JP3126304B2 - Incorrect inspection method of engine bracket - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an erroneous product inspection method for an engine bracket for inspecting an erroneous product of an engine bracket and discriminating whether or not the erroneous product has occurred and a product type.

[0002]

2. Description of the Related Art There are several types of engine brackets for a vehicle depending on the mounting position and the type of vehicle. Therefore, after mounting, it is necessary to inspect whether or not a predetermined bracket is mounted on a corresponding engine or position. An example of the means will be described below with reference to FIG. FIG. 3 (a) is a front view of the engine bracket incorrect inspection apparatus (1). The inspection table (3) for positioning and holding the engine (2) and the inspection table (3) are set in a predetermined direction (perpendicular to the paper). (4), and first and second limit switches (Sa) (Sb) arranged vertically at an inspection position in the middle of the transport path.
And Then, engine brackets (5) and (5) are attached to side surfaces of the engine (2), and as shown in FIG. 3 (b), a short engine bracket (5a) is provided.
And long engine brackets (5b).

In the above configuration, the engine (2) with the engine bracket (5) is mounted on the inspection table (3) and transported. Then, the engine bracket (5) is inspected at the inspection position for erroneous products. For example, in the case of the long engine bracket (5b), since the shape extends obliquely upward from below, the first and second upper and lower limit switches ( Sa) (S
b) turns on upon detection. Short bracket (5a)
In this case, only the lower second limit switch (Sb) is turned on without being detected by the upper first limit switch (Sa).

[0004]

The problem to be solved is that when the shape of the engine bracket is slightly changed, it is necessary to set the positioning and repetition accuracy for the inspection in the order of several millimeters, which takes a lot of work, Further, when the engine bracket (5) is increased in length to four types by changing the length of the engine bracket (5) to two types due to a design change, the two limit switches (Sa) and (Sb) determine, for example, the longer one or the shorter one and the respective mounting positions. It is not possible to perform wrong product inspection. Accordingly, the present invention provides an erroneous product inspection method for an engine bracket that can determine an external appearance and inspect erroneous products even if the type or shape of the engine bracket changes.

[0005]

SUMMARY OF THE INVENTION According to the present invention, there is provided a method of capturing an image of an engine bracket from a predetermined direction, dividing the captured screen into a plurality of small areas, and measuring a characteristic amount of an image in each small area. Detecting the amount of change in the image feature amount between the regions,
The detection value is compared with the reference value, and the degree of coincidence with the reference value of the change amount is determined by fuzzy inference for each small area,
A step of inspecting the engine bracket for an incorrect product based on the determination over the entire small area.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a method for inspecting an engine bracket for erroneous parts according to the present invention will be described below with reference to FIGS. 1 (a) and 1 (b) and FIGS. 2 (a) and 2 (b). First, Figure 1
(A) shows an apparatus configuration for carrying out the inspection method according to the present invention. In the drawing, (6) is an article transfer inspection table,
(7) is a CCD camera, (8) is an image processing device, (9)
Is a personal computer for fuzzy inference. The inspection table (6)
Moves and moves on the transport path (12) in the direction perpendicular to the plane of the paper while positioning and holding the engine (11) to which the engine bracket (10) is attached. At this time, the engine bracket (10) was changed from the conventional two types of long and short to two
The number of types has increased for all four types.

[0007] The CCD camera (7) takes an image of the engine bracket (10) from a predetermined direction, for example, from the upper right side of the side, and an imaged image (F) is shown in FIG. 1 (b). In the figure, two circular portions are mounting holes for the engine bracket (10). As shown in FIG. 1 (b), the image processing device (8) divides the bracket imaging screen (F) by the CCD camera (7) vertically and horizontally into a plurality of small areas (F1) (F2).
, And the image feature amount, for example, the area (S1) of the image and the perimeter (L1) are measured for each of the small areas (F1) (F2). At this time, the information processing is simplified by measuring with a digital amount. However, the measurement may be performed by grayscale image processing (gray search). And each small area (F1) (F2)
... the amount of change (X1) between image feature amounts (S1) ... (L1) ...
(Y1)... Are detected for each small area. Therefore, the detected value is compared with a reference value {the area difference (A1)... And the perimeter difference (B1)... The difference from the reference image (P1 = X1-A1) (Q1 = Y1-B1) is calculated.

