JP3055322B2 - Circular container inner surface inspection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention inspects the inner surface of a circular container such as a beer can conveyed by a conveyor or the like, and inspects the inner surface of the circular container as an image processing device for detecting deformation, foreign matter, dust, scratches, etc. of the container. Related to the device. In the drawings, the same reference numerals indicate the same or corresponding parts.

[0002]

2. Description of the Related Art FIG. 2 is an explanatory view of a high-brightness area when an aluminum can container for beer is observed from above as an example. FIG. 2A is a top (image) view of the can container and FIG. () Is a side sectional view. And 102 is a container, 101 is this container 10
A ring-shaped illuminator illuminating 2 from above, 103 is a high-brightness portion at the mouth, and 104 is a high-brightness portion at the bottom. By irradiating the inner surface of the container 102 with the light beam using the ring illuminator 101 in this manner, the inner and outer surfaces of the container are 10
High luminance portions such as 3,104 are generated. This is particularly noticeable when the inner surface of the container has a metallic luster such as a can.

FIG. 3B is a top view image of the container 102 (FIG. 3).
FIG. 4A shows a density change on the scanning line Q-Q1 with respect to (A)), and is classified into five regions W1 to W5 according to the characteristics of the density change. The first area W1 is the high-brightness area 10 of the mouth.
3, the second area W2 is the upper middle part of the side of the container where the change in density is relatively small, and the third area W3 is a container that is darker than the other areas because light rays by the illumination 101 described in FIG. The fourth region W4 is the bottom high brightness part 104, and the fifth region W5 is the bottom.

Conventionally, a window is provided in each of these areas W1 to W5, and a threshold value for detecting a defect of black stain (black point) or white stain (white point) is set according to the optical characteristics of the area. Was. As a method of detecting a defect, for example, a multi-valued grayscale image signal such as 8 bits obtained by A / D conversion of an analog video signal (analog grayscale image signal) obtained by scanning a target image is converted to a predetermined threshold value. There are known a binarization method, a differentiation method of differentiating the video signal to extract a defect signal, and the like. In this differentiation method,
A differential signal is also output at the contour of the outer shape of the object, but either the positive pulse or the negative pulse is generated by the differentiation at the contour, while the positive and negative pulses are simultaneously generated at the minute defect. The defective part can be extracted by utilizing this.

That is, a value P (i, j) at a point of interest (coordinate values X = i, Y = j) for a signal P (X, Y) obtained by differentiating an analog grayscale image signal based on raster scanning,
The value P (i−p) at a point separated by a predetermined minute α pixel and β pixel respectively before and after the point of interest on the scanning line in the X direction.
α, j) and P (i + β, j), P (i, j) −P (i−α, j)> TH1 and P (i + β, j) −P (i, j) )> TH1 (where TH1 is a predetermined threshold value (positive value)) A binarization function value PD (i, j) for detecting a defect at a point of interest
= 1, this point of interest is regarded as a black level defect point, otherwise PD (i, j) = 0, and this point of interest is regarded as a normal point.

[0006]

However, the above-described method is advantageous for defects such as minute black spots, but in the case of dents in a container, local contrast cannot be obtained and detection is difficult. Therefore, an object of the present invention is to provide a circular container inner surface inspection apparatus that can solve this problem.

[0007]

According to a first aspect of the present invention, there is provided a circular container inner surface inspection apparatus which illuminates the inner surface side of an axisymmetric circular container from the axial direction of the circular container, and further includes a TV. In a circular container inner surface inspection apparatus that images an illumination surface of the circular container from the axial direction through a camera, analyzes the captured image, and inspects the circular container for deformation or dirt, the image is stored in a frame memory 1 or the like. Means for obtaining a plurality of radial scanning lines for scanning an image of the inner surface of the circular container as passing through the center of the image and having both ends at the outermost periphery of the image (processing area determination circuit 18A, etc.) The image of the inner surface of the circular container is sequentially scanned by a radial scanning line (via the radial address generator 4 or the like), and the density of the pixels on the radial scanning line arranged at one or a predetermined plurality of pixel intervals. scanning The absolute value of the density difference between each corresponding pixel and the density of a pixel at a position corresponding to the pixel on the radial scanning line having a predetermined interval is determined according to the coordinates of the corresponding pixel on the scanning line. The threshold value is compared with each other, and this comparison is also sequentially performed on the other pairs of radial scanning lines having the predetermined interval. When an absolute value of the density difference larger than the threshold value is detected, it is determined that the inner surface of the container is defective. It is assumed that the apparatus includes means for determining that there is a failure (the failure detection circuit 12, the failure detection determination circuit 13, and the like).

