JP2013029350A - Appearance inspection method and device for the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an appearance inspection method of an object which quantitatively evaluates a defect in the shape which is difficult to be detected in a visual inspection to be detected in an appearance inspection of the object in the complicated shape and a device for the same.SOLUTION: A method for inspecting appearance of an object is constituted so that an inspection target object is loaded, the inspection target object is picked up while continuously moving the inspection target object in at least one direction, surface unevenness information on the inspection target object is acquired while acquiring an image of the inspection target object including texture information on the surface of the inspection target object, the three-dimensional shape of the inspection target object is restored from the acquired surface unevenness information on the inspection target object, appearance information on an object with surface texture is obtained from the acquired image and the restored three-dimensional shape of the inspection target object, a plurality of features are extracted from the obtained appearance information, at least one feature of the plurality of extracted features is compared with a feature corresponding to the at least one feature of preset reference data, and the appearance of an object to be an inspection target is evaluated.

Description

  The present invention relates to an appearance inspection for detecting a defect of an object to be inspected from a surface texture (image) and a surface displacement (shape) of the object to be inspected, and a complicated shape such as a cutting tool such as a drill or a cutter or a turbine wheel. The present invention relates to an appearance inspection method and an apparatus suitable for inspecting a defective object.

  In the production site for metal workpieces, the quality of the processed product of a tool such as a drill or a cutter used for processing varies greatly depending on the shape of the blade shape. Therefore, management of the blade shape is important. To manage the blade shape, it is necessary to quantify the angle of the blade edge and grasp the presence or absence of chipping and the degree of wear. However, visual inspection and management of an object having a complicated shape such as a drill or a cutter has a shielding area, and a person views the object from various directions while moving the target object, and detects defects. For this reason, there is a limit in quantitatively measuring the subtle angle difference and surface irregularities of the cutting edge, and it is possible to prevent defective products by replacing the blade in advance or The blades are checked and replaced in the wake of the large number of occurrences. Therefore, there is a problem that the quality of the processed product is not stable, the tool replacement cost is unnecessarily increased, and it is difficult to detect visually, and it is possible to detect delicate shapes and surface irregularities with high accuracy and quantitative. The need for is arising.

  Three-dimensional shape measurement has been conventionally performed. In the method described in Japanese Patent Laid-Open No. 2002-92632 (Patent Document 1), the measurement object is irradiated with laser light, and the formation position of the irradiation spot of the laser light is changed by scanning to change the formation position. The reflected light at each irradiation spot is received, and the distance of the position of the irradiation spot is detected by the principle of triangulation. Then, the obtained three-dimensional shape data is displayed with a shadow using the light receiving width of the reflected light. As another technique, the method described in Japanese Unexamined Patent Application Publication No. 2009-204425 (Patent Document 2) controls the emission intensity of laser light applied to an object and receives light according to the controlled emission laser intensity. The reflection intensity to be made constant. As a result, the portion where the calculation of the three-dimensional data is impossible due to the high or low reflectance is reduced, and the three-dimensional shape measurement with high accuracy is enabled.

JP 2002-92632 A JP 2009-204425 A

  As described above, the quality inspection of a three-dimensional shape is often a visual inspection by an inspector. For this reason, the inspector makes a difference in the determination result of the surface unevenness and the determination result of the presence or absence of micro scratches on the surface, which causes variations in quality and outflow of defects. In order to solve these problems, it is necessary to measure the shape and the state of the surface, integrate them, and suppress the quality of the inspection object according to a comprehensive and quantitative determination criterion.

  Patent Documents 1 and 2 describe that a measurement target is irradiated with a laser and its reflected light is detected to obtain three-dimensional shape data of the measurement target. It is not described that the quality of the shape is determined from the data quantitatively.

  An object of the present invention is to inspect an object to be inspected from a surface texture (image) and a surface displacement (shape) of an object to be inspected in a visual inspection of an object having a complicated shape such as a cutting tool such as a drill or a cutter or a turbine wheel. It is an object of the present invention to provide an appearance inspection method and apparatus for quantitatively detecting and visualizing defects in the apparatus.

  In order to solve the above-mentioned problems, in the present invention, an apparatus for inspecting the appearance of an object is placed on a table means that can place an object to be inspected and can move continuously in at least one direction, and the table means. An image acquisition unit that captures an image of the inspection target object and acquires an image of the inspection target object including texture information of the surface of the inspection target object, and a surface that acquires surface unevenness information of the inspection target object placed on the table unit Displacement information acquisition means, three-dimensional shape restoration means for restoring the three-dimensional shape of the inspection target object from the surface unevenness information of the inspection target object acquired by the surface displacement information acquisition means, and the image and three-dimensional shape restoration means acquired by the image acquisition means Appearance data generation means for obtaining the appearance information of the object having the surface texture from the three-dimensional shape of the inspection target object restored in step 3, and whether the appearance information obtained by the appearance data generation means A feature extraction means for extracting a plurality of features, and an inspection object in comparison with a feature corresponding to the at least one feature of reference data set in advance at least one of the plurality of features extracted by the feature extraction means And an appearance evaluation means for evaluating the appearance of the object.

  In order to achieve the above object, according to the present invention, an apparatus for inspecting the appearance of an object is mounted on a table means on which an object to be inspected can be placed and continuously moved in at least one direction. An image acquisition unit that captures an image of the placed inspection target object and acquires the image, a surface displacement information acquisition unit that acquires surface unevenness information of the inspection target object placed on the table unit, and the surface displacement information acquisition Information from the surface irregularities of the object to be inspected acquired by the means, shadow information calculating means for adding a shadow to the surface texture acquired by the image acquiring means, and surface defects of the object to which the shadow calculated by the shadow information calculating means is added And display means for displaying the appearance data including.

  Furthermore, in order to achieve the above object, according to the present invention, in the method for inspecting the appearance of an object, the inspection object is imaged while the inspection object is placed and continuously moved in at least one direction. Acquire the surface unevenness information of the inspection target object while acquiring the image of the inspection target object including the texture information of the surface of the object, and restore the 3D shape of the inspection target object from the acquired surface unevenness information of the inspection target object The appearance information of the object having the surface texture is obtained from the obtained image and the restored three-dimensional shape of the inspection target object, and a plurality of features are extracted from the obtained appearance information, and at least one of the extracted features The appearance of the object to be inspected is evaluated by comparing the feature with the feature corresponding to the at least one feature of the reference data set in advance.

