JP2004317431A - Foreign matter inspection method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect foreign matters by using a threshold which can be used on any background and does not make a detection error. <P>SOLUTION: A set of data of the intensity of foreign matters and the threshold for detecting the foreign matters from a plurality of sample images is obtained in advance to calculate a regression curve and the following formula; threshold = coefficient × minimum intensity + offset. The threshold is determined based on the formula and is used as the threshold to threshold process the image for foreign matter inspection. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a foreign matter inspection method and apparatus, for example, a foreign matter inspection method and apparatus for detecting foreign matter adhering to an injector that performs fuel injection.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, in a manufacturing process of a product requiring high precision and high quality, foreign substances such as dust and burrs may adhere to the product, which often causes a problem in product quality. For example, the clearance of the needle 41 of the injector 40 that performs the fuel injection of the engine of the vehicle shown in FIG. 5 is about 50 μm, and if the foreign matter 42 is clogged there, the needle 41 is fully opened and the control becomes impossible. Such foreign materials include metal burrs and cutting powder generated in the manufacturing process, plastic trash such as polyacetal on a pallet on which the injector is placed, fiber trash or paper for work clothes, and metal or nonmetal such as human skin or nails. Miscellaneous things. Conventionally, a cleaning step of washing away foreign matter is performed every time one of the injector manufacturing steps is completed, and the washing is performed many times so that the foreign matter does not adhere. Further, as a part of the process control, a foreign substance inspection is performed to measure what kind of foreign substance adheres in which step, and a measurement of a foreign substance size and the like are performed.
[0003]
In the foreign substance inspection, the foreign substances washed out in the cleaning process are collected on a measurement filter such as a filter paper, a filter image enlarged by a microscope is photographed, and the obtained image is subjected to image processing to cut out the foreign substances from the background. The image is binarized in order to cut out the foreign matter. However, when the foreign matter is collected on the measurement filter, the background color of the filter differs for each process. Therefore, when binarizing foreign matter on the filter, it is necessary to adopt a dynamic threshold method in which a threshold is determined in conjunction with the background. Conventionally, the threshold is determined by the following formula.
Threshold = minimum luminance + coefficient × (maximum luminance−minimum luminance)
[0004]
Since the background and the foreign matter only need to be distinguished, it does not matter which one has higher or lower brightness. Here, the foreign matter is assumed to have lower brightness than the background. Therefore, the minimum luminance corresponds to the luminance of the foreign matter in the image.
[0005]
FIG. 6 is a histogram showing a light amount frequency distribution in a foreign matter image in one step of the injector manufacturing process. The horizontal axis of the graph in the figure is the luminance, and the vertical axis is the number of pixels having the luminance. The mountain B of the histogram is the luminance of many pixels, and represents the luminance of the measurement filter itself, that is, the luminance of the background. Since the foreign matter A is darker than the background and has a smaller number, the foreign matter A slightly appears in a portion where the luminance is lower than the peak of the histogram. The lines of the maximum luminance M and the minimum luminance L and the line of the threshold value H are shown on the histogram. In the prior art, the threshold value H was calculated from the maximum luminance and the minimum luminance and the set coefficient according to the above equation.
[0006]
[Problems to be solved by the invention]
However, as described above, the background color of the measurement filter from which the foreign matter has been collected differs for each process, and it is necessary to change the coefficients of the above equation in accordance with the background color. That is, it was necessary to know from which process the target foreign matter came out. In addition, when there is no foreign matter on the measurement filter, the maximum luminance and the minimum luminance match, and the luminance of the background becomes a threshold, and the background is erroneously detected as a foreign matter.
[0007]
In view of such a problem, an object of the present invention is to provide a foreign matter detection method and a foreign matter detection device that can be used in all processes and that use a threshold value without erroneous detection.
[0008]
[Means for Solving the Problems]
The present invention, in order to achieve the above object, a foreign matter inspection method or apparatus for inspecting foreign matter by binarizing an image, in which a regression curve is obtained from the luminance of the foreign matter and a threshold value for detecting the foreign matter in advance, Provided is a foreign substance inspection method or a foreign substance inspection apparatus that obtains a threshold based on the regression curve and uses the threshold as a threshold for binarizing the image.
[0009]
This regression curve can be determined by a linear equation.In addition, in order to obtain a regression curve, a sample at a predetermined distance from the regression curve obtained for the first time is excluded, and the regression curve is obtained again. Is also good.
[0010]
When the threshold value thus obtained is used, it is not necessary to know the process in which the foreign matter is obtained, and the background is not erroneously detected as the foreign matter.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION
An embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a diagram showing an outline of a foreign matter inspection apparatus for inspecting, for example, an injector foreign matter, which is one embodiment of the present invention.
[0012]
The microscope 1 is, for example, a stereo microscope, and has a CCD camera 11. The eyepiece unit 12 is a pair of binoculars, and an XY table 14 on which a sample is placed is provided below the objective unit 13. The XY table 14 is provided with a measurement filter for collecting foreign substances. For example, a table on which four measurement filters can be mounted may be used. A known attachment required for other operations can be attached to the microscope 1. The image analysis device 2 is a device for analyzing a sample image acquired by the microscope 1 and includes, for example, a personal computer, and includes a main body 21, a display device 22, a keyboard 23, and the like. The main body 21 includes a storage device such as a hard disk, and may include other known input devices and output devices (not shown).
[0013]
Next, a foreign substance detection step of the present embodiment will be described.
First, foreign substances are collected in a measurement filter by cleaning the injector in each cleaning step. The diameter of the measurement filter used is 50 mm.
[0014]
Next, a predetermined number of samples are extracted from each step. For example, in the case of five steps, about 6 to 8 samples may be taken in each step. FIG. 3 shows a sample image on the display device 22 (FIG. 1). The sample image is obtained by dividing the measurement filter by about 60 and has a size of 4 to 5 mm in length and width. Since the number of pixels is 640 × 480, one pixel is about 7.5 μm. The foreign matter 3 shown in FIG. 3 is fiber waste.
[0015]
While observing such a foreign substance sample image, an optimal threshold value at which the foreign substance can be clearly identified is visually determined. In this case, the foreign substance often has the minimum luminance value in the image, and the foreign substance may be represented by the minimum luminance value in the image. In the present embodiment, the luminance is divided into 256 levels from 0 (dark) to 255 (bright). For example, if the optimal threshold value which can be clearly separated from the background at the minimum luminance of 100, which is the luminance of a foreign substance, is 120, A data set of the minimum luminance value 100 and the threshold value 120 is obtained.
[0016]
In this way, a data set of the minimum luminance value and the threshold value is obtained. This data is plotted on a graph with the horizontal axis representing the minimum luminance value and the vertical axis representing the threshold value. The plotted result is shown in FIG. The obtained data did not show a relationship depending on the process from which the data was collected (for example, plotted collectively for each process). That is, the foreign substances collected from each step were plotted in a manner scattered throughout the graph. As shown in FIG. 3, a regression curve L1 (regression line in this embodiment) was obtained from this data, and its coefficient was determined. The resulting linear equation is
y = 0.4961x + 93.32
Met. In this equation, since y is a threshold and x is a minimum luminance value,
Threshold = coefficient × minimum luminance value + offset
[0017]
By using the threshold calculated using this equation, it is possible to detect a foreign substance without having to know in advance which process the measurement image has come from. For an image in which no foreign matter is present, the background luminance is the minimum value and the maximum value. However, the threshold value according to the present invention is obtained by adding a coefficient to the minimum luminance value and adding an offset portion. The background is not picked up as foreign matter.
[0018]
Further, in obtaining the regression curve, after obtaining the regression curve shown in FIG. 4, a sample S (shown by a dashed line in FIG. 3) apart from the obtained regression curve by a certain distance or more is excluded. To find the regression curve again. FIG. 5 shows the regression curve obtained again. The resulting equation is
y = 0.6772x + 64.185
Met.
[0019]
In the case of FIG. 4, the regression curve is obtained by excluding the four points showing unique values, so that the value of the correlation coefficient becomes higher than that of FIG. 3, and the threshold value is determined more accurately. I was able to.
[Brief description of the drawings]
FIG. 1 is a diagram showing a foreign object detection device according to an embodiment of the present invention.
FIG. 2 is a diagram showing an example of a micrograph of a foreign substance.
FIG. 3 is a diagram illustrating an example of a threshold value determining unit according to the present invention.
FIG. 4 is a diagram showing another example of the threshold value determining means according to the present invention.
FIG. 5 is a diagram showing the influence of foreign matter on the injector.
FIG. 6 is a diagram showing a histogram of a foreign object image and a threshold.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Microscope 11 ... CCD camera 12 ... Eyepiece 13 ... Objective lens 14 ... XY table 2 ... Image analyzer 21 ... Main body 22 ... Display device 23 ... Keyboard 3 ... Foreign matter