The personal computer (9) has a built-in fuzzy inference element, and the amount of change (X1) for each small area (F1) (F2).
It is determined by fuzzy inference how much (Y1)... Conforms to the reference values (A1)... (B1). For example, based on the difference (P1)... (Q1)... Calculated by the image processing device (8), each change amount (X1).
.., (B1)... (V1). Then, the determination is performed over all the small areas (F1), (F2)... To inspect the appearance of the engine bracket (10), and to determine whether or not an erroneous product has occurred and the type of the bracket.

The operation of the present invention based on the above configuration will now be described. First, an image of the engine bracket (10) is taken by a CCD camera (7) from a predetermined direction (rightward on the paper surface).
As shown in (b), the bracket imaging screen (F) is divided into a plurality of small areas (F1) by the image processing device (8). Therefore, the image feature amount, for example, the area (S1) and the perimeter (L1) are measured for each small area (F1). Therefore, when attention is paid to the small regions (F9), for example, the image feature amounts (area and peripheral length) between the adjacent upper, lower, left, and right small regions (F2) (F16) (F8) (F10) , The small area (F9) and its surrounding small areas (F2) (F16) (F8) (F10)
(X92), (X916), (X98), and (X910) and the perimeter differences (Y92), (Y916), (Y98), and (Y910). When the fuzzy inference is performed by the center-of-gravity calculation, the detected value and a preset reference value (A92) (A916) (A98)
The difference between (A910) and (B92) (B916) (B98) (B910) (P92 = X92-A92) ... (Q92 = Y92-B92) ... is calculated. Alternatively, the difference may be divided by a reference value to calculate a percentage value of the difference and standardized.

Then, the difference (P92)... (Q92)... Is input to the register of the personal computer (9), and fuzzy inference is performed. Of each area difference and perimeter difference between the reference values (U9
2) Judge (V92) ... Further, a fuzzy judgment is performed over all the small areas (F1)..., A plurality of data are judged, and the engine bracket is inspected for an incorrect product.

In this way, even if the shape and type of the engine bracket (10) change, an incorrect product inspection can be performed correspondingly. When the illumination conditions change due to a change in the amount of light in the surroundings, the value of the image feature amount such as the area itself changes, but the change amount of the image feature amount between the small regions does not change. for that reason,
Irregularity inspection of the engine bracket (10) can be performed accurately and stably regardless of changes in lighting conditions. In addition, if high reliability of the inspection cannot be obtained only with the data of the area and the perimeter, other image feature data such as the outline and the position of the mounting hole are added to the above data as needed to improve the reliability of the inspection. Can be done. Furthermore, considering the information processing time, small area (F1)
May be increased or decreased as appropriate.

At this time, the algorithm in the fuzzy inference described above, for example, if the difference between the change amount of the image feature value between the small areas and the reference value is small, the degree of coincidence becomes large and the captured image of the engine bracket (10) is previously If the set reference image is approached and the difference is large, the degree of coincidence is reduced and the reference image is deviated. Therefore, the following fuzzy rules are created according to the above algorithm.