  Furthermore, in order to achieve the above object, according to the present invention, in the method for inspecting the appearance of an object, the object to be inspected is placed and continuously moved in at least one direction, and continuously moved in this one direction. The image of the inspection target object is captured and the image is acquired. The surface unevenness information of the inspection target object continuously moving in this one direction is acquired, and the surface unevenness information of the acquired inspection target object is acquired. A shadow is added to the surface texture of the image of the inspection target object, and appearance data including the surface defect of the inspection target object to which this shadow is added is displayed.

  According to the present invention, the non-defective product standard setting means for creating and holding a good product standard used for the quality determination of the appearance of the inspection target object from an image obtained by imaging the inspection target in advance from design data, By collating non-defective product standards with measured values, quantitative evaluation values for shape defects and minute surface scratches, which are difficult for humans to quantify, can now be performed. In addition, a final visual confirmation can be easily performed by means of estimating a three-dimensional shape shadow from the relationship between the measured surface displacement and the detection optical system, and adding and displaying the shadow on the surface unevenness.

It is a block diagram which shows the schematic structure of the external appearance inspection apparatus in the Example of this invention. It is a block diagram which shows the concept of the system configuration | structure of the external appearance inspection apparatus in this invention. It is a flowchart which shows the flow of a process of the signal processing part in the Example of this invention. It is a flowchart which shows the flow of the connection process performed in the connection process part in the Example of this invention. It is a flowchart which shows the flow of the collation process performed in the defect determination part in the Example of this invention. It is an outline figure of a test object and a good quality reference explaining a concept of defect judgment processing to which a different threshold is applied for each part in an example of the present invention. The outline figure of a test object and a quality standard explaining the concept of the surface defect determination process which applied a different threshold value for every site | part in the Example of this invention. It is a flowchart which shows the flow of a process of another example of the collation process performed in a defect determination part with respect to the test target object with an unevenness | corrugation in the surface in the Example of this invention. It is a block diagram which shows the concept of the position collation performed in the defect determination part in the Example of this invention. It is a flowchart which shows another example of the flow of a process of the signal processing part in the external appearance inspection apparatus of the Example of this invention. It is a perspective view of a test subject in the example of the present invention. It is the shadow image estimation figure estimated from the uneven | corrugated information of the test target object in the Example of invention. 2 is an optical image of an inspection object obtained by optical imaging in an embodiment of the invention. It is a graph which shows the active power sensor signal in processing the sample in the Example of invention with a drill. It is a front view of the drill which shows the state which is processing the sample with the drill in the Example of invention. It is a flowchart which shows the flow of a process when processing the sample in the Example of invention with a drill. It is an example of the inspection object which has a free-form surface, and is a side view of a cutter showing a state where cutting is performed under such a condition that chips are thin. It is an example of the inspection object which has a free-form surface, and is a side view of the cutter showing a state where cutting is performed under conditions such that chips are thick. It is a perspective view of an appearance inspection apparatus and an inspection object which show an example of a use of an appearance inspection apparatus in an example of an invention. It is a flowchart which shows the flow of a process of the external appearance inspection in an example of the use of the external appearance inspection apparatus in the Example of invention.

  Embodiments of an object appearance inspection method and apparatus according to the present invention will be described with reference to the drawings. First, an embodiment of an appearance inspection apparatus that targets a rotationally symmetric object such as a turbine wheel composed of a plurality of wings as an inspection object will be described.

  FIG. 2 is a conceptual diagram showing an embodiment of an appearance inspection apparatus for a rotationally symmetric object according to an embodiment of the present invention. The appearance inspection apparatus in the present embodiment includes a measurement unit 1, an A / D conversion unit 2, a signal processing unit 3, and an overall control unit 9.

  The measurement unit 1 includes one or more illumination units 4a and 4b and one or more detection units 7a, 7b, and 7c. The illumination units 4a and 4b can irradiate the inspection object 5 with light having different illumination conditions (for example, one or more of the irradiation angle, the illumination direction, the illumination wavelength, and the polarization state are different). Reflected light 6a, 6b is generated from the inspection object 5 by the illumination light emitted from each of the illuminating units 4a, 4b, and the generated reflected light 6a, 6b is generated by each of the plurality of detecting units 7a, 7b, 7c. The image detectors 7a and 7b detect the light intensity distribution. Each of the detected light intensity distributions is amplified and A / D converted by the A / D converter 2 and input to the signal processor 3 as an image signal.

The detection unit 7c irradiates the inspection object 5 with a laser, measures the irradiation position on the inspection object 5 and the optical path distance to the detection unit 7c by the principle of triangulation, and outputs a three-dimensional displacement detection unit (laser displacement) Etc.). The optical path distance data obtained by detection by the three-dimensional displacement detector 7c is also A / D converted by the A / D converter 2 together with the light intensity signals obtained by the image detectors 7a and 7b, and input to the signal processor 3. The
The signal processing unit 3 includes a three-dimensional model generation unit 8-1 and a defect determination unit 8-2. First, the 3D model generation unit 8-1 superimposes the 3D shape data of the inspection object 5 and the 2D surface texture data on the corresponding position of the inspection object 5 from the input digital signal. Generate a dimensional model. Then, the defect determination unit 8-2 compares the generated three-dimensional model with a pre-stored three-dimensional good product standard, and detects defective portions such as shape defects, surface micro unevenness, and surface texture defects. And output to the overall control unit 9.
In the configuration shown in FIG. 2, the reflected light 6a, 6b is detected by the separate detectors 7a, 7b, but may be detected by a single detector. Further, the illumination units 4a and 4b and the detection units 7a and 7b are not limited to two, and may be one or three or more. Moreover, the number of the illumination parts 4a and 4b and the number of the detection parts 7a and 7b do not necessarily need to be the same number.

  Each of the reflected light 6a and the reflected light 6b indicates a reflected light distribution generated corresponding to each of the illumination units 4a and 4b. If the optical condition of the illumination light by the illumination unit 4a and the optical condition of the illumination light by the illumination unit 4b are different, the reflected light 6a and the reflected light 6b generated by each differ from each other. In this embodiment, the optical properties and characteristics of the reflected light generated by a certain illumination light are referred to as the reflected light distribution of the reflected light. More specifically, the reflected light distribution refers to a distribution of optical parameter values such as intensity, amplitude, phase, polarization, wavelength, and coherency with respect to the emission position, emission direction, and emission angle of the reflected light.

  Next, FIG. 1 shows a block diagram as an embodiment of a specific appearance inspection apparatus for realizing the configuration shown in FIG. That is, the appearance inspection apparatus according to the present embodiment includes a plurality of illumination units 4a and 4b that irradiate the inspection object 5 (for example, a sample such as a drill blade) with illumination light, and reflection from the inspection object 5. An image detection unit 7a that forms an image of light, an image detection unit 7b that forms an image of reflected light from the inspection object 5 at an angle different from that of the image detection unit 7a, and a laser displacement that measures the distance to the inspection object 5 A three-dimensional displacement detector 7c provided with a meter, an A / D converter 2 for amplifying an optical image obtained from each detector, and A / D converting, a signal processor 3, and an overall controller 9. It is prepared for.

  The inspection object 5 is mounted on a stage (XYZ-θ-R stage) 101 that can move and rotate in the XY plane, move in the Z direction perpendicular to the XY plane, and rotate in the XZ plane. The XYZ-θ-R stage 101 is driven by the mechanical controller 10. At this time, the inspection object 5 is fixedly mounted on the XYZ-θ-R stage 101 with the work chuck 103 and moved in parallel in any of the X, Y, and Z directions while rotating θ, or By detecting the reflected light from the inspection object 5 by rotating it R, the reflected light intensity distribution of the entire surface of the inspection object 5 is obtained as a two-dimensional image (texture information). Obtained as unevenness information. The mechanical controller 10 performs stage control for obtaining the surface texture and unevenness information covering the entire inspection object 5.

Each illumination light source of the illumination units 4a and 4b may use a laser or a lamp. Moreover, the wavelength of the light of each illumination light source may be a short wavelength, or may be light with a broad wavelength (white light). Illumination conditions (for example, irradiation angle, illumination azimuth, illumination wavelength, polarization state, etc.) are selected or automatically selected by the user.
Similarly, in the image detection units 7a and 7b, the user can set detection conditions (for example, a detection azimuth angle and a detection elevation angle).
The image detection units 7a and 7b employ a one-dimensional image sensor (CCD linear sensor) or a CCD area sensor configured by two-dimensionally arranging a plurality of one-dimensional image sensors, and XYZ-θ. -A two-dimensional image is obtained by detecting the reflected light distribution from the object 5 moving at the same speed in the Y-axis direction while rotating R by the image sensor in synchronization with the movement / rotation of the R stage 101. The image detection units 7a and 7b are optical filters 71a and 71b, that is, optical elements capable of adjusting light intensity such as ND filters and attenuators, polarizing optical elements such as polarizing plates, polarizing beam splitters, and wave plates, or bandpass filters. Or any combination of wavelength filters such as dichroic mirrors, or a combination thereof, and controls any one or a combination of the light intensity, polarization characteristics, and wavelength characteristics of the detection light.

  The signal processing unit 3 detects a defect of the inspection object 5, and generates a three-dimensional model of the inspection object 5 from signals input from the image detection units 7a and 7b and the three-dimensional displacement detection unit 7c. By comparing the generated three-dimensional model with the non-defective product standard input in advance, the three-dimensional model generation unit 8-1 detects a shape defect, a surface minute unevenness, a surface texture defect, and the like, and outputs them to the overall control unit 9. It is comprised by the defect determination part 8-2. In the signal processing unit 3, for example, the defect determination unit 8-2 is configured by connecting a database.

  More specifically, the three-dimensional model generation unit 8-1 includes an image integration processing unit 301 that integrates image signals input from the image detection units 7a and 7b, and a three-dimensional shape measurement input from the three-dimensional displacement detection unit 7c. A three-dimensional shape restoration processing unit 302 that restores a three-dimensional shape (three-dimensional shape) from data, and an image integrated by the image integration processing unit 301 and a three-dimensional shape restored by the three-dimensional shape restoration processing unit 302 are connected. And a connection processing unit 303 that superimposes and generates three-dimensional shape data. In addition, the defect determination unit 8-2 includes a matching processing unit 304 that collates the three-dimensional shape data connected by the connection processing unit 303 with the good product reference data stored in the storage device 12 and extracts defects. Yes.

  The overall control unit 9 includes a CPU (incorporated in the overall control unit 9) that performs various controls, accepts parameters from the user, and displays the generated three-dimensional model of the object to be inspected, detected defective portions, and the like. A user interface unit (GUI unit) 11 having a display unit and an input unit and a storage device 12 for storing a model processed by the signal processing unit 3, a good product reference model, and the like are appropriately connected. Further, the signal processing unit 3, the illumination units 4 a and 4 b, the image detection units 7 a and 7 b, the three-dimensional displacement detection unit 7 c, the mechanical controller 10, and the like are also driven by commands from the overall control unit 9.

  Here, the inspection object 5 has a rotationally symmetric shape (a certain axial symmetry and is composed of parts having the same shape. In the configuration shown in FIG. 1, the inspection object 5 is a rotational object with respect to the Y axis). The overall control unit 9 rotates the work chuck 103 around the Y axis (R rotation) by the R axis rotation drive unit 102 of the XYZ-θ-R stage 101 for the inspection object 5 XYZ-. It was obtained by continuously moving the surface texture in the Y-axis direction by the θ-R stage 101 and synchronizing the surface texture from the image detectors 7a and 7b and the surface displacement from the three-dimensional displacement detector 7c. An appearance defect is detected from the images of the two types of reflected light (6a, 6b) and the three-dimensional shape measurement data.

  Next, an example of a processing flow in the signal processing unit 3 will be described. FIG. 3 illustrates the image detection units 7 a and 7 b and the three-dimensional image by scanning the inspection target 5 in the Y-axis direction while rotating the inspection target 5 R on the XYZ-θ-R stage 101. A basic process of the signal processing unit 3 that detects a defective portion by using a digital signal (image 31, image 32, surface displacement 33) obtained from the displacement detection unit 7c as an input is shown. Of the input images 31 and 32 and the surface displacement 33, the image 31 and the image 32 are captured with one or more different illumination conditions and one or more different azimuth and elevation angle combinations. It is an image. For this reason, the image 31 and the image 32 are different in the degree of emphasis, that is, the appearance of minute scratches (irregularities), stains and dirt (textures) on the surface of the inspection object 5. Therefore, first, in the three-dimensional model generation unit 8-1, the obtained two images 31 and 32 are integrated by the image integration processing unit 301.