[0008] The circular container inner surface inspection apparatus according to claim 2 is
The inner surface side of the circular container is illuminated from the axial direction of the axially symmetric circular container, and the illumination surface of the circular container is imaged from the axial direction via a TV camera, and the image is analyzed to analyze the circular shape. In a circular container inner surface inspection apparatus for inspecting deformation and dirt of the container, means for subtracting the image of the bottom of the circular container and calculating the center of the image with reference to the edge of the difference image (new image edge detection circuit 14 etc. ) And the image of the inner surface of the circular container are sequentially scanned by a plurality of radial scanning lines passing through the center of the image obtained by the calculating means, and pixels on the radial scanning line are arranged at one or a predetermined plurality of pixel intervals. The absolute value of the density difference between each corresponding pixel between the density and the density of a pixel at a position corresponding to the pixel on the radial scanning line having a predetermined interval with respect to this scanning line is represented by the coordinates of the corresponding pixel on the scanning line. Each of the comparisons is performed with respect to another pair of radial scanning lines having the predetermined interval, and the absolute value of the density difference larger than the threshold is detected. In such a case, means for determining that the inner surface of the container is defective (the defect detection circuit 12, the defect detection determination circuit 13, etc.) is provided.

[0009]

The inner side of the circular container (eg, the ring illuminator 101) is illuminated from the axial direction of the axially symmetric circular container (eg, 102), and the circular container is illuminated from the axial direction via the TV camera. When imaging the illumination surface and analyzing the image, paying attention to the fact that the density distribution of the inner surface of the container occurs on a concentric circle as shown in FIG. 3, the image is radially scanned around the center O of the image. In step 3, after scanning the radiation scanning line Q-Q1 for the density change of the upper image, the radiation scanning line Q'-Q
1 'is scanned, and the density difference (density difference) between pixels at the corresponding coordinates (addresses) on both scanning lines is sequentially obtained, and each density difference is determined by a predetermined threshold T
Compare with HP. For example, a dent 10 as shown in FIG.
5, there is a change in shading of the scanning line Q'-Q1 'in FIG.
(C). 3B and 3C are sequentially obtained in the order of pixel arrangement.
Since the portion changed by 5 exceeds the threshold value THP, it can be detected as defective.

According to the first aspect of the present invention, a radiation scanning line is obtained from a fixed binary image of a circular container so that the outermost periphery of the image is located at both ends, thereby scanning an image outside the container. , The processing time can be reduced without performing useless scanning. Further, in the invention according to claim 2, the image of the inner surface of the container generated around the bottom high brightness portion 104 in FIG. 3 is differentiated, and the container center point O is estimated with reference to the edge of the difference image. Since the position of each radiation scanning line is accurately corrected so that all the radiation scanning lines pass through, the effect of the positional deviation of the container can be reduced.

As described above, according to the present invention, by scanning the image radially from the center of the circular container image and examining the change in the density of the pixel row obtained thereby, it is possible to detect dents in the container and to detect black spot defects. Can be inspected.

[0012]

FIG. 1 is a block diagram of hardware as one embodiment of the present invention. In the figure, PO is a TV outside the figure
The video signal obtained by raster scanning the camera screen is A
/ D conversion (for example, 8 bits) gray-scale image signal,
Reference numeral 1 denotes a frame memory which receives the multi-valued grayscale image signal PO and stores it as multi-valued screen data. Reference numeral 3 denotes a raster address generator which generates a raster address for the frame memory.

Reference numeral 4 denotes a radial address generator for generating an address to the frame memory 1 for radial image scanning, and 2 stores a ring-shaped mask pattern for each window as shown in FIG. 3B. A window address memory 5 for generating a raster address for the window memory 2 in the same manner as the generator 3;
Numeral 6 denotes a radial address generator for generating an address for radial scanning on the window memory 2 similarly to the generator 4. The generation of the raster address and the emission address can be switched by switching between the generators 3, 5, 4 and 6.

