JPH03268178A - Preprocessing method of image in parts installation inspecting device - Google Patents

Abstract

PURPOSE:To prevent the generation of misdetection by applying processing for converting a brightness value more than a reference value for brightness conversion into a constant value to an input image and automatically converting the distribution of brightness in a searching range so that only a brightness difference between a target to be detected and its shadow appears. CONSTITUTION:When the searching range (s) is formed in an image forming its shadow 2 by lighting a part 3 to be inspected from an upper oblique position and the distribution of brightness is measured, the shadow (2) part of the parts 3 can be detected as the minimum brightness value 1c. The reference value is set up by adding a constant 1a (a comparatively small value) to the minimum brightness value 1c, and when the clipping processing of a high brightness part divided through the reference value is executed, the image of only the shadow (2) part is formed and the boundary of the shadow 2 can be detected as the outline of the parts 3. Even when the tone of the parts 3 is dark and the brightness different between the parts 3 and the shadow 2 is small, the outline of the parts 3 can be detected without generating misdetection due to the influence of a part having a large brightness difference from the shadow 2 such as the surface of the parts 3 and a printing part 1 around the parts 3.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention is a component mounting method for inspecting whether or not a component of a predetermined size, which is an object to be inspected, is mounted at a specified position in a component mounting inspection process. The present invention relates to an image preprocessing method in an inspection machine.

Conventional technology A conventional method for inspecting whether a component of a predetermined size mounted on a board is mounted at a specified position is to set the brightness value to a certain value for an input image of the object to be inspected and its surroundings. The component and its surroundings are displayed in white and black as a binary image set to a darkness value of A method of detecting the position, or adjusting the angle of illumination to obtain an input image of a part with a shadow, setting a darkness value for the brightness distribution of the input image and binarizing it, and then detecting the part and its shadow. There is a method of detecting the boundary line as a point of change in brightness value.

Problems to be Solved by the Invention However, according to the conventional method, in the method of setting the measurement window, in order to distinguish between parts and other parts, the luminance is divided into white and black at a certain value by 241Yf. It is necessary to set a dark value, and even if you take into account the color of the part and the color of the background when setting the dark value, the brightness value will vary due to the difference in reflection on the part surface, and the dark value will vary. It is not easy to set the value, and detection accuracy often becomes unstable. In addition, with the method of detecting the boundary line between a component and its shadow area, if it is difficult to set the lighting conditions depending on the component, or if the contrast between the component and its background is poor, it is possible to detect text on the component or print on the board. There were cases where the external shape was incorrectly detected.

Means for Solving the Problems The present invention obtains an input image in which the object to be inspected is illuminated diagonally from above to create a shadow, measures the luminance distribution within a prespecified search range, and calculates the luminance. The process of determining the minimum brightness value corresponding to the shadow area from the distribution, adding an arbitrary constant to the minimum brightness value to set a reference value for brightness conversion, and converting the brightness values above the reference value to a constant value is performed using the input described above. This method detects changes in brightness values that meet preset conditions by specifying the direction in which to search for the brightness distribution within the search range by applying this to the image. To detect the boundary line (edge) between the bright and dark parts when the area is dark, search in the direction of the arrow shown in the figure and specify the point where the bright part changes to the dark part as the detection target. do. Assuming that the brightness change on a certain line P-ρ' within the search range S is as shown in Fig. 1 (g)), the position detection process detects Kff as the brightness change point, and Determine the boundary as K. The boundary line work within the search range S is obtained by scanning the entire search range S in the search direction, detecting boundaries for each scanning line in the same manner, and connecting the boundary points.

It is easy to detect the boundary line when there is a clear difference in brightness like this, but in actual inspection objects, the surface color of the part is dark, so the difference in brightness from the shadow clearly appears. When there is no target or when there is a high brightness area within the search range, such as on a board or printed part, the input image is automatically converted to a state where only the difference in brightness between the target and the shadow appears. This is an image preprocessing method for a component placement inspection machine, which is characterized by the following.

According to the present invention, when a search range is provided in an image in which a shadow is generated by illuminating the part to be inspected from diagonally above and the brightness distribution is measured, the shadow part of the part can be detected as the lowest brightness value. Therefore, if you set a reference value by adding a constant (relatively small value) to the minimum brightness value and then clip the high-brightness part using this as a boundary, the image will contain only the shadow part, and the border of the shadow will be set to the outline of the part. can be detected as Therefore, even if the color tone of the part is dark and the difference in brightness from the shadow is small, the outline of the part can be detected without erroneous detection due to the influence of areas with a large difference in brightness from the shadow, such as the part surface or printed parts around the part. I can do it.