There are various integration methods. For example, (1) average, (2) logical product, (3) logical sum, and (4) brightness of the pixel with the larger standard deviation in the vicinity region. is there. Assuming that the brightness of each corresponding pixel (x, y) of the images 31 and 32 is f (x, y) and g (x, y), the integrated image h (x, y) Calculated.
Average: h (x, y) = {f (x, y) + g (x, y)} / 2 (Expression 1)
Logical product: h (x, y) = minimun {f (x, y), g (x, y)} (Expression 2)
OR: h (x, y) = maximun {f (x, y), g (x, y)} (Equation 3)
Standard deviation: if (σ _f > σ _g ) h (x, y) = f (x, y) (Expression 4)
else h (x, y) = g (x, y)
However, σ _f = [Σ {f (x + i, y + j) ² } − {Σf (x + i, y + j)} ² / (N × N)] / (N × N−1) i, j = 0, 1 ..N-1 (same for σ _g )
In addition to the above, the integration method can perform various calculations depending on how noise is generated. Naturally, without performing the integration process, only one of the images 31 and 32 may be used for the subsequent processing, or three or more images may be integrated.

On the other hand, the three-dimensional shape restoration processing unit 302 of the signal processing unit 3 uses the three-dimensional shape measurement data (surface displacement 33) obtained from the three-dimensional displacement detection unit 7c and the stage controller 10 input via the overall control unit 9. The three-dimensional shape (three-dimensional shape) 35 of the inspection region of the inspection object 5 is restored using rotation information obtained by driving the R-axis rotation driving unit 102 to rotate the inspection object 5 around the Y axis (R rotation). Next, the integrated processing unit 303 connects the integrated image 34 generated by the image integration processing unit 301 and the three-dimensional shape 35 of the inspection area of the inspection object 5 restored by the three-dimensional shape restoration processing unit 302. Thereby, the three-dimensional model 36 is generated from the digital signal of the inspection object 5.

  Next, in the defect determination unit 8-2, the generated three-dimensional model 36 and the non-defective product standard 37 previously stored in the storage device 12 are read out and collated in the collation unit 304, and a portion having a large difference from the non-defective product standard 37 is detected. Extract as a defective part.

  FIG. 4 is an example of the flow of processing performed by the connection processing unit 303. For an object having two narrow grooves, 40 is data of the three-dimensional shape (corresponding to the three-dimensional shape 35 in FIG. 3) of the inspection region of the inspection object 5 restored by the three-dimensional shape restoration processing unit 302. Reference numeral 41 denotes an integrated image of the surface texture generated by the image integration processing unit 301 (corresponding to the integrated image 34 in FIG. 3). Corresponding positions between the data to be connected are known from the positional relationship and resolution of the detectors 71a, 71b, 71c in FIG. 1, but in reality, there is a possibility that the rotational eccentricity of the inspection object 5 may occur. is there. Further, fluctuations (yawing and / or pitching) of the XYZ-θ-R stage 101 may occur, and a deviation occurs with respect to a known corresponding position at the stage of acquiring data. there is a possibility. For this reason, in the image integration processing unit 301, position matching is performed between the three-dimensional shape data 40 and the integrated image 41 to perform precise alignment.

  A description will be given along the flow of processing shown in FIG. First, one or more feature quantities for position matching are calculated from the three-dimensional shape data 40 and the integrated image 41 (S401, S402). The feature amount calculation method may be the same or different in S401 and S402. As an example, S401 is characterized by the amount of change of the z value at each (x, y) coordinate position with respect to the vicinity (x, y) position, and in S402, the brightness at the nearby position in the xy coordinates. It is characterized by the differential value of.

  Next, the degree of coincidence between the three-dimensional shape data 40 and the integrated image 41 is evaluated for the feature quantity at the calculated xy coordinate position while shifting one data in each of the x direction and the y direction. Each shift amount when the value becomes the highest is set as a shift amount between the three-dimensional shape data 40 and the integrated image 41 (S403). A typical evaluation value for the degree of coincidence is normalized correlation. Then, the calculated deviation of the x and y coordinates is corrected, and the data is superimposed (S404). Reference numeral 42 denotes an example of a three-dimensional model (corresponding to the three-dimensional model 35 in FIG. 3) generated by superimposing the three-dimensional shape data 40 and the integrated image 41. The three-dimensional model 42 is generated by adding height information of the three-dimensional shape data 40 to the integrated image 41 having surface texture information.

  FIG. 5 is an example of the flow of collation processing performed by the collation processing unit 304 inside the defect determination unit 8-2. The three-dimensional model 42 is a three-dimensional model (corresponding to the three-dimensional model 35 in FIG. 3) of the inspection object 5 generated by the three-dimensional model generation unit 8-1 described in FIG. 4, and 50 is input from the storage device 12 in advance. This is a finished good standard (corresponding to the good standard 36 in FIG. 3). Here, the non-defective product standard 50 is design data.

  First, feature quantities at the three-dimensional coordinate positions of the three-dimensional model 42 and the non-defective product reference 50 of the inspection object 5 are calculated (S501, S502). Examples of the feature amount include a gradient with respect to a position near the surface displacement and a curvature of the surface. Then, the degree of coincidence between the feature amounts of the three-dimensional model 42 and the non-defective product standard 50 calculated from S501 and S502 is evaluated while shifting one data in the x, y, and z directions, respectively. The amount of shift in each of the x, y, and z directions when becomes the highest is taken as the amount of shift in the x, y, and z directions between the three-dimensional model 42 and the non-defective product reference 50 (S503). A typical evaluation value for the degree of coincidence is normalized correlation. Then, the deviation of the calculated x, y, and z coordinates is corrected, and the data at the corresponding coordinate points of the three-dimensional model 42 and the non-defective product reference 50 is superposed (S504).

  Next, a difference in contour coordinates between the superimposed three-dimensional model 42 and the non-defective product standard 50 and a difference in brightness between corresponding contour positions are calculated (S505). Then, the calculated difference value is compared with a preset threshold value, and a point where the difference in the coordinate position of the contour is larger than the threshold value is determined as a shape defect, and a point where the brightness difference is larger than the threshold value. Defect determination is performed to determine surface defects such as stains, dirt, and minute scratches (S506). The result of the defect determination (S506) is sent to the overall control unit 9.

  Here, the range that is permitted as normal or the reference that is detected as a defect may differ depending on the part of the inspection target object. For this reason, the threshold value can be set differently depending on the location. FIG. 6 is an example of a conceptual diagram of a defect determination process in which different threshold values are applied to each part. For simplicity, an example in which a shape defect is detected from a two-dimensional model is shown.

  In FIG. 6, 61 is a diagram showing the contour of the inspection object, and 60 is a diagram showing a non-defective standard contour such as the design data. 62 is a result of collating the positions of the contour diagram 61 of the inspection object and the contour diagram 60 of the non-defective product reference and superimposing the inspection target 61 and the good product reference 60 on the two-dimensional coordinates. The broken line 621 is the outline of the external shape of the non-defective product standard 60, and the solid line 622 is the outline of the external shape of the inspection object 61. 63, 64, and 65 are locations where the deviation of the coordinate position of the outline is large. 66 is a threshold line diagram showing the threshold width with respect to the non-defective product standard 60, the broken line 67 is the outline of the external shape of the good product standard 60, the solid line 68 on the outside and the solid line 69 on the inside are the threshold values, A portion where the contour 61 of the object is outside the solid line 68 or inside the solid line 69 is determined to be normal if it is defective in shape and between the solid lines 68 and 69. In the example of the threshold diagram 66, the normal range is narrow in the acute angle portions (for example, 661, 662, 663, etc.), and even a slight contour shift is detected as a shape defect. Become. When the points 63, 64, and 65 where the deviation of the outline is large are compared with the threshold diagram 66, 63 and 65 at the acute angle portion where the normal allowable range is narrow are detected as defects, and 64 does not become a defect. .

  Similarly, the threshold for detecting a surface defect can be set for each part. FIG. 7 is an example of a conceptual diagram of defect determination processing in which different threshold values are applied to each part. For simplicity, an example in which a surface defect is detected from a two-dimensional model is shown. 71 is a view showing the surface state of the inspection object, and 70 is a view showing the surface state of the non-defective product standard. 71 representing the surface state of the inspection object is surface texture information obtained from the integrated image 34, and 70 representing the surface state based on the non-defective product is surface texture information having no defect. 72 indicates the surface state of the inspection target object 71 and the position 70 of the non-defective product reference surface state 70 are collated, and the two-dimensional coordinates 71 indicate the surface state of the inspection target object and the non-defective product standard surface state 70. It is a figure which shows the result of having calculated the difference of the texture information (for example, the brightness of each pixel). In 72 representing the result of calculating the difference of the texture information, 73 and 74 are portions where the difference of the texture information is large, and 74 is larger than 73. 75 is a threshold value diagram showing the threshold width with respect to 70 representing the surface state of the non-defective product, and the threshold values of the regions 76, 77, 78, 79 having sharp angles indicated by the stripe hatching are near the center. The state is lower than the threshold value of 750.

Then, in FIG. 72 showing the difference between the texture information of 71 representing the surface state of the inspection object and 70 representing the surface state of the non-defective product, the portion above the threshold shown in the threshold diagram 75 is stained and stained. Surface defects such as minute flaws, and portions smaller than the threshold value are determined to be normal (Equation 5).
if f (x, y) ≧ th (x, y) then defect ・ ・ ・ （Equation 5）
if f (x, y) <th (x, y) then normal
f (x, y) is the difference in texture information for each coordinate
th (x, y) is the threshold value of each coordinate
That is, in the example of the threshold diagram 75, even if the texture information difference is slight, it is detected as a surface defect. When the areas 73 and 74 having a large difference in the texture information difference 72 are compared with the threshold diagram 75, the area 73 in the acute angle portion having a low normal tolerance is detected as a defect, and the vicinity of the center portion. This region 74 does not become a defect.

  Threshold values that differ from region to region, such as threshold diagram 66 and threshold diagram 75, can be set manually by the user in advance. It can also be generated and set. It can also be set statistically based on past appearance inspection data. For example, an area where important defects have frequently occurred in the past and an area where noise that is unnecessary for detection is likely to occur are known if past data are aggregated. Therefore, the probability of occurrence of important defects and noise is calculated, and accordingly, for example, on the threshold value diagram 66 or threshold value diagram 75, the region where the probability of occurrence of important defects is high lowers the threshold value. In a region where the probability of occurrence of noise is high, the threshold value can be set automatically. The minimum unit of threshold regions that can be set individually is a pixel.

  In the above description, the defect determination process S506 shown in FIG. 5 indicates that the shape defect is extracted independently from the outline of the outer shape (that is, three-dimensional displacement), and the surface defect is extracted independently from the image (that is, two-dimensional texture). However, it is also possible to calculate at the same time. FIG. 8 shows an example of the processing flow.

  In FIG. 8, reference numeral 80 denotes a surface texture image (corresponding to the integrated image 34 generated by the integrated image processing unit 301 in FIG. 3) of a rotationally symmetric inspection object having an uneven surface. Is a developed image generated by imaging the surface of each of the small areas with the image detection unit 7a or 7b and connecting them.

  In the present embodiment, a development view of the inspection object 5 such as the surface texture image 80 can be generated and held. 81 is a diagram showing the three-dimensional shape of the inspection object 5 restored by the three-dimensional shape restoration processing unit 302 from the unevenness information on the surface of the inspection object 5 detected by the three-dimensional displacement detection unit 7c. 82, the connection processing unit 303 shown in FIG. 1 collates the positions of the surface texture image 80 and FIG. 81 showing the three-dimensional shape, and applies the development image (x, y) coordinates of the integrated image of the surface texture image 80. 81 is a three-dimensional development model to which the height information z-coordinate of FIG. 81 showing the three-dimensional shape of the inspection object 5 is added. This corresponds to the three-dimensional model 35 of FIG.

  On the other hand, a non-defective product reference 83 that is a non-defective three-dimensional developed model corresponding to the three-dimensional developed model 82 is created in advance and stored in the storage device 12. On the other hand, as in the example shown in FIG. 5, the matching processing unit 304 extracts feature amounts for each of the three-dimensional model 82 and the non-defective product reference 83 (S801, S802). The degree of coincidence is evaluated (S803), and the positions of the three-dimensional model 82 and the good product reference 83 are collated (S804).

Next, in the processing example of FIG. 8, multidimensional vectorization is performed for each point (S805). This calculates one or a plurality of texture feature quantities for each pixel at the (x, y) coordinates. The feature amount of the texture only needs to represent the feature of the pixel. For example, the brightness of the integrated image itself (f (x, y)), contrast (Equation 6), brightness dispersion value of neighboring pixels (Equation 7), increase / decrease in brightness with neighboring pixels (Equation 8) There are secondary differential values and the like.
Contrast: max {f (x, y), f (x + 1, y), f (x, y + 1), f (x + 1, y + 1)}
−min {f (x, y), f (x + 1, y), f (x, y + 1), f (x + 1, y + 1) (Equation 6) variance; [Σ { f (x + i, y + j) ² } − {Σf (x + i, y + j)} ² / M] / (M−1)
i, j = -1, 0, 1 M = 9 (Expression 7)
Luminance increase / decrease distribution (x direction);
If (f (x + i, y + j) −f (x + i + 1, y + j)> 0) then g (x + i, y + j) = 1
else g (x + i, y + j) = 0 (Equation 8)
Then, the calculated one or more features, f1 (x, y), f2 (x, y), f3 (x, y)... And the z coordinate value (displacement information), and the threshold value at that point Th (x, y) is vectorized as a feature of the pixel at the (x, y) coordinate position (Equation 9).
Feature vector V of coordinates (x, y); V = (x, y, z, f1, f2,..., Th) (Equation 9)
Then, the value of each (x, y) coordinate is plotted in a multidimensional feature space having each feature vector element as a feature axis. The non-defective product 83 is also plotted in the same manner. Then, the distribution of the feature vector of the non-defective product standard 83 plotted in the multidimensional feature space is set as a normal part distribution, and a point whose distance from the normal part distribution is larger than the threshold value Th (84) is detected as a defect ( S806). Various classifiers are prepared for classification (defect / normal two-class classification) based on multidimensional feature values. Typical examples include SVM (support vector machine), k nearest neighbor classifier, Bayes classifier, and the like.

  In order to detect a minute and weak signal defect by collation with the non-defective standard described above, it is necessary to accurately collate the non-defective standard with the position of the inspection object. An example of position verification will be described with reference to FIG.

  90 in FIG. 9 is a partial profile of the surface of the inspection object, and 91 is a partial profile of the corresponding non-defective standard surface. Regarding the collation of the positions of the inspection object surface profile 90 and the non-defective product reference surface profile 91, in this example, the Z-value cross-sectional profile in the Y direction (between A and B in the figure) at each X coordinate of the inspection object surface profile 90. Z value) is compared and scanned while scanning the Z value cross-sectional profile in the Y direction (Z value between A ′ and B ′ in the figure) of the non-defective reference surface profile 91 in the X axis direction. This is performed at each X coordinate of the inspection object surface profile 90 (A1-B1, A2-B2, A3-B3,... Of the Z-value cross-sectional profile 92 in the Y direction). Then, the collation results of both profiles at each X coordinate are tabulated (for example, a majority decision is taken), and the amount of displacement of the position in the X direction with respect to the non-defective product reference surface profile 91 of the inspection object surface profile 90 is calculated. The deviation in the Y direction and the deviation in the Z direction are calculated in the same manner, and the deviation amounts in the x, y, and z directions are determined.

  The non-defective product standard used for comparison is design data, and it has been shown in the above examples that it is set in advance, but it can also be generated from an actual object. FIG. 10 shows a configuration of a signal processing unit 310 having a function of generating a non-defective product standard from an actual object, instead of the signal processing unit 3 using design data as the non-defective product standard described in FIG. The signal processing unit 310 performs digital signals (images 31 and 7) obtained from the image detection units 7 a and 7 b and the three-dimensional displacement detection unit 7 c on the inspection target 5 by scanning the XYZ-θ-R stage 101. Using the image 32 and the displacement 33) as input, a defective portion is detected.

  The signal processing unit 310 generates a three-dimensional model of the inspection object 5 from the signals input from the image detection units 7a and 7b and the three-dimensional displacement detection unit 7c, and generates a non-defective product reference 100. 1 and a defect determination unit 80-2 that detects a shape defect, surface minute unevenness, surface texture defect, and the like by comparing the generated three-dimensional model with a good product standard and outputs the defect to the overall control unit 9. The

  More specifically, the three-dimensional model generation unit 80-1 integrates the images 31 and 32 input from the image detection units 7a and 7b to create an integrated image 34, and a three-dimensional displacement detection. A 3D shape restoration processing unit 312 that restores a 3D shape (three-dimensional shape) from the 3D shape measurement data input from the unit 7 c, and an image integrated by the image integration processing unit 311 and a 3D shape restoration processing unit 312. The non-defective product generation for generating the non-defective product reference 100 using the connection processing unit 313 that connects and superimposes the generated three-dimensional shapes to create the three-dimensional model 36, and the integrated image 34 and the restored three-dimensional shape data 35. And a processing unit 1001. Further, the defect determination unit 80-2 includes a matching processing unit 314 that collates the three-dimensional model 36 connected by the connection processing unit 313 with the good product standard 100 generated by the good product generation processing unit 1001 and extracts defects. ing.

  In the signal processing unit 310, the flow of processing of the signal processing unit that generates a non-defective product standard from the inspection target will be described with reference to FIG.

  Of the images 31 and 32 and the surface displacement 33 input to the signal processing unit 310, the image 31 and the image 32 have one or more different illumination conditions and one or more different azimuth and elevation angles. It is the image imaged by the combination of. Similar to the example shown in FIG. 3, first, in the three-dimensional model generation unit 80-1, the obtained two images 31 and 32 are integrated by the image integration processing unit 311. There are various integration methods. For example, (1) average, (2) logical product, (3) logical sum, and (4) brightness of the pixel with the larger standard deviation in the vicinity region. is there. The integration method is as shown in (Expression 1) to (Expression 4).

  On the other hand, the three-dimensional shape restoration processing unit 312 restores the three-dimensional shape (three-dimensional shape) 35 from the three-dimensional shape measurement data input from the three-dimensional displacement detection unit 7c.

  Next, the integrated image 34 generated by the image integration processing unit 311 and the three-dimensional shape (three-dimensional shape) 35 restored by the three-dimensional shape restoration processing unit 312 are connected by the connection processing unit 313. Thereby, the three-dimensional model 36 is generated from the digital signal of the inspection object 5.

  In the present embodiment, the non-defective product standard 100 is generated by the non-defective product generation unit 1001 from the integrated image 34 and the three-dimensional shape (three-dimensional shape) 35 restored by the three-dimensional shape restoration processing unit 312 simultaneously with the generation of the three-dimensional model 36. Then, in the defect determination unit 80-2, the generated three-dimensional model 36 and the good product reference 100 are checked in the check unit 314, and a portion having a large difference from the good product reference 100 is extracted as a defective portion.

