown in FIG. 1(a), the ball portion 6. The needle portion 7 and the spider portion 8 appear white due to the reflected light. Then, in step (2), a fixed circle mask is applied to mask and erase the spider part by 8 square meters as shown in FIG. 1(b). In this case, the bearing 57 is fixed to the jig 1 as shown in Fig. 3, but there may be a positional deviation between it and the fixed circular mask due to positioning error in the bearing 11. , the center coordinates, radius, etc. of the fixed circle mask should be determined empirically in advance so that the needle 5 (S7) will not be masked.

Next, for object recognition in step ■, see Figure 1(b).
The number and area of objects in the black part are determined in the screen, and the presence or absence of the bearing 5 as a workpiece is determined. That is, when the bearing 5 is in the predetermined position, the portion A inside the inner circumferential surface of the ball portion 6 and the portion A inside the ball portion 6
The black object is divided into two parts, part B, which is outside the outer peripheral surface of
The existence of the bearing 5 can be recognized by looking at the number of bearings and the area of the inner part A. In the unlikely event that the bearing 5 is not in the specified position, the image in Figure 11(b) will be in a so-called pitch black state, and it will be recognized that there is one black object.
Outputs abnormal signal.

Then, in steps ■ and ■, the predetermined area of the object recognized in the previous step is determined! The holding side is determined to be the inner part A of the ball part 6, and the center coordinates (center of gravity) and radius R are determined based on the area previously calculated for the inner part of the ball part 6.
and! demand. In this case, the determined radius R is smaller than the actual radius of the inner peripheral surface of the ball portion 6. The reason for this is that although the radius R is calculated from the area of the inner public figure, the total area calculated in the previous step does not include the area shown in Figure 1 (b).
This is because the areas of the individual needles 7a of the needle portion 7 are not included, as shown in T in C).

Next, in steps (1) and (2), the radius Rt (R+a) is calculated by adding 7 points inside and outside of the previously determined radius R as shown in FIGS. 1 and 2.

R, (R-b) is applied with a circle mask, and the radius R1* R2 surrounds the area t, which is masked except for the t part, and the radius R1t R2
Only the submolecule surrounded by , that is, the arrangement region G of the needle part is extracted.

Furthermore, in step (2) of FIG. 1, the area of each needle 7a is calculated in the extracted image of FIG. 1(d), and it is determined whether the individual area is within a predetermined control limit Then, the number of needles 7a having an area l within the control limit is counted, and the number of needles 7a is detected. Then, by comparing the detected number of needles with a predetermined number, it is determined whether or not a needle is missing.

Here, when calculating the area of each needle 7a individually in Figure @1 (d), we also calculate the distance between the centers of gravity (center-to-center distance) l of each needle 7a to check whether it is indeed a needle or not. It is also possible to judge.

In this case, screen noise and erroneous recognition of the part leaking from the mask as a needle are definitely eliminated, and the inspection accuracy is further improved.

Effects of the Invention According to the present invention, only the arrangement region of the rolling elements is extracted by performing mask processing on a radius of 7 bases calculated from the area, and the number l of rolling elements having a predetermined area is counted from this arrangement region. [1] Misrecognitions are less likely to occur when there is a difference in gloss between individual rolling elements or when noise occurs, and inspection accuracy can be improved.

[Brief explanation of the drawing]

@Figure 1 is a flowchart showing one embodiment of the present invention 1;
The figure is an enlarged view of the main part of FIG. 1 (6), FIG. 3 is an explanatory diagram of the configuration of a defect inspection system, and FIG. 4 is an explanatory diagram showing a conventional image processing method. 5... Needle bearing, 6... Ball part as outer ring, 7... Needle part, 7a... Needle as rolling element, 8... Spider part as inner ring, A...
-Inner part, G...Array area. Figure 3

Claims (1)

[Claims] (1) A method of inspecting the number of rolling elements by pattern recognition for a bearing in which a large number of rolling elements are interposed between an inner ring and an outer ring, the inner ring portion of a binarized image being a step of masking and erasing and finding the area of the part inside the inner diameter of the outer ring minus the area of the rolling elements; and a step of finding the center coordinates and radius of the inner part based on the area of the inner part. , a step of masking a portion other than the arrangement region of the rolling elements based on the value of the radius and extracting only the arrangement region, and a step of counting the number of rolling elements having a predetermined area from the arrangement region. A bearing defect inspection method characterized by comprising: