JP2011043446A - Surface inspection apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a surface inspection apparatus for detecting the existence of a flaw on the surface of a machined workpiece, and estimating its position and size. <P>SOLUTION: The surface inspection apparatus 9 for inspecting the surface 3A inside a bore 3 cut and worked in a cylinder block 5 has: a one-dimensional power spectrum processing section 63 for extracting a one-dimensional digital intensity image extended by a predetermined length in the direction orthogonal to the cut/processed direction from a digital intensity image; a tool scale pitch transforming section 64 for transforming a pitch corresponding to a frequency f of a one-dimensional power spectrum data into a pitch data indicating a signal strength; an evaluation image generating section 55 for juxtaposing the pitch data, and generating an evaluation image; and an estimating section 57 for detecting the existence of the flaw based on the evaluation image. The one-dimensional power spectrum processing section calculates the one-dimensional power spectrum data while a length L is varied in the one-dimensional direction. The tool scale pitch transforming section synthesizes a plurality of the pitch data, and generates one set of pitch data. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a surface inspection apparatus for inspecting a workpiece surface subjected to machining.

In the automobile manufacturing process, a bore is bored in a cylinder block of an engine. In the bore boring process, the bore is formed by advancing and retreating with respect to the cylinder block while rotating the boring tool, so that a spiral machining mark is formed on the inner surface of the bore, and this machining mark is passed through the engine oil ( Oil pit).
In addition, since the inner surface of the bore becomes the sliding surface of the piston, the sliding surface has an appropriate surface roughness and surface properties in order to suppress the sliding resistance with the piston and allow the engine to exhibit desired performance. It needs to be formed. Therefore, usually, after the boning of the bore, honing is performed on the inner surface of the bore so as to leave an oil pit. Further, after the honing process, an inspection of the remaining polishing of the bore that causes sliding resistance is performed.
In this inspection, for example, a two-dimensional Fourier transform is performed on a digital image obtained by photographing the inner surface of the bore to generate a two-dimensional power spectrum image, and based on each pixel value of the two-dimensional power spectrum image, A technique for evaluating the smooth state of the inner surface of the bore is known (for example, see Patent Document 1).

JP 2004-132900 A

  However, in the inspection using the power spectrum image, although it is possible to determine the overall roughness of the inner surface of the bore, there is no spatial information in the power spectrum image, so polishing is performed based on the power spectrum image. It is not possible to know the range or size where the rest can be seen. Therefore, in order to specify the remaining polishing portion, it is necessary for the operator to visually find the digital image obtained by photographing the bore, or to directly observe the bore. Then, after confirming the remaining polishing portion, the operator makes a final decision as to whether it is an oil pit or the remaining polishing in consideration of the size and shape thereof.

In this way, the conventional inspection only knows the degree of overall roughness of the inner surface of the bore, and does not know the range or size of the remaining polishing. There was a problem that it was necessary to find out and confirm it visually, and it took a long time for the inspection.
The present invention has been made in view of the above-described circumstances, and can detect the presence or absence of scratches on the surface of a workpiece subjected to machining, and can estimate the position and size thereof. An object of the present invention is to provide a surface inspection apparatus capable of reducing time.

  In order to achieve the above object, according to a first aspect of the present invention, there is provided a surface inspection apparatus that inspects for the presence or absence of scratches based on a digital image of a surface of a machined workpiece, and is orthogonal to the machining direction. A one-dimensional power spectrum calculating means for extracting a one-dimensional digital image extending a predetermined length in the direction from the digital image and calculating a one-dimensional power spectrum, and the one-dimensional power spectrum corresponding to the frequency of the one-dimensional power spectrum. Pitch data conversion means for converting into pitch data indicating the signal intensity for each pitch; evaluation image generation means for generating an evaluation image by arranging pitch data sequentially obtained along the machining direction; and Detection means for detecting the presence or absence of scratches on the surface based on the evaluation image, and the one-dimensional power spectrum calculation means A plurality of the primary power spectra are calculated while varying a length in a one-dimensional direction of the one-dimensional digital image, and the pitch data converting unit is configured to calculate the plurality of one-dimensional power spectra calculated by the one-dimensional power spectrum calculating unit. Each is converted into pitch data, and each pitch data is synthesized to generate one pitch data.

  According to the present invention, a one-dimensional power spectrum is generated in which a direction orthogonal to the machining direction is a one-dimensional direction. In this one-dimensional power spectrum, since a signal corresponding to the machining depth is obtained at a frequency corresponding to the machining pitch, machining traces on the inner surface 3A of the bore 3 are separated from other irregularities. Separately, it can be extracted efficiently.

In addition, according to the present invention, pitch data obtained by converting the one-dimensional power spectrum is arranged in parallel, and an evaluation image that is an image in which the parallel direction corresponds to the machining direction is generated. With this configuration, it is possible to efficiently detect the presence or absence of a scratch due to a machining mark based on the evaluation image, and it is possible to specify a location where the scratch exists. Furthermore, based on the spread in the parallel direction of the location where the scratch exists, the size of the scratch (extended length) can be estimated.
Therefore, the operator can estimate the presence / absence of a flaw, its location and size without visually observing the workpiece, and the quality of the workpiece surface can be easily judged. Further, even when the operator actually confirms visually, the inspection time can be shortened because the scratched portion can be narrowed down.

  Furthermore, according to the present invention, when generating pitch data from a one-dimensional digital image, a plurality of primary power spectra are calculated while varying the length in the one-dimensional direction of the one-dimensional digital image, The power spectrum is converted into pitch data, and all pitch data is synthesized to generate one pitch data. According to this configuration, since the pitch sample points are interpolated by each pitch data, the resolution of the pitch data can be easily increased, and the detection accuracy of the flaw can be increased.

  To achieve the above object, according to a second aspect of the present invention, there is provided a surface inspection apparatus for inspecting an inner surface based on a digital image of an inner surface of a bore formed and polished in a cylinder block. A one-dimensional power spectrum calculating means for extracting a one-dimensional power spectrum from the digital image by extracting a one-dimensional digital image extending a predetermined length in a direction orthogonal to the processing direction; Pitch data conversion means for converting into pitch data indicating signal intensity for each pitch corresponding to the spectrum frequency, and pitch data sequentially obtained along the cutting direction are arranged in parallel to generate an evaluation image Image generating means, and detecting means for detecting the presence or absence of scratches on the inner surface based on the evaluation image, The evaluation image generating means calculates the plurality of primary power spectra while changing the length in the one-dimensional direction of the one-dimensional digital image, converts each one-dimensional power spectrum into pitch data, and converts each pitch data To generate one pitch data.

  According to this invention, there exists an effect similar to the 1st aspect of this invention. In addition, the inner surface of the bore that is difficult to visually recognize can be inspected with high accuracy.

  In order to achieve the above object, according to a third aspect of the present invention, there is provided a one-dimensional digital that extends to a predetermined length in a surface inspection apparatus that inspects for the presence or absence of scratches based on a digital image of a surface of a workpiece subjected to machining. One-dimensional power spectrum calculating means for extracting an image from the digital image and calculating a one-dimensional power spectrum, and pitch data indicating the signal intensity for each pitch corresponding to the frequency of the one-dimensional power spectrum. The pitch data conversion means for converting to the above and pitch data sequentially obtained along a certain direction are arranged in parallel to generate an image, and the certain direction is rotated by a predetermined angle with respect to the digital image, The image is generated at a rotation angle, and an image containing the highest high signal intensity is selected from the images as an evaluation image. Image generation means, and detection means for detecting the presence or absence of scratches on the surface based on the evaluation image, wherein the one-dimensional power spectrum calculation means is a length in the one-dimensional direction of the one-dimensional digital image. The pitch data conversion means converts each of the plurality of one-dimensional power spectra calculated by the one-dimensional power spectrum calculation means into pitch data, and changes each pitch. One pitch data is generated by combining the data.

  According to this invention, there exists an effect similar to the 1st aspect of this invention. In addition to this, it is possible to inspect the surface of a workpiece whose machining direction is not known in advance.

  According to a fourth aspect of the present invention, in the first to third aspects of the present invention, the length of the one-dimensional digital image is set to a length that provides an S / N that can detect a predetermined scratch. To do.

  According to the present invention, although the trade-off occurs between the resolution of the pitch data and the S / N depending on the length of the one-dimensional digital image, a good S / N is achieved despite the high resolution. be able to.

  According to a fifth aspect of the present invention, in the first to fourth aspects of the present invention, the one-dimensional power spectrum calculating means calculates the first power spectrum in the one-dimensional direction of the one-dimensional digital image. The calculation is performed while shortening the length.

  When calculating a plurality of primary power spectra while increasing the length of the one-dimensional digital image in the one-dimensional direction, it is necessary to re-extract the one-dimensional digital image from the digital image each time. Therefore, in order to calculate a plurality of one-dimensional power spectra while reducing the length of the one-dimensional digital image in the one-dimensional direction, the one-dimensional digital image only needs to be extracted once. Thereby, simplification of processing and reduction of processing time can be achieved.

According to the present invention, since the evaluation image in which the pitch data obtained by converting the one-dimensional power spectrum is arranged in parallel along the machining direction is generated, the scratch due to the machining mark is generated based on the evaluation image. It is possible to efficiently detect whether or not there is a flaw and to identify a place where the flaw exists. Furthermore, based on the spread in the parallel direction of the location where the scratch exists, the size of the scratch (extended length) can be estimated.
In addition, according to the present invention, when generating pitch data from a one-dimensional digital image, a plurality of pitch data are generated while varying the length of the one-dimensional digital image in the one-dimensional direction, and all pitches are generated. By synthesizing the data to generate one pitch data, pitch data obtained by interpolating the sample points can be obtained. As a result, the resolution of the pitch data can be easily increased, and the flaw detection accuracy can be increased.
Further, in the inspection of the inner surface of the bore, the inner surface of the bore that is difficult for the operator to visually recognize can be easily and accurately inspected.
In addition, the pitch data obtained sequentially along a certain direction is arranged in parallel to generate an image, and the certain direction is rotated by a predetermined angle with respect to the digital image, and the image is generated at each rotation angle. By selecting the image with the highest signal strength among the images as the image for evaluation, even if the direction of machining is unknown, the image with pitch data aligned in the direction of machining is used for evaluation. It can be an image.
In addition, by setting the length of the one-dimensional digital image to a length that provides an S / N that can detect a predetermined scratch, it is possible to realize a high S / N with high resolution.
Further, when calculating a plurality of primary power spectra, each time a plurality of primary power spectra are calculated, the one-dimensional digital image is calculated while reducing the length of the one-dimensional digital image in the one-dimensional direction. There is no need to re-extract the image from the digital image, and the processing can be simplified and the processing time can be shortened.

It is a figure which shows schematic structure of the cylinder block in which the bore | bore inner surface inspection system which concerns on one Embodiment of this invention, and the bore | bore used as test object were formed. It is the figure which showed the image produced | generated by the hole inner surface inspection along the flow of inspection. It is a figure which shows the production | generation process of the one-dimensional power spectrum image by the image production | generation part for evaluation. It is a figure which shows the relationship between a one-dimensional digital luminance image and a one-dimensional power spectrum. It is a figure which shows the relationship between the height of the extraction window which extracts a one-dimensional digital luminance image, the resolution of a tool eye pitch, and S / N. It is a figure which shows the shift of the sample point of the tool eye pitch by the height change of an extraction window. It is explanatory drawing of the interpolation process of a tool eye pitch. It is a flowchart of a bore | bore inner surface inspection process. It is a figure which shows schematic structure of the surface inspection system 100 which concerns on the modification of this invention with the workpiece | work of inspection object. It is a figure for demonstrating the determination of a process direction.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing a schematic configuration of a bore inner surface inspection system 1 according to an embodiment of the present invention and a cylinder block 5 in which a bore 3 to be inspected is formed.
The bore 3 is formed by a so-called boring process in which a cutting tool is protruded in a radial direction on a boring head provided on a rotating shaft, and the boring head is moved forward and backward with respect to the cylinder block 5 as a workpiece. By this boring, a spiral cutting trace having directionality is formed on the inner surface 3A of the bore 3. Thereafter, a honing grindstone is disposed on the inner surface 3A of the bore 3 in order to obtain a surface roughness and surface properties that can exhibit the desired performance of the engine while leaving an oil pit on the inner surface 3A of the bore 3. Honing is performed using the processed head.

The bore inner surface inspection system 1 evaluates the presence or absence of a polishing residue based on a digital image obtained by photographing the inner surface 3A of the bore 3 after the honing process. That is, the bore inner surface inspection system 1 includes a sensor head 7 that scans the inner surface 3A of the bore 3, and a surface that generates a digital image based on an output signal of the sensor head 7 and evaluates a polishing residue based on the digital image. An inspection device 9 and a drive mechanism 11 for moving and driving the sensor head 7 are provided.
The sensor head 7 has a cylindrical shape that can enter the bore 3 and is attached to the drive mechanism 11 so as to be rotatable around the central axis 12 and movable in the height direction. 3 is configured to be capable of sensing the entire inner peripheral surface. In detail, the sensor head 8 irradiates the laser light toward the inner surface 3A of the bore 3 from the opening 15 provided on the peripheral surface thereof, and reflects the amount of light according to the shape of the cutting trace. Is detected and output to the surface inspection apparatus 9. That is, the sensor head 7 includes an LD (laser diode) 17 as a light source, an optical fiber 19, and a condensing optical unit 21, and guides the light from the LED 17 to the condensing optical unit 21 through the optical fiber 19. And the laser beam is irradiated from the opening 15 onto the inner surface 3 </ b> A of the bore 3. The sensor head 7 includes a light receiving sensor 23 that receives the reflected light, and a plurality of optical fibers 25 that guide the reflected light returning from the inner surface 3A of the bore 3 to the light receiving sensor 23 via the condensing optical unit 21. Is disposed adjacent to the optical fiber 19.

The drive mechanism 11 includes a rotation drive mechanism 31 that rotates the sensor head 7 and an advance / retreat mechanism 33 that advances and retracts the rotation drive mechanism 31.
The rotation drive mechanism 31 includes a housing 34, a shaft 35 provided with the sensor head 7 at the tip and penetrating the housing 34 vertically, and a shaft that rotates the shaft 35 under the control of the surface inspection apparatus 9. A motor 37 and a rotary encoder 39 that detects the rotation speed and rotation angle of the shaft 35 and outputs the detected rotation speed to the surface inspection device 9 are provided.
The advancing / retreating mechanism 33 is a feed screw mechanism, and detects a shaft portion 41 in which a screw is engraved, an advancing / retreating motor 43 that rotationally drives the shaft portion 41, and a rotation speed and a rotation angle of the shaft portion 41 to detect a surface. 9 and a rotary encoder 45 that outputs to N. The shaft portion 41 is screwed into the nut portion 36 of the housing 34, and the shaft portion 41 is rotated by driving the advance / retreat motor 43, so that the rotation drive mechanism 31 is advanced and retracted.

  The surface inspection apparatus 9 controls the drive mechanism 11 to control the position of the sensor head 7 and an imaging that generates a digital image of the inner surface 3A of the bore 3 based on the light reception signal of the sensor head 7. An evaluation image generation unit 55 that generates an evaluation image 73 (FIG. 2) for evaluating the polishing residue based on the digital image, and an evaluation image 73 based on the evaluation image 73. And an evaluation unit 57. The surface inspection apparatus 9 can be configured by causing a personal computer to execute a program for realizing each unit, for example.

  Explaining each part of the surface inspection apparatus 9, the position control unit 51 incorporates a servo mechanism that drives the shaft motor 37 and the advance / retreat motor 43, and controls the position and rotation angle of the sensor head 7 on the central axis 12. That is, the position control unit 51 inserts the sensor head 7 into the bore 3 at the start of inspection and positions the opening 15 at the lower end position Ka of the inspection range K. Then, the opening 15 of the sensor head 7 reaches the upper end position Kb of the inspection range K in such a manner that the sensor head 7 is lifted while rotating around the central axis 12 so as to follow the locus of the boring bar during boring. The sensor head 7 scans the entire surface of the inspection range K in a spiral shape. This inspection range K is determined by a range that functions as a sliding surface with the cylinder.

The imaging unit 53 A / D-converts the light reception signal from the sensor head 7 and outputs it as a digital signal indicating luminance, and the inner surface 3A of the bore 3 based on this digital signal. And an imaging unit 61 that constitutes a digital luminance image 70 for the inspection range K.
As shown in FIG. 2A, the digital luminance image 70 is obtained by imaging the reflected light intensity obtained by the sensor head 7 at each inspection position in the bore 3 in correspondence with the inspection position. In the embodiment, the height position of the sensor head 7 and the rotation angle of the sensor head 7 are imaged with the vertical axis and the horizontal axis, respectively. In addition, the broken line in the digital luminance image 70 in the figure schematically shows a cutting trace (so-called tool eye) P at the time of boring.

As shown in FIGS. 2 and 3, the evaluation image generation unit 55 generates a one-dimensional power spectrum in a direction orthogonal to the direction of the cutting mark P based on the one-dimensional digital luminance image 70 </ b> A extracted from the digital luminance image 70. A one-dimensional power spectrum processing unit (one-dimensional power spectrum calculation means) 63 that sequentially generates data 80 along the direction of the cutting mark P, and a frequency of the one-dimensional power spectrum data 80 a tool pitch conversion unit (pitch data conversion means) 64 for converting the pitch data 81 into the pitch data 81 indicating the signal strength for each pitch corresponding to f, and the pitch data 81 is varied according to the signal strength of each pitch data 81. The evaluation image 73 is generated by arranging the converted light and dark replacement images 71 in parallel. Further, the evaluation image generation unit 55 calculates a plurality of one-dimensional power spectrum data 80 while reducing the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A, and converts each one-dimensional power spectrum data 80 into pitch data. An interpolation processing unit 65 that converts the data into 81 and combines the pitch data 81 to generate one pitch data 81 is provided. The one-dimensional power spectrum data 80, the pitch data 81, and the evaluation image 73 will be described in detail later.
The evaluation unit 57 functions as a detection unit that evaluates the polishing residue (the presence or absence of scratches) of the inner surface 3A of the bore 3 based on the luminance value of each pixel of the evaluation image 73.

FIG. 3 is a diagram illustrating a generation process of the light and dark replacement image 71 by the evaluation image generation unit 55.
In the evaluation image generation unit 55, an extraction window (column) 75 having a width W of 1 pixel (pix) and a length L of 200 pixels is defined as a window for defining a region to be subjected to one-dimensional power spectrum processing in the digital luminance image 70. Is preset. The direction of the length L of the extraction window 75 corresponds to the one-dimensional direction of the one-dimensional power spectrum data 80.
As shown in FIG. 2A, the evaluation image generation unit 55 arranges the extraction window 75 in the digital luminance image 70 so that the length L direction is orthogonal to the cutting mark P direction. As shown in (A), a digital luminance image in a range corresponding to the extraction window 75, that is, a one-dimensional digital luminance image 70A having a width W of one pixel is extracted. FIG. 3A schematically shows a cutting trace P that is not polished by the honing process (having a deep machining trace) as a polishing residue Q.

Next, the evaluation image generating unit 55 performs one-dimensional Fourier transform on the one-dimensional digital luminance image 70A, and as shown in FIG. 3B, one-dimensional power spectrum data indicating the signal intensity for each frequency f. 80 is generated. In the one-dimensional power spectrum data 80, an intensity signal corresponding to the depth of the cutting mark P appears at a frequency f corresponding to the pitch of the cutting mark P.
More specifically, as shown in FIG. 4A, in the one-dimensional digital luminance image 70A, when black and white changes for each pixel, the luminance value of each pixel is as shown in FIG. A pattern in which luminance values “250” (white) and “0” (black) appear alternately is represented by a luminance change in a one-dimensional direction (length L direction) as shown in FIG. A waveform is obtained. On the other hand, in the case of the power spectrum, black and white are switched every two pixels, so that a signal having a pitch corresponding to 2 pixels / cycle appears as shown in FIG. A signal appears at a frequency f corresponding to the reciprocal of this pitch (cycle / pixel). In the power spectrum, the frequency f takes a discrete value.
In the boring process, the cutting trace P has a spiral shape with a substantially constant pitch. Therefore, the cutting trace P also appears as a signal having a frequency f corresponding to the helical pitch in the one-dimensional power spectrum data 80. At this time, the signal intensity increases as the cutting mark P becomes deeper. When the inner surface 3A of the bore 3 has irregularities due to the dents or the like in addition to the cutting traces P, a signal having an intensity corresponding to the depth of the irregularities appears as another frequency f.

  Further, as shown in FIG. 3 (C), the evaluation image generation unit 55 uses a pitch T (hereinafter referred to as “tool”) obtained by discretizing the one-dimensional power spectrum data 80 indicating the signal intensity for each discrete frequency f. It is converted into pitch data 81 indicating the signal intensity for each eye pitch T). In the pitch data 81, the pitch and depth (size) of the cutting trace P distributed in the range of the extraction window 75 are indicated by the signal intensity of each tool pitch T. The unit of this pitch is expressed in pixels. By using the actual size (mm) per pixel in the digital luminance image 70, the pitch length expressed in pixels is expressed in actual size (mm). You can also.

The length L of the extraction window 75 will be described in detail. In the length L, values that make the resolution and S / N of the pitch data 81 good are set. That is, the resolution of the pitch data 81 depends on the length L of the extraction window 75. As shown in FIGS. 5A to 5C, the resolution decreases as the length L of the extraction window 75 decreases. . Therefore, in order to obtain a high resolution, it is necessary to increase the length L of the extraction window 75 to about 1200 pixels, for example, as shown in FIG.
However, when the length L is long, the component (signal) of the cutting trace P in the extraction window 75 is relatively reduced, so that the S / N is lowered. Further, in order to detect a fine (shallow) cutting mark P, the length L of the extraction window 75 is preferably small.
Therefore, by reducing the length L of the extraction window 75 to about 25 pixels, the S / N is improved and the fine cutting trace P can be detected, but the resolution is insufficient as described above. There is a problem. Thus, when the length L is varied, a trade-off occurs between the resolution of the pitch data 81 and the S / N.

  Therefore, the inventors first obtained the length L of the extraction window 75 from which an S / N that can detect significant scratches satisfactorily is obtained through experiments and the like. As a result, the knowledge that the length L = 200 pixels (pix) is optimal is obtained, and this length L is used for the extraction window 75. However, at this length L, since the resolution of the pitch data 81 is relatively low, there is a possibility that a significant flaw detection failure may occur. Therefore, the inventors decided not to increase the resolution of the pitch data 81 by increasing the length L of the extraction window 75 but to increase the substantial resolution as follows.

  That is, when the length L of the extraction window 75 of the one-dimensional digital luminance image 70A is varied, the sample point (discrete value) of the tool eye pitch T is shifted accordingly. Specifically, as shown in FIG. 6, when the length L of the extraction window 75 is reduced from 200 pixels to 199 pixels, the sample points of the tool eye pitch T are entirely increased by the amount corresponding to the decrease in the length L. Shifts in the direction of decreasing pitch length (left side). As a result, the signal intensity of the discrete value of the tool eye pitch T different from that when the length L is 200 pixels is interpolated. At this time, in the example shown in FIG. 6, since the length L of the extraction window 75 is reduced by 1 pixel, the information included in the extraction window 75 is lost by 1 pixel / 200 pixels (= 0.5%), but the pitch data 81 The number of sample points of the tool eye pitch T can be increased approximately twice, and the resolution can be substantially increased (200%).

  In this embodiment, as shown in FIG. 7, the length L is reduced by one pixel at a time until the length L is changed from 200 pixels to 195 pixels with respect to the same one-dimensional digital luminance image 70A. Five pieces of one-dimensional power spectrum data 80 are calculated, each one-dimensional power spectrum data 80 is converted into pitch data 81, and these pitch data 81 are synthesized. As a result, the sample point of the tool eye pitch T is increased by about 5 times, and the resolution of the pitch data 81 is substantially increased by 5 times.

  Instead of obtaining the pitch data 81 while reducing the length L of the one-dimensional digital luminance image 70A, the resolution may be substantially increased by obtaining the pitch data 81 while increasing the height L. However, when calculating a plurality of one-dimensional power spectrum data 80 while increasing the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A, the one-dimensional digital luminance image 70A is obtained from the digital luminance image 70 each time. Must be extracted. On the other hand, when calculating a plurality of one-dimensional power spectrum data 80 while reducing the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A, the one-dimensional digital luminance image 70A is extracted only once. Often, this simplifies processing and shortens processing time.

  Further, in the present embodiment, when the plurality of one-dimensional power spectrum data 80 is calculated, the length L of the one-dimensional digital luminance image 70A is reduced by one pixel. For example, it may be varied by 5 pixels). However, as shown in FIG. 6, the number of pixels to be reduced per time is the sampling point of the pitch data 81 obtained from the one-dimensional digital luminance image 70 </ b> A having the length L of the sampling point shift width M <b> 1 of the pitch data 81. It is preferable to set so as to be within the range of the interval width (discrete width) M2.

  Such processing is performed by the evaluation image generation unit 55. That is, as shown in FIG. 3C, the evaluation image generation unit 55 converts one piece of one-dimensional power spectrum data 80 (FIG. 3B) into pitch data 81 and the same one-dimensional digital signal. One-dimensional power spectrum data 80 is calculated while reducing the length L of the luminance image 70 </ b> A, and each is converted into pitch data 81. Then, all the pitch data 81 are synthesized, and as shown in FIG. 3D, the pitch data 81 is generated by interpolating the sample points of the tool eye pitch T.

  Next, as shown in FIG. 3E, the evaluation image generation unit 55 multi-values the pitch data 81 according to a legend in which the luminance value is lowered as the signal intensity increases, that is, as the scratch becomes larger (deeper). To generate a bright / dark replacement image 71. It should be noted that the size of the bright / dark replacement image 71 is the same as that of the extraction window 75.

  As shown in FIG. 2A, the evaluation image generation unit 55 moves the extraction window 75 in the direction of the length L along the direction A of the cutting mark P to move the rotation angle from 0 degrees to 360 degrees. Then, a bright and dark replacement image 71 is sequentially generated, and as shown in FIG. 2B, these are arranged in parallel along the cutting direction to generate an evaluation image 73. As a result, an image in which the light and dark replacement images 71 are arranged in parallel corresponding to the rotation angle of the sensor head 7 is obtained.

Based on the evaluation image 73 obtained in this way, the evaluation unit 57 evaluates the polishing residue Q. More specifically, as shown in FIG. 2C, the evaluation unit 57 excludes the oil pit and leaves only the strength corresponding to the polishing remaining Q more reliably. A binarized image 78 is generated by performing binarization processing with the luminance threshold value.
Then, as shown in FIG. 2D, the evaluation unit 57 extracts, for each pixel remaining in the binarization process, an extraction window 75 (that is, the light / dark replacement image that includes the pixel). 71) is applied to generate an abnormal area extraction image 79 in which the area corresponding to the extraction window 75 is color-coded by coloring. Thereby, in this abnormal area extraction image 79, the range R in which the polishing residue Q exists is clearly indicated by color. In addition to the remaining polishing Q, the color-coded range R includes various kinds of scratches such as dents caused by collision of a cutting tool or the like.

FIG. 8 is a flowchart of the bore inner surface inspection process performed by the bore inner surface inspection system 1.
After the cylinder block 5 in which the bore 3 to be inspected is formed is set at a predetermined position directly below the drive mechanism 11, the position control unit 51 causes the sensor head 7 to enter the bore 3 and advance and retract while rotating the bore 3. The inner surface 3A is scanned over the inspection range K, and the digital luminance image 70 of the inspection range K is generated by the imaging unit 53 based on the signal obtained by this scanning (step S1). Next, the one-dimensional power spectrum processing unit 63 of the evaluation image generating unit 55 sets an extraction window 75 extending in the direction perpendicular to the cutting trace P in the digital luminance image 70 by a length L, and the extraction window 75. A one-dimensional digital luminance image 70A is extracted from the range (step S2).

  Next, the one-dimensional power spectrum processing unit 63 calculates a plurality of one-dimensional power spectrum data 80 while reducing the length L of the one-dimensional digital luminance image 70A by one pixel (step S3). Then, each one-dimensional power spectrum data 80 is converted into pitch data 81 by the tool eye pitch conversion unit 64 (step S4), and all pitch data 81 is synthesized by the interpolation processing unit 65 to obtain sample points of the tool eye pitch T. Is generated (step S5). Next, the evaluation image generation unit 55 generates a multi-valued bright / dark replacement image 71 based on the signal intensity of the pitch data 81 (step S6).

  The evaluation image generation unit 55 moves the extraction window 75 along the cutting trace P in the digital luminance image 70 until the bright / dark replacement image 71 is generated for all of the inspection range K (step S7 is NO). (Step S8), and the generation of the bright / dark replacement image 71 is repeated. Then, the evaluation image generation unit 55 arranges these bright and dark replacement images 71 in parallel to generate an evaluation image 73 (step S9).

  Next, the evaluation unit 57 generates a binarized image 78 by binarizing the evaluation image 73 with a predetermined luminance threshold value in order to leave only the intensity corresponding to the polishing remaining Q (step S10). An abnormal area extraction image 79 is generated by color-coding the range corresponding to the extraction window 75 from which the remaining pixels have been extracted (that is, all the pixels of the bright / dark replacement image 71 including the remaining pixels) (step SS11). Next, when there is no coloring range in the abnormal area extraction image 79 generated in this way (step S12: NO), the evaluation unit 57 evaluates that there is no polishing residue Q on the inner surface 3A of the bore 3. If there is a coloring range (step S12: YES), it is evaluated that there is a polishing residue Q (step S14).

  If there is a polishing residue Q, the abnormal area extraction image 79 is displayed on a monitor device (not shown) and presented to the operator. According to the abnormal area extraction image 79, the size of the polishing residue Q can be estimated from the size of the colored range. Further, the position where the polishing residue Q exists can be easily estimated from the position of the coloring range, and it becomes easy to find the polishing residue Q when actually checking visually.

As described above, according to the present embodiment, the following effects can be obtained.
That is, according to the present embodiment, the one-dimensional power spectrum data 80 is generated in which the direction orthogonal to the direction of the cutting mark P is a one-dimensional direction. In this one-dimensional power spectrum data 80, a signal corresponding to the depth of the cutting trace P is obtained at a frequency f corresponding to the pitch of the cutting trace P, so that the cutting trace from the inner surface 3A of the bore 3 is obtained. P can be efficiently extracted while being distinguished from other irregularities.

In addition to this, according to the present embodiment, the bright and dark replacement images 71 that are multi-valued images of the pitch data 81 obtained by converting the one-dimensional power spectrum data 80 are arranged in parallel, and the parallel direction is in the direction of the cutting mark P. The evaluation image 73, which is a corresponding image, is generated. With this configuration, it is possible to efficiently evaluate the presence or absence of the polishing residue Q of the cutting trace P based on the pixel value of the evaluation image 73, and to specify the range R in which the polishing residue Q exists. . Further, the size (extended length) of the polishing residue Q can be estimated based on the spread of the range R in the parallel direction.
Therefore, the operator can estimate the presence / absence, the location and size of the polishing residue Q without visual observation, and the quality of the bore 3 can be easily determined. Even when the operator actually confirms visually, the inspection remaining time Q can be narrowed down, so that the inspection time can be shortened.

  Further, according to the present embodiment, when the pitch data 81 is generated from the one-dimensional digital luminance image 70A, a plurality of one-dimensional power spectra are reduced while shortening the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A. Data 80 is calculated, each one-dimensional power spectrum data 80 is converted into pitch data 81, and all pitch data 81 are combined to generate one pitch data 81. According to this configuration, since the sample point of the tool pitch P is interpolated by each pitch data 81, the resolution of the pitch data 81 can be easily increased, and the detection accuracy of the polishing residue Q can be increased.

  In particular, according to the present embodiment, since the length L of the one-dimensional digital luminance image 70A is set to a length that is S / N that can detect a flaw of a desired depth, the resolution is determined by the length L. Despite a trade-off in S / N, it is possible to achieve a good S / N with high resolution.

Further, according to the present embodiment, when calculating a plurality of one-dimensional power spectrum data 80 from the one-dimensional digital luminance image 70A, the calculation is performed while reducing the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A. Due to the configuration, the following effects are obtained.
That is, when calculating a plurality of one-dimensional power spectrum data 80 while increasing the length L in the one-dimensional direction of the one-dimensional digital luminance image 70A, the one-dimensional digital luminance image 70A is obtained from the digital luminance image 70 each time. Although it is necessary to extract, according to this configuration, since the one-dimensional digital luminance image 70A is calculated while reducing the length L in the one-dimensional direction, it is only necessary to extract the one-dimensional digital luminance image 70A once. Thus, the processing can be simplified and the processing time can be shortened.

  Further, according to the present embodiment, with respect to the abnormal area extraction image 79 obtained by binarizing the evaluation image 73, the pixels remaining by binarization are color-coded together with the pixels of the light / dark replacement image 71 including the pixels. It was set as the structure to do. By doing so, the range R in which the polishing residue Q exists is clarified, and it becomes easier for the operator to find the portion of the polishing residue Q when actually observing.

In addition, embodiment mentioned above shows the one aspect | mode of this invention to the last, and can change arbitrarily within the scope of the present invention.
For example, in the embodiment, the apparatus for inspecting the inner surface 3 </ b> A of the bore 3 is illustrated, but the present invention is not limited to the apparatus for inspecting the machined surface of the hole like the bore 3. That is, as shown in FIG. 9, the present invention can also be applied to an apparatus for inspecting a machined surface in which a flat surface of a workpiece 90 is cut in the same direction at a substantially equal pitch. In this case, since the machined surface is a flat surface, a digital luminance image 70 of the machined surface can be obtained by photographing the entire image with the camera 91 at a time.

Even if the machining direction (direction of the cutting mark P) is not known in advance, the machining direction can be specified as follows. That is, as shown in FIGS. 10A to 10C, each time the digital luminance image 70 on the machined surface is rotated by a predetermined angle, the light / dark replacement image 71 (pitch data 81) is converted into the light / dark replacement image. The evaluation image 73 is generated by sequentially generating and arranging in parallel along a direction B orthogonal to the one-dimensional direction (direction of length L) of 71. At this time, when the digital luminance image 70 is rotated so that the one-dimensional direction of the light / dark replacement image 71 is orthogonal to the direction of the cutting mark P, the image for evaluation whose strong signal intensity is most contained in each light / dark replacement image 71. 73 is obtained. Therefore, the machining direction can be specified by specifying such an evaluation image 73.
Then, as shown in FIG. 9, the machining direction determination unit 92 that determines the machining direction in this way is provided to constitute the surface inspection device 109 of the surface inspection system 100, so that the direction of the machining trace is known in advance. The surface inspection device 109 that can evaluate the machined surface of the workpiece 90 that has not been formed can be configured.

DESCRIPTION OF SYMBOLS 1 Bore inner surface inspection system 3 Bore 3A Inner surface 5 Cylinder block 7 Sensor head 9, 109 Surface inspection apparatus 55 Evaluation image generation part (Evaluation image generation means)
57 Evaluation part (detection means)
63 One-dimensional power spectrum processing unit (one-dimensional power spectrum calculation means)
64th tool pitch converter (pitch data converter)
65 Interpolation processing unit 70 Digital luminance image (digital image)
70A 1D digital luminance image (1D digital image)
71 Light-and-dark replacement image 73 Evaluation image 75 Extraction window 79 Abnormal area extraction image 80 One-dimensional power spectrum data 81 Pitch data 90 Work 91 Camera 92 Processing direction determination unit 100 Surface inspection system L Length of extraction window (one-dimensional digital image) P Cutting trace Q Polishing residue T Tool pitch (pitch)

Claims (5)

In a surface inspection device that inspects the presence or absence of scratches based on a digital image of the surface of a machined workpiece,
One-dimensional power spectrum calculating means for extracting a one-dimensional digital image extending a predetermined length in a direction orthogonal to the machining direction from the digital image and calculating a one-dimensional power spectrum;
Pitch data conversion means for converting the one-dimensional power spectrum into pitch data indicating signal intensity for each pitch corresponding to the frequency of the one-dimensional power spectrum;
An evaluation image generating means for generating an evaluation image by arranging pitch data sequentially obtained along the machining direction in parallel;
Detecting means for detecting the presence or absence of scratches on the surface based on the image for evaluation,
The one-dimensional power spectrum calculation means calculates a plurality of the primary power spectra while varying the length in the one-dimensional direction of the one-dimensional digital image,
The pitch data converting means converts each of a plurality of one-dimensional power spectra calculated by the one-dimensional power spectrum calculating means into pitch data, and synthesizes each pitch data to generate one pitch data. Surface inspection equipment. In a surface inspection apparatus for inspecting the inner surface based on a digital image of the inner surface of a bore formed and polished in a cylinder block,
A one-dimensional power spectrum calculating means for extracting a one-dimensional digital image extending from the digital image by a predetermined length in a direction orthogonal to the direction of the cutting, and calculating a one-dimensional power spectrum;
Pitch data conversion means for converting the one-dimensional power spectrum into pitch data indicating signal intensity for each pitch corresponding to the frequency of the one-dimensional power spectrum;
An evaluation image generating means for generating an evaluation image by arranging pitch data sequentially obtained along the cutting direction in parallel;
Detecting means for detecting the presence or absence of scratches on the inner surface based on the evaluation image,
The one-dimensional power spectrum calculation means calculates a plurality of the primary power spectra while varying the length in the one-dimensional direction of the one-dimensional digital image,
The pitch data converting means converts each of a plurality of one-dimensional power spectra calculated by the one-dimensional power spectrum calculating means into pitch data, and synthesizes each pitch data to generate one pitch data. Surface inspection equipment. In a surface inspection device that inspects the presence or absence of scratches based on a digital image of the surface of a machined workpiece,
A one-dimensional power spectrum calculating means for extracting a one-dimensional digital image extending to a predetermined length from the digital image and calculating a one-dimensional power spectrum;
Pitch data conversion means for converting the one-dimensional power spectrum into pitch data indicating signal intensity for each pitch corresponding to the frequency of the one-dimensional power spectrum;
The pitch data sequentially obtained along a certain direction is arranged in parallel to generate an image, the certain direction is rotated by a predetermined angle with respect to the digital image, and the image is generated at each rotation angle, An evaluation image generating means for selecting an image containing the largest amount of high signal intensity from each image as an evaluation image;
Detecting means for detecting the presence or absence of scratches on the surface based on the image for evaluation,
The one-dimensional power spectrum calculation means calculates a plurality of the primary power spectra while varying the length in the one-dimensional direction of the one-dimensional digital image,
The pitch data converting means converts each of a plurality of one-dimensional power spectra calculated by the one-dimensional power spectrum calculating means into pitch data, and synthesizes each pitch data to generate one pitch data. Surface inspection equipment.   The surface inspection apparatus according to claim 1, wherein the length of the one-dimensional digital image is set to a length that provides an S / N that can detect a predetermined scratch.   The one-dimensional power spectrum calculation means calculates the plurality of primary power spectra while reducing the length in the one-dimensional direction of the one-dimensional digital image. Crab surface inspection device.

Images