  In the non-defective product standard 100, if the object to be inspected is a rotationally symmetric object composed of a plurality of parts having the same shape, the parts made to have the same shape while rotating the object by the period of the symmetric part are the same. Images of multiple parts by taking images with the same field of view, the same illumination angle, and the same detection angle, and the average of the brightness of the corresponding positions or the median value of the brightness between the obtained images Generate by taking.

  As described above, the example of the form of the specific appearance inspection apparatus that realizes the configuration shown in FIG. 2 has been described. However, the measurement results and inspection results of the appearance by the appearance inspection apparatus according to the present invention are shown in FIG. Displayed on a user interface unit (GUI unit) 11 having display means and input means. For example, the surface unevenness information (surface displacement 33) obtained by the three-dimensional displacement detector 7c is only a measurement point group. For this reason, the visual inspection apparatus according to the present invention presents the measurement point group to the user more visually. Examples thereof are shown in FIGS. 11A to 11C.

  In FIG. 11A, 110 is an inspection object. If the illumination angle (112 in the figure) of the illumination light 111 when acquiring the surface texture, the angle of the detection unit (114 in the figure) for detecting the reflected light 113, etc. are known with respect to the surface irregularities of the inspection object 110 From the unevenness information (surface displacement 33), it is possible to estimate how the shadow is added. In this embodiment, the shadow is estimated from the unevenness information (surface displacement 33) and displayed on the GUI unit 11 as a two-dimensional image.

  11B of FIG. 11B is an example of a shadow image estimated from the unevenness information (surface displacement 33) of the inspection object 110. In addition to the display, the shadow image estimation function based on the unevenness information (surface displacement 33) enables a more sensitive inspection. For example, by comparing the actual surface texture information detected by the image detection unit with the shadow image estimated from the unevenness information, the specular reflection component that becomes noise and the shadow due to the shielding of parts are excluded from the actual surface texture information It becomes possible to do. This facilitates discrimination between noise and weak surface defects.

  Reference numeral 116 in FIG. 11C denotes an optical image detected by the image detection unit 7a or 7b. Although the shape defect on the surface of the inspection object 110 can be displayed in the shaded image of FIG. 11A or FIG. 11B, it is difficult to display non-shape defects such as discoloration and stains on the surface of the inspection object 110. . On the other hand, by displaying an optical image as shown in FIG. 11C, it is possible to display non-shape defects 117 such as discoloration and stains on the surface of the inspection object 110. Therefore, by displaying the optical image as shown in FIG. 11C side by side with the shadow image as shown in FIG. 11A or FIG. 11B, the shape defect on the surface of the inspection object 110 and the non-shape defect such as discoloration or stain on the surface can be removed. It becomes possible to display simultaneously.

  The appearance inspection apparatus described above measures surface texture information and surface displacement information obtained from the detection unit. In particular, by quantitatively measuring the cracks, wear, and chipping of cutting tools such as drills and cutters, it is possible to manage the tools and estimate the quality of the processing target. When the target is a tool for machining, it is possible to efficiently perform inspection in cooperation with sensor time-series data at the time of machining. Examples thereof are shown in FIGS. 12A to 12C.

  In FIG. 12A, a graph 120 is an active power sensor signal detected by a sensor (not shown) during processing of the workpiece 123 by the drill 122 shown in FIG. 12B. In the graph 120 of FIG. 12A, sensor signals for three times (t1, t2, t3) are plotted in time series. The horizontal axis is the time during processing. A signal portion at t3 surrounded by an ellipse 121 indicates an active power sensor signal when processing failure occurs. This indicates that a processing failure has occurred at t3.

  In the appearance inspection apparatus according to the present embodiment, an inspection is performed in conjunction with an abnormality sign by monitoring sensor data such as the processing state. The flow of this inspection is shown in FIG. 12C.

In the processing flow shown in FIG. 12C, sensor signals such as active power values are monitored while a tool to be inspected such as a drill is in operation (S1201). A sign of abnormality is diagnosed from the data obtained by this monitoring (S1202), and if there is no abnormality, the inspection object is in a standby state as soon as the processing is completed (S1203). When there is a sign of abnormality in the sensor signal, the appearance inspection according to the embodiment described with reference to FIGS. 1 to 10 is executed using the sensor signal as a trigger (S1204). As a result, the frequency of inspection can be reduced.
Furthermore, the object of the appearance inspection apparatus is not limited to a rotationally symmetric shape, but includes an object having a free-form surface. Examples thereof are shown in FIGS. 13A and 13B. 13A in FIG. 13A and 131 in FIG. 13B are cutting cutters, each having a different shape. 13A in FIG. 13A and 133 in FIG. 13B are chips generated after cutting, and the shapes are different as in the case of the chips 132 in FIG. 13A and the chips 133 in FIG. 13B according to the difference in the shape of the cutter. In this visual inspection, the shape of the cutters 130 and 131 and the shape of the chips 132 and 133 to be processed are measured, and compared with each other to optimize the cutter shape to achieve a more ideal workmanship. Is possible.
14A and 14B show another example of the use of the appearance inspection apparatus. 140 of FIG. 14A is the appearance inspection apparatus described with reference to FIGS. 1 and 2, and a roll-shaped thin plate 141 is an inspection target. Here, the surface texture (image 142) on the entire surface of the thin plate is sequentially detected by the image detection unit described with reference to FIGS. 1 and 2 and is input to the signal processing unit 3.

FIG. 14B illustrates a processing procedure in the signal processing unit 3. The signal processing unit 3 detects a defect from the image 142 (S1401). Thereby, when there is a defect on the thin plate, the position coordinate 143 of the defect is detected. Defect detection may be performed by a general defect inspection method such as simple brightness binarization processing or comparison with surrounding brightness. Next, the unevenness near the position coordinates of the defect is detected by the three-dimensional displacement detector (S1402). Then, based on the detected displacement 144 of the defective portion, it is determined whether or not the defect is fatal (S1403), and the defect type is further specified (S1404). As described above, in this visual inspection apparatus, it is determined only by the surface texture whether the defects on the thin plate can be eliminated by a subsequent cleaning process such as a stain or dirt on the surface, or cannot be excluded such as a hole. Defects that cannot be determined can also be determined using 3D displacement information.
As described above, according to the present invention, it becomes possible to quantitatively evaluate the shape defect and the level of fine scratches that could not be quantified by direct visual inspection by a person such as an inspector. As a result, it is possible to evaluate the quality of processed products, quality control of processed tools, and determine the criticality of defects. This achieves product stabilization. In addition, it is possible to prevent quality defects by interlocking with simulation results.