Reference numeral 7 denotes a window gate circuit which masks the multi-valued gray-scale signal PO or the image signal 1a read from the frame memory 1 with the mask pattern data from the window memory 2, and outputs the image signal only for the designated window area. This is a circuit for passing the PO or the image signal 1a read from the frame memory 1. Reference numerals 8 and 14 denote image edge detection circuits, which have a function of detecting an edge of an image, specifically, an outer end (outer peripheral point) and an inner end (inner peripheral point) of a ring-shaped high luminance portion. The binarized image signal is binarized with a predetermined threshold value for detecting the position of the target image and circularity inspection, and the coordinates of the rising point and the coordinates of the falling point of the binarized signal as an image edge are calculated. Store it in its own memory. Reference numeral 9 denotes a circuit for performing a circularity test on the coordinate values of the outer peripheral point or the inner peripheral point detected by the image edge detection circuit 8.

Reference numeral 15 denotes a position of the center of the actual target image detected by the image edge detection circuit 14 by inputting the latest multi-valued grayscale image signal PO so that a window is generated at a correct position with respect to the target image. This is a circuit for detecting a deviation from the set center position of the window. Reference numeral 10 denotes an image signal 1a read from the frame memory 1 by raster scanning of the frame memory 1 through a window gate circuit 7, and a defect detection circuit for detecting a dirty defective pixel such as a black spot. This is a failure detection / judgment circuit that counts defective pixels and determines a failure.

Reference numeral 12 denotes a defect detection circuit for inputting an image signal 1a read out from the frame memory 1 by the radial scanning of the frame memory 1 through the window gate circuit 7 and detecting a defective pixel as described later. , 1
Reference numeral 3 denotes a failure detection / judgment circuit that counts the detected defective pixels and determines a defect. 19 is a circuit for inputting the judgment results of the circularity judgment circuit 9 and the defect detection judgment circuits 11 and 13 to make a comprehensive judgment, and 20 is a circuit for this comprehensive judgment circuit 1.
9 is an output circuit that outputs pass / fail based on the output determination signal of No. 9.

Next, 16 is an X projection circuit for obtaining an X direction projection circuit pattern of the target image using the multi-valued image signal PO passed through the window gate circuit 7, and 17 is a Y projection for similarly obtaining a Y direction projection pattern of the target image. A circuit 18 is a circuit that uses the output data of the two projection circuits 16 and 17 to determine only the region of the target container image that is not connected to another container image.

That is, when the circular container to be inspected passes through a predetermined position, an image to be inspected is captured from the upper portion of the container by strobe light emission or exposure limitation by a shutter. During this capture period, the X projection circuit 16 and the Y projection circuit 1
The image is sent to 7 to obtain projection features. In the case of calculating the projection of a fixed binary image, the projection circuits 16 and 17 have a binarization function.

FIG. 4 shows a method for detecting a circumscribed rectangle of a container (that is, a hatched processing area) from a projection amount 203 in the X direction and a projection amount 204 in the Y direction using a fixed binary image 201 of the container. When moving along the transport path, the processing area determination circuit 18 separately performs connection separation processing to determine the processing area 202. Note that this connection separation is performed, for example, in Japanese Patent Application No. 3-2, filed by the applicant of the present invention.
No. 49946.

FIG. 5 shows an embodiment of the configuration of the failure detection circuit 12 of FIG. As described above, from the radiation address generator 4,
A radiation address is generated for the container inner surface image of the frame memory 1, and a certain radiation scanning line K (K and K + 1, K + 2, and K + 3 described below are sequentially arranged at predetermined regular intervals (equal angles). When scanning is performed, the density data of each pixel on this scanning line is stored in a FiFo (first-in first-out memory) 21-1 in the order of scanning. When the next scanning line K + 1 is scanned, the data on the scanning line K + 1 is stored in the FIFO 21-1, and at the same time, the data on the previous scanning line K is moved to the FIFO 21-2 and stored. Similarly, when scanning the next scanning line K + 2,
The data on the scanning line K + 2 moves to the Fifo 21-1, the data on the previous scanning line K + 1 moves to the Fifo 21-2, and the data on the previous scanning line K also moves to the Fifo 21-3.
And each is accumulated.

