JPH01250806A - Method and device for measuring width - Google Patents

Abstract

PURPOSE:To highly reliably measure the width of a protrusion part based on a center position of a contour by providing a binarization part, a contour center position detecting part and a contour center-to-center distance calculating part, etc. CONSTITUTION:An image horizontally scanning signal 1 is emphasized in its contour by a contour emphasizing part 2. Then, this is binarized by the binarization part 3 to extract the contour only. Then, the center position of each contour is calculated in the contour center position detecting part 4, and successively, the distance between the centers of two contours is calculates in the contour center-to-center distance calculating part 5. And, the above process is performed N-times with different horizontal scanning signals. N-number of data on the distance between the centers obtained by this method is relisted in descending order of value by a data relisting part 6, and M-number of data in upper order only is extracted by a data extracting part 7. After calculating statistical data, such as the max. value, min. value and mean value, etc., on the finally extracted data by a statistical calculation part 8, they are displayed by a display and decision part 9.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a width dimension measuring method and apparatus for measuring the track width of an ultra-precision machined part such as a magnetic head.

(Prior Art) Recently, with improvements in processing technology, ultra-precision machined parts that are machined with submicron dimensional accuracy have appeared. Typical examples are video tape recorders and magnetic heads of various magnetic storage devices. A magnetic head is usually manufactured by processing a block of ferrite magnetic material using an ultra-precision processing machine such as a micro grinder to a width of about 100 microns and an accuracy of ±0.5 microns. At this time, the most important thing is dimensional accuracy after processing. FIG. 6 shows an example of an external view after processing. 101 is a base material such as ferrite, 102 is a processed groove, and 103 is a convex portion formed by the groove. In the case of a magnetic head, since the portion 103 serves as the magnetic head, the dimensional accuracy of the width of 103, that is, the track width, is important.

Conventionally, an inspector visually inspects the dimensions of a processed object using a microscope.

However, this method has the problem that the inspection accuracy and inspection standards change due to individual differences among inspectors, the degree of fatigue, etc., and it is difficult to always measure dimensions using a constant standard.

Therefore, automating this process can be expected to have a very large effect because it can standardize the inspection standards and save manpower.

As shown in FIG. 7, the most commonly used automation method is to take an image of the inspected object magnified with a microscope using an ITV camera and measure it using an image processing device. 110 is a microscope, 111 is an ITV camera, 1
12 is an object to be inspected.

Further, 113 is an image processing device, 114 is a television monitor for confirmation, and 115 is a stage on which an object to be inspected is placed, which shows a measurement method using a conventional image processing device.

FIG. 8 is an example of an image (microscopically enlarged) of the object to be inspected taken with an ITV camera. Here, for simplicity, a diagram in which grooves are formed in the vertical direction of the screen is shown. In the example of the magnetic head shown above, since the surface before processing is mirror-polished, the shaved portion, ie, the groove portion 120, is dark, and the remaining convex portion 121 is photographed brightly. Therefore, the image processing device uses a certain brightness as a threshold value, converts brighter areas into a binary image with white pixels and darker areas as black pixels, and then calculates the number of white pixels in the horizontal direction of the image. do. If the dimension of one pixel in a photographed image is measured in advance, the calculated number of white pixels can be converted into length. Therefore, the width dimension 121 in FIG. 8 can be measured. The width of this 121 becomes the width of the convex portion 103 in FIG. For example, if you use a microscope with a 60x objective lens,
Since the size of each pixel is approximately 0.2 microns, submicron measurements are possible.

(Problem to be Solved by the Invention) However, a problem with this method is that when the binarization level is changed, the measurement width changes. The reason for this is that in reality, the boundaries between the grooves and the protrusions cannot be completely separated. This will be explained using FIG. 9. FIG. 9 shows an arbitrary horizontal image signal 130 for the image shown in FIG. The horizontal axis indicates the horizontal coordinates of the image, and the vertical axis indicates the brightness at each horizontal coordinate. As shown in FIG. 9, in a normal image, a convex region 131
A transition M region 133 exists at the boundary between the groove region 132 and the groove region 132 .

This occurs due to light diffraction and focal mismatch. Therefore, the length of the white pixel area is different between the case of binarization with a brightness of 134 and the case of binarization with a brightness of 135. Therefore, in the conventional method in which the length of the white pixel area is used as the measurement width, the measurement width differs when the binarization level is changed or when the surrounding brightness changes.

Furthermore, although not shown here, if there is chipping at the boundary or if grinding fluid from machining is attached to the surface, simply setting the length of the white pixel part as the length of the convex part will cause the measured value to change. There were problems such as the width being significantly different from the actual width of the convex portion, resulting in large measurement errors.

An object of the present invention is to solve the problems of dimension measurement using conventional image processing devices and to provide a highly reliable width dimension measurement method and device that can also measure submicron dimensions.

(Means for Solving the Problems) Therefore, the present invention has solved the above-mentioned problems of the prior art by providing a width dimension measuring method and an apparatus therefor as set forth in the claims.