  As described above, the embodiment of the present invention has been described by taking the appearance inspection of an object having a complicated shape such as a machining tool as an example. However, the present invention can also be applied to similarity determination of an object having a free-form surface. One of them is the different identification of bullets fired from handguns.

DESCRIPTION OF SYMBOLS 1 ... Measurement part 2 ... A / D conversion part 3,310 ... Signal processing part 4a, 4b ... Illumination part 5 ... Test object 7a, 7b ... Image detection part 7c ... Three-dimensional displacement detection part
8-1, 80-1 ... 3D model generation unit 8-2, 80-2 ... Defect determination unit
DESCRIPTION OF SYMBOLS 9 ... Overall control part 10 ... Mechanical controller 11 ... User interface part 12 ... Memory | storage device 101 ... XYZ- (theta) -R stage 301 ... Image integration process part 302 ... Three-dimensional shape restoration process part 303 ... Connection process part 304 ... Verification processing section

Claims (14)

A device for inspecting the appearance of an object,
Table means for placing the object to be inspected and continuously moving in at least one direction;
Image acquiring means for capturing an image of the inspection target object including the texture information of the surface of the inspection target object by imaging the inspection target object placed on the table means;
Surface displacement information acquisition means for acquiring surface irregularity information of the inspection object placed on the table means;
A solid shape restoring means for restoring the solid shape of the object to be inspected from the surface unevenness information of the object acquired by the surface displacement information acquiring means;
Appearance data generation means for obtaining appearance information of an object having a surface texture from the image acquired by the image acquisition means and the three-dimensional shape of the inspection target object restored by the three-dimensional shape restoration means;
Feature extraction means for extracting a plurality of features from the appearance information obtained by the appearance data generation means;
Appearance evaluation means for evaluating the appearance of an object to be inspected by comparing at least one of the plurality of features extracted by the feature extraction means with a feature corresponding to the at least one feature of preset reference data When,
An apparatus for inspecting the appearance of an object. The appearance data generation means includes the information on the three-dimensional shape of the inspection target object restored by the three-dimensional shape restoration means and the surface texture information of the inspection target object obtained from the image obtained by the image acquisition means. The position information is integrated as vector information to obtain the appearance information,
The feature extraction unit integrates the vector information integrated by the appearance data generation unit and the vector information of the reference data on a feature space,
2. The object appearance inspection apparatus according to claim 1, wherein the appearance evaluation unit evaluates an outlier value on the feature space integrated by the feature extraction unit as an object defect.   The object visual inspection apparatus according to claim 1, wherein the reference data is generated from design data.   3. The object appearance inspection apparatus according to claim 1, wherein the reference data is generated from an image of the inspection object acquired by the image acquisition unit. The reference data is the three-dimensional shape restoring means based on an image obtained by imaging an object that should originally have the same surface displacement and texture by the image obtaining means and surface unevenness information obtained from the object by the surface displacement information obtaining means. 3. The appearance inspection of the object according to claim 1, wherein the appearance information of the object has surface texture information generated by the appearance data generation unit using the three-dimensional shape of the object restored in step 3. apparatus.   The appearance evaluation unit compares at least one feature among the plurality of features extracted by the feature extraction unit with a feature corresponding to the at least one feature of the reference data, and a feature of the object to be inspected. 3. The object appearance inspection apparatus according to claim 1, wherein whether or not the object to be inspected is similar to the reference data is determined based on a degree of coincidence of features of reference data. A device for inspecting the appearance of an object,
Table means for placing the object to be inspected and continuously moving in at least one direction;
Image acquisition means for imaging the inspection object placed on the table means and acquiring the image;
Surface displacement information acquisition means for acquiring surface irregularity information of the inspection object placed on the table means;
Shadow information calculation means for adding a shadow to the surface texture acquired by the image acquisition means, information from the surface unevenness of the inspection object acquired by the surface displacement information acquisition means;
Display means for displaying appearance data including surface defects of the object to which the shadow calculated by the shadow information calculating means is added;
An apparatus for inspecting the appearance of an object. A method for inspecting the appearance of an object,
While inspecting the object to be inspected while placing the object to be inspected and continuously moving in at least one direction, acquiring the image of the object to be inspected including the texture information of the surface of the object to be inspected. Get surface roughness information of the object,
The three-dimensional shape of the inspection target object is restored from the obtained surface unevenness information of the inspection target object,
Obtaining appearance information of an object having a surface texture from the acquired image and the restored three-dimensional shape of the inspection target object,
Extracting a plurality of features from the obtained appearance information;
Appearance of an object characterized by evaluating the appearance of an object to be inspected by comparing at least one of the extracted features with a feature corresponding to the at least one feature of preset reference data Inspection method. Obtaining the appearance information integrates the restored three-dimensional information of the inspection target object and the surface texture information of the inspection target object obtained from the acquired image as vector information at each position of the inspection target object. Obtained by
Extracting the feature integrates the integrated vector information and the vector information of the reference data on a feature space;
9. The object appearance inspection method according to claim 8, wherein the evaluation of the appearance includes evaluating an outlier value on the integrated feature space as a defect of the object.   10. The object visual inspection method according to claim 8, wherein the reference data is generated from design data.   The object reference inspection method according to claim 8 or 9, wherein the reference data is generated from the acquired image of the object to be inspected.   The reference data is surface texture information generated using an image obtained by imaging an object that should originally have the same surface displacement and texture, and the three-dimensional shape of the object restored from the surface unevenness information acquired from the object. 10. The method for inspecting the appearance of an object according to claim 8 or 9, characterized in that the information is appearance information of the object having a point. The evaluation of the appearance is performed by comparing at least one of the extracted plurality of features with a feature corresponding to the at least one feature of the reference data, and the feature of the object to be inspected and the reference data 10. The object appearance inspection method according to claim 8, wherein the object appearance inspection method is performed by determining whether or not the object to be inspected is similar to the reference data based on the degree of coincidence of the features. A method for inspecting the appearance of an object,
Place the object to be inspected and move it continuously in at least one direction,
Image the object to be inspected that is continuously moving in the one direction to obtain the image;
Obtaining surface roughness information of the object to be inspected that is continuously moving in the one direction;
A shadow is added to the surface texture of the acquired image of the inspection object from the acquired surface unevenness information of the inspection object,
An appearance inspection method for an object, characterized by displaying appearance data including a surface defect of the inspection target object to which the shadow is added.

Images