When scanning the next scanning line K + 3 in this manner, the data on this scanning line K + 3 is
-1 and at the same time
The data of 2, 21-3 are extruded in order, and
From 1-3, data on the scanning line K is given as one input of the subtractor 24. On the other hand, since the data on the scanning line K + 3 is also provided as the other input of the subtractor 24, the subtractor 24 eventually obtains the density data of the pixel at the coordinates (address) corresponding to the scanning lines K and K + 3, respectively. The comparison operation is carried out one pair at a time. The absolute value of the output (density difference) of the subtracter 24 is obtained by the absolute value conversion circuit 24, and the output of the subtractor 24 is further described by the comparator 26 for each scanning address (accordingly, for each pixel pair). It is compared with a threshold value THP provided from the threshold value table 23. Here, one pulse is sent from the radiation address generator 4 to the counter 27 every time one pixel advances on the scanning line, and the counter value indicates the address of the pixel on each scanning line. By using the address input, a threshold value THP corresponding to each pixel position can be given to the comparator 26 as described in FIG.

FIG. 6 shows a change in density of a pixel on a scanning line (Q-Q1 in this example) and a threshold value THP for each pixel address.
This is in comparison with a setting example of (THP1 to THP4). That is, a large threshold value TH is set in a portion where the shading changes greatly.
P1 and THP3 are given, and small threshold values THP2 and THP4 are given to a portion where the shading is small. FIG. 7 shows FIG.
The relationship between the absolute value of the density difference between the pixels at the corresponding addresses (B) and (C) and the threshold value (broken line) in FIG. 6 is plotted on the horizontal axis of the scanning line, and FIG. It is shown correspondingly. As shown in FIG. 7, the absolute value of the density difference in the portion of the dent 105 is larger than the threshold value THP2 in this portion, and is detected as defective.

FIG. 8 is a block diagram of a processing area determining circuit according to an embodiment of the present invention. This processing area determination circuit 18A is provided in place of the circuits 16, 17, and 18 in FIG. In the processing area determination circuit 18A, the fixed binarization circuit 31 converts the multi-valued image signal 1a on the horizontal line extracted from the frame memory 1 by raster scanning so that the mouth high brightness portion 103 of the circular container is cut out. Binarization is performed at a predetermined threshold value. The image change point detection circuit 32
A change point (ie, a rising point and a falling point) of the binarized image signal 31a output from the digitizing circuit 31 is detected, and a latch signal is supplied to the first rising point memory 33 and the final falling point memory.

As a result, the first rising point memory 33 stores 2
The X address of the first rising point of the digitized image signal 31a is stored, and the final falling point memory 34 stores the binary image signal 3
The X address of the last falling point of 1a is stored. By performing such horizontal scanning in the order of the Y address, the memories 33 and 34 store the X addresses of the first rising point and the final falling point in the order of the Y address, respectively.

FIG. 9 is a diagram for explaining the operation of determining the radial scanning lines by the circuit of FIG. 8. FIG. 9B shows an example of the data storage state of the first rising point memory 33 and the last falling point memory 34. It is shown. That is, the first rising point memory 3
3 shows a ring-shaped high-intensity portion 103 shown in FIG.
For each horizontal scan of the left half of the outer circumference of (i.e., for each Y coordinate)
, XSn are stored in the final falling point memory 34. Similarly, the X coordinate, XE, for each horizontal scan (Y coordinate) of the right half of the outer periphery of the mouth high luminance portion 103 is stored in the final falling point memory 34.
1, XE2,... XEn are stored. L (L1 to L3) in FIG. 9A indicates a radiation scanning line. First, the radiation scanning line L1 is determined as a straight line connecting the two points by reading the data XS1 of the leading Y address from the rising point memory 33 and reading the data XEn of the final Y address from the falling point memory 34. You. Similarly, the radiation scanning line L2
Is the data XS2 of the Y address next to XS1 and XEn, respectively.
And XEn-1 are read and determined as a straight line connecting these two points. Hereinafter, the scanning lines L3,... Are determined by repeating the same procedure.

The process for determining such a radiation scanning line is shown in FIG.
Is performed by the scanning function determination circuit 35. In this case, the radiation address generator 4 in FIG. 1 sequentially outputs the coordinates (address) of each pixel on each radiation scanning line L using the output data of the determination circuit 35. When only horizontal scanning is performed, the scanning lines become sparse near the upper and lower portions of the outer peripheral circle of the high-brightness portion of the mouth. To prevent this, the scanning direction is set to the vertical direction as shown in FIG. Changes can be made, which result in denser radiation scan lines.