(Examples) Hereinafter, the present invention will be described in detail with reference to Examples. FIG. 1 is a block diagram showing the configuration of a width dimension measuring device according to the present invention.

One image horizontal scanning signal 1 photographed by an ITV camera or the like is first subjected to contour enhancement in a contour enhancement section 2 .

Next, this is binarized in the binarization section 3 and only the contour line is extracted. Next, the contour center position detecting section 4 calculates the center position of each contour, and then the distance between contour centers calculating section 5 calculates the distance between the centers of the two contour lines.

The above processing is executed N times for different horizontal scanning signals.

The data sorting unit 6 rearranges the N distance data between the centers thus obtained in descending order of the value, and the data extracting unit 7 extracts only the top M data. After calculating statistical data such as the maximum value, minimum value, average value, and standard deviation for the finally extracted data in the statistical calculation section 8, the process is then displayed in the display/judgment section 9 and ends.

Furthermore, based on this statistical data, it can also be determined whether the machined object is machined to specified dimensions. The processing of each block 2 to 5 shown in FIG. 1 is executed using what is called a normal window that limits the image processing range. FIG. 2 shows an image taken by an ITV camera and a window.

10 is the periphery of the image taken by the ITV camera. The gray portion 11 is a valley portion that has been removed during processing. 12 is an unprocessed convex portion, which has a glossier surface than 11, so it is photographed brighter than 11. 1
Boundaries 13 and 14 between 1 and 12 are contour lines indicating edge portions. 15 is a window, and the area surrounded by 15 becomes the target area for image processing. For example, if N=128, the vertical width 16 of the window is 128.

In the horizontal direction, it is sufficient that the edge contour lines 13° and 14 at both ends of the convex portion are included. Here, 17, which will be used in the explanation later, is a chipped portion that occurred during processing. When the outline of the image shown in FIG. 2 is emphasized by the contour emphasizing section 2 of FIG. 1 and then binarized by the binarization section 3, an image as shown in FIG. 3 is obtained. Contour enhancement emphasizes parts where the brightness of an image changes, and a typical example is the Sobel operator. Therefore, when the contour-enhanced image is binarized, only the edge portion contour lines 13 and 14 in FIG. 2 become white pixels, and the rest become black pixels. That is, two boundary lines 18° and 19 between the convex portion and the adjacent grooves on both sides are obtained. The contour lines 13.14 in Fig. 2 are respectively the boundary lines 18.1 in Fig. 3.
Corresponds to 9. Ideally, the width of the border lines 18 and 19 is one pixel, but in reality, it has a width of several pixels. This is caused by light interference and out of focus, and is impossible to completely eliminate, especially in images taken through a microscope. Therefore, the boundary line 18.19
After determining the position of the center of the width, the distance between 18 and 19 for each horizontal pixel column (corresponding to the horizontal scanning signal) is determined as the distance between the centers.

If you measure based on the center position of the boundary line in this way,
Since the variation due to the width of the transition region 133 in FIG. 9 can be suppressed, it is possible to eliminate the problem of the measurement width varying due to the threshold value for binarization as in the conventional example. If the vertical width shown above is 128 pixels, 128 center-to-center distance data are obtained6. The edge after processing has unevenness in the vertical direction due to chipping, etc. If large chipping occurs, a chip 17 as shown in FIG. 2 will occur. Therefore, the distance data (measurement width) for a pixel column in the horizontal direction has a value width as shown in FIG. In particular, when a large chip like No. 17 occurs, the part with a small measurement width has a large width (FIG. 5). If statistical calculations are performed for all measurement widths as is, data on the large chip 17 will also be included in the calculations, resulting in a large error. When measuring the width of a convex part, chipping due to chipping occurs on the smaller side of the measurement width, so
The measurement width data is sorted in descending order of value, and only M pieces of data, for example, 50 pieces of data are extracted from the largest value, and statistical calculations are performed on this data. By doing so, measurement errors due to chipping or the like can be prevented. In preparation for actually using the measuring device according to the present invention, the size of the chip on the object to be measured is predicted and the value of M is determined. If the vertical size of the screen shown in FIG. 3, which occurs relatively frequently, is converted into the number of pixels and this is M', then the value of M can be set as follows: M=N-M'.

The statistical calculation results obtained by the above processing are measurement data excluding the chipped part of the workpiece, and the maximum value, minimum value, and average value of the measurement width of the part other than the chipped part are chipped. The average value and standard deviation of the measurement width excluding the part shown in the figure indicate the variation in the measurement width in the other part.

(Effects of the Invention) According to the width dimension measuring method and device of the present invention, the width of the convex part can be measured based on the center position of the contour line, so it is possible to measure the width of the convex part based on the center position of the contour line. Since fluctuations in the measurement width due to changes in can be removed, a highly reliable width measurement device can be realized. Furthermore, since only M pieces of data having a large value smaller than N are extracted from the N pieces of measurement width data and statistical calculations are performed on this data, measurement errors caused by chipping due to chipping can be removed. Therefore, it has become possible to provide a highly reliable width measurement method and device as an automatic measuring device.

When the width measuring device according to the present invention is installed in a processing device that processes to a predetermined processing size while feeding back measurement results, it is possible to improve the accumulation rate and eliminate disturbances such as chips, making it possible to fully automate such processing devices. It became.

The embodiment shown here is one example of the present invention and is not necessarily limited thereto. For example, in the description of the example,
Although the case where the width to be measured is in the horizontal direction of the screen has been described, it is not necessarily limited to this. In a photographed image, if the inclination of the object to be inspected is known, the position to be measured can be easily calculated as is well known. In addition, although we used an ITV camera and a window to measure the width of the convex portion at N different positions, it is also possible to use a method that combines a line sensor and a step-feeding single-axis stage. You can expect great effects.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a width dimension measuring device according to the present invention, FIG. 2 is an explanatory diagram showing an image of an object to be inspected and a window taken by an ITV camera, and FIG. 3 is shown in FIG. An explanatory diagram showing the contour line and window after the image signal photographed by the television camera is processed by the contour enhancement section and the binarization section shown in FIG. 1, and FIGS. 4 and 5 are shown in FIG. 1. 7 is a graph showing the relationship between the data values output by the data sorting unit and the number of measured data at different positions. FIG. 6 is a perspective view of an object to be inspected whose width dimension is to be measured, and FIG. 7 is a block diagram showing a schematic configuration of a conventional width dimension measuring device.
FIG. 8 is an explanatory diagram showing an image of the object to be inspected taken by an ITV camera, and FIG. 9 is a graph showing an image in an arbitrary horizontal direction with respect to the image of FIG. 1... Image horizontal scanning signal, 2... Contour emphasis unit, 3...
...Binarization section, 4...Contour line center position detection section, 5...
・Contour line center distance calculation unit (boundary line distance calculation unit), 6.
...Data sorting section, 7...Data extraction section, 8...
- Statistical calculation section, 9... Display/judgment section, 101... Inspection object, 102... Groove, 103... Convex portion, 13.1
4...Contour line (border line), 18.19...Border line. Agent Patent Attorney Jun Kawauchi 1st 2nd 3rd 4th 5th Tedahime Boo y
'+ Eup% 6 Diagram UI % 7 Mouth 8th 9th

Claims (1)

[Scope of Claims] (1) In a method for measuring the width of a convex portion between adjacent grooves of an object to be inspected, in which a plurality of grooves having a constant width are dug in parallel on a surface, The convex part is photographed by a television camera whose optical system is set so that the convex part is within the field of view, and the contour part is extracted from this photographed image, and the two adjacent grooves on both sides of the convex part are extracted. After extracting the contour line, that is, the boundary line of the book, measure the distance between the two boundary lines at any N different positions, and select the data in order from the largest value among these N measurement results. Extract M pieces of predetermined smaller data, calculate at least the minimum value, maximum value, average value, and deviation value for the extracted data to determine the convexity between the grooves provided on the object. A width dimension measuring method characterized by measuring the width of a portion. (2) The width dimension measuring method according to claim 1, wherein the contour portion is an area where brightness changes rapidly in a two-dimensional space formed by a photographed image of the object. (3) When the two border lines have a certain finite width, the width between the two border lines is measured based on the center position of the width of each border line. 1
Width dimension measurement method described in section. (4) In a width dimension measuring device that measures the width of a convex part between adjacent grooves of an object to be inspected, in which a plurality of grooves with a constant width are dug in parallel on the surface, the convex part is photographed. a television camera that outputs an image horizontal scanning signal; an edge enhancement unit that inputs the image horizontal scanning signal and extracts an edge portion; and an edge enhancement unit that inputs the output of the edge enhancement unit and binarizes it to generate an outline, that is, the edge A binarization unit extracts the two boundaries between the convex portion and the adjacent grooves on both sides, measures the distance between the two boundary lines at arbitrary different N positions, and outputs the measured data. a data sorting section that sorts the N pieces of measurement data in descending order of values; and a data extraction section that inputs the output of the data sorting section and extracts M pieces of data smaller than the N pieces. a statistical calculation unit that calculates at least the maximum value, minimum value, average value, and standard deviation of the M pieces of data extracted by the data extraction unit; and displays the output of the statistical calculation unit or the output of the statistical calculation unit. A width dimension measuring device comprising: a display/judgment unit that measures the width of a convex portion between grooves provided on the object. (5) The width dimension measuring device according to claim 4, wherein the outline portion is an area where brightness changes rapidly in a two-dimensional space formed by a photographed image of the object. (6) When the two boundary lines have a certain finite width,
The width dimension measuring device according to claim 4, further comprising a contour line center position detection unit that detects the center position of each boundary line, and wherein the boundary line distance calculation unit is a contour line center distance calculation unit. .