Claims (8)

A foreign matter inspection method for binarizing an image and inspecting the foreign matter, wherein a regression curve is obtained in advance from the luminance of the foreign matter and a threshold for detecting the foreign matter, and a threshold obtained based on the regression curve is obtained by converting the image to A foreign substance inspection method used as a threshold for binarization. The foreign matter inspection method according to claim 1, wherein the foreign matter is a minimum luminance portion of an image. The foreign matter inspection method according to claim 1, wherein the regression curve is determined by a linear equation. The foreign matter inspection method according to any one of claims 1 to 3, wherein the regression curve is a regression curve obtained by excluding a sample at a predetermined distance from the obtained regression curve. What is claimed is: 1. A foreign matter inspection device for binarizing an image and inspecting foreign matter, wherein a threshold value for binarizing the image is based on a regression curve previously obtained from a luminance of the foreign matter and a threshold value for detecting the foreign matter. Foreign matter inspection device using a threshold value obtained by The foreign matter inspection device according to claim 5, wherein the foreign matter is a minimum luminance portion of an image. The foreign matter inspection device according to claim 5, wherein the regression curve is determined by a linear expression. The foreign substance inspection device according to any one of claims 5 to 7, wherein the regression curve is a regression curve obtained by excluding a sample at a predetermined distance from the obtained regression curve.

Images