Since the radiation scanning line L obtained as described above does not always pass through the center point O of the circular container image, the center point O is obtained by another means described below, and the rising point and the falling point are obtained. It is also possible to determine a straight line connecting any one point and the center point O. Next, a method of determining the center O of the circular container as an embodiment of the invention according to claim 2 will be described. That is, the image edge detection circuit 14 in FIG. 1 detects the ring-shaped high-brightness portion of the mouth to obtain the center position of the target image in the above, but when the radiation scanning line is used,
Since it is desirable to increase the accuracy of the center O, the center O may be determined using an image of the bottom surface of the circular container as described below.

The vicinity of the boundary between the side surface and the bottom surface of the circular container has low brightness because the illumination light hardly reaches it, but the difference in density from the surroundings is not so large as to be cut out at a fixed duplex level. Therefore, as a means for stably detecting an image, a binarization method using a difference is effective. FIG. 11A schematically shows the luminance distribution at the bottom of the circular container. FIG. 11A shows the bottom low-luminance section 11 around the bottom section 110 and the bottom high-luminance section 104.
1 shows an example in which 1 has occurred. In FIG. 2B, as shown in FIG. 2A, the Y-side position detection window 112 and the X-side position detection window 113 are set at specific positions on the screen. An example is shown in which detection and binarization are performed using a valley detection binarization method described later.

FIG. 11 (B) 2
The image edges of the value image are obtained in the X direction and the Y direction, respectively, a position reference point 114 determined by the X and Y coordinates of the edge is obtained, and a predetermined offset amount is added to the coordinates of the position reference point to obtain the center of the container. The position O can be determined accurately. Regarding the valley detection binarization method described above, a method for detecting a valley as a defective portion is described in detail in Japanese Patent Application No. 3-265134, which is a prior application of the present applicant. explain.

FIG. 12 is a diagram for explaining the principle of this valley detection binarization method. FIG. 12A shows an example of a multi-level grayscale image signal PO (X, Y) on a scanning line (Y = j) QQ1. Is shown. Here, 51 is a point of interest in the normal part, and 52 and 53 are a normal part front background point and a non-defective part rear background point as background points separated by the number of pixels α before and after the point of interest 51 on the scanning line, respectively. is there.

Similarly, reference numeral 54 denotes a point of interest in the valley portion,
And 56 respectively represent a valley front background point and a valley rear background point as background points separated by the number of pixels α before and after the point of interest 54 on the scanning line. Here, the coordinates of the point of interest are X = i, Y
= J, the difference term of the image signal PO (the following equation (1),
PO (i−α, j) −PO (i, j)> THD──── (1), and PO (i + α, j) −PO (i, j) > THD──── (2) (where THD is a predetermined threshold value (positive value)), a binary function value (valley binary image) for valley detection at the point of interest This signal is referred to as a valley by setting POD (i, j) = 1). In the normal part of FIG. 12, the above equation (2) does not hold and no valley is detected. (1) and (2) hold in the valley of
The valley can be detected. FIG. 12B shows the valley binarized image signal POD (X, j) as a valley determination output corresponding to FIG.

In this manner, the waveform of FIG. 12A is divided into small areas, and the values of the number of pixels α and the threshold value THD in the equations (1) and (2) are appropriately given for each of the small areas. Optimal valley detection can be performed. FIG. 13 shows an example in which the functions of the circuits except for the frame memory 1 and the raster address generator 3 in FIG. In this case, in particular, the radiation address generation circuits 4 and 6 as hardware for accessing the frame memory 1
It is important to reduce the processing time of the CPU 41 because there is no independent. To this end, it is effective to appropriately increase the interval between the radiation scanning lines or to appropriately increase the interval between pixels to be calculated on the radiation scanning line.

[0033]

According to the present invention, an image of a circular container is scanned by a radiation scanning line passing through the center thereof, and a density difference between pixels at corresponding coordinates on a pair of radiation scanning lines having a predetermined interval is determined. When the absolute value of the density difference exceeds a predetermined threshold value in accordance with the coordinates, it is determined that the defect is defective, so that local dents on the side of the circular container with low contrast can be accurately detected. Can be detected. In particular, in the invention according to claim 1, since the radiation scanning line is obtained from the fixed binary image of the circular container so that the outermost periphery of the image is at both ends, it is unnecessary to scan the image outside the container. ,
Processing time can be reduced without performing useless scanning. In the invention according to claim 2, the image of the inner surface of the container generated around the bottom high-luminance portion is differentiated, and the center point O of the container is estimated based on the edge of the difference image.
Since the position of each radiation scanning line is accurately