Example First, position detection processing will be explained using FIG.
Even if the odor position detection process is performed within the search range S specified in advance, the edge of the component that is the detection target may not be accurately detected. For example, as shown in FIG. 2(a), a case may be considered in which a dark-colored component is mounted on a board and the boundary between the component and the shadow is set as the detection target.

In Fig. 2(a), the part is illuminated from the upper right side of the figure, 1 is the silk-printed part on the board drawn around the part, 2 is the shadow of the part, 3 is the part, and 4 is the part surface. It is a character. In this situation, if the search range S is set as shown in the figure and the change in brightness value is as shown in Figure 2(b), edge candidates within the search range S can be considered at positions a to e. . Assuming that the search direction is designated as the direction of the arrow in the figure, the condition for detecting the detection target d is a change point from a dark area to a bright area. However, this condition is satisfied even at positions a and e, and the change in brightness at these positions is greater than at d, so it is possible that a and e will be detected as d and d in the position detection process. becomes more sexual. If we consider the search direction of the detection condition within the search range and add the detection order among the several detected candidates as a condition, it is possible to detect d in Fig. 2 (in this case, the detection (order is set to 2), there are generally variations in the position of the parts to be detected, and the detection range S is assumed to be set arbitrarily, so
The method of detecting by specifying the order described above is not universal and cannot be said to be practical.

Therefore, if we pay attention to the brightness changes in Figure 2 (g)), we can see that the brightness corresponding to the shadow 2 part of the part is the darkest, and the boundary between the part 1 and the shadow 2, which is the detection target, is the shadow 2 part. It can be seen that this is the position where the brightness first changes in the direction of becoming brighter.

Therefore, as shown in FIG. 3(a), the area corresponding to the search range is divided into a plurality of small areas (h).

h, ), the brightness values in each small area are measured and a brightness graph is drawn as shown in Figure 3i (b).The lowest value is set as Ic, and based on this, limping is performed as shown in Figure 4. Set conditions. Note that Ia is a constant value that can be set arbitrarily. In the case of FIG. 4, (5) a search range is set for the luminance-converted image, and position detection processing is performed.

Effects of the Invention According to the present invention, the brightness within the search range of the input image of the detection target is measured, and the clipping value is automatically set based on the measured value. When position detection processing is applied to
Even in cases where detection accuracy decreases due to the detection of silk-screen printing drawn on a board, letters on the surface of a component, etc., the brightness distribution within the search range is converted to a state where only the brightness difference near the detection target is detected. Therefore, compared to conventional methods, stable detection is possible that is less susceptible to the effects of color variations in target parts and extra noise information.

[Brief explanation of drawings]

The first is an explanatory diagram of a position detection processing method according to an embodiment of the present invention, FIG. 2 is an explanatory diagram of an image preprocessing method, FIG. 3 is an explanatory diagram of brightness measurement in image preprocessing, and FIG. To explain the clipping process, all parts having a luminance of Ic-11a or higher are converted to a luminance of Ic→-1a. When this clipping condition is applied to the image shown in FIG. 2(a), a brightness curve as shown in FIG. 2(C) can be obtained by selecting an appropriate value for Ia. Therefore, if the position detection process shown in FIG. 1 is applied to the state shown in FIG. 2(C), the target boundary position d will be detected correctly. The order of preprocessing for component position detection obtained from the above explanation is as follows. (1) Shine light diagonally above the object to create a shadow,
The image is manually created. (2) In a pre-specified search range, the range is divided into a plurality of small regions, and the luminance value of each of the divided regions is measured. (3) The lowest value among the measured luminance values is set as Ic, and rc-1-Ia, which is obtained by adding an arbitrarily settable constant 1a, is set as a clipping value. (4) Perform brightness conversion on the image input in (]) using clipping values. There is a graph. I a--- Minimum brightness value constant

Claims (1)

[Claims] (1) Obtain an input image in which the object to be inspected is illuminated from diagonally above to create a shadow, measure the brightness distribution within a pre-specified search range, and determine the area corresponding to the shadow from the brightness distribution. By determining the minimum brightness value, adding a constant to the minimum brightness value to set a reference value for brightness conversion, and applying processing to the input image to convert brightness values greater than the reference value to a constant value, An image preprocessing method for a component placement inspection machine characterized by automatically converting a brightness distribution into a state in which only a brightness difference between a detection target and a shadow appears.