(I) IF Pn (n = 1...) = ZR (Zero), THEN
Un (n = 1 ...) = PL (Positive Large) (II) IF Pn (n = 1 ...) = PS (Positive Small), THEN U
n (n = 1 ...) = PS (III) IF Pn (n = 1 ...) = PM (Positive Medium), THEN
Un (n = 1 ...) = ZR (IV) IF Qn (n = 1 ...) = ZR, THEN Vn (n = 1 ...) = PL (V) IF Qn (n = 1 ...) = PS, THEN Vn ( n = 1 ...) = PS (VI) IF Qn (n = 1 ...) = PM, THEN Vn (n = 1 ...) = ZR Also, as a membership function for executing a fuzzy rule, FIG. As shown in (1), the membership functions (Ma) and (Mb) of the triangle are set. The membership function (Ma) relates to the input part (% value), and the membership function (Mb) relates to the output part (degree of coincidence). Therefore, for example, assuming that Pn = 5% as input data, the matching degree is 0.5 in rule (I), 0.5 in rule (II), and 0 in other cases. Therefore, the output (coincidence) is calculated by the center of gravity
(Un) is near 2, which is quite close to the reference image. The above calculation is performed for all the image feature amounts over all the small areas, and the engine bracket (10) is inspected for an incorrect product.

In the fuzzy inference, there is a determination means based on a probability in addition to the determination based on the calculation of the center of gravity. For example, FIG.
As shown in (b), the fuzzy set membership function (Mc) of the amount of change (X1)... (Y1)... Value, (P
In Sa) and (NSa), each fuzzy set 位置 of the displacement and the determination probability (Dn) are set for each small area. Therefore,
The degree of conformity (Ra) (Rb) to the reference value and the displacement is detected from the amount of change (X1)... (Y1). Since the closer to the reference value, the higher the fitness (Ra) and the lower the fitness (Rb), they are compared with the judgment probability (Dn), and the change amounts (X1). The degree of coincidence (U1) (V1) is determined. For example, when Ra>Dn> Rb, the degree of coincidence (U1)... (V1)... Is determined to be within the normal range, and the determination operation is performed for each small area. Therefore, for example, the number of normal determinations is determined as a criterion from all determination results over all the small areas as a determination criterion, and the engine bracket (10) is inspected for an incorrect product.

[0015]

According to the present invention, the imaging screen of the engine bracket is divided into a plurality of small areas, the image feature amount in each small area is measured, and the reference value of the change amount of the image feature amount between each small area is measured. The engine bracket was inspected for erroneous parts by judging the degree of coincidence with fuzzy inference, so that even if the type and shape of the engine bracket changed, erroneous parts could be inspected correspondingly, and changes in lighting conditions such as disturbance light fluctuations etc. Irrespective of this, an accurate inspection becomes possible, and the inspection is stabilized, and the accuracy is also improved.

[Brief description of the drawings]

FIG. 1A is an apparatus configuration diagram showing an embodiment of a method for inspecting an engine bracket for erroneous parts according to the present invention.
(B) is a front view of the imaging screen of the engine bracket according to the present invention.

FIG. 2 (a) is a waveform diagram of each membership function of an input / output unit for performing fuzzy inference of a method for inspecting an incorrect product of an engine bracket according to the present invention. (B) is a waveform diagram of another membership function for implementing fuzzy inference of the method for inspecting an incorrect product of an engine bracket according to the present invention.

FIG. 3A is an apparatus configuration diagram showing an example of a conventional engine bracket incorrect part inspection method. (B) shows the conventional 2
It is a front view which shows a kind of engine bracket.

[Explanation of symbols]

 7 CCD camera 8 Image processing device 9 PC for fuzzy inference 10 Engine bracket F Imaging screen F1 ... Small area

Continuation of the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01B 11/00-11/30 102 G06T 7/00 G06T 7/60

Claims (1)

(57) [Claims] 1. A step of taking an image of an engine bracket from a predetermined direction, dividing the imaged screen into a plurality of small areas, and measuring an image feature amount in each of the small areas; The amount of change is detected, the detected value is compared with a reference value, the degree of coincidence with the reference value of the change amount is determined for each small area by fuzzy inference, and the engine bracket is determined based on the determination over all the small areas. A method of inspecting an engine bracket for an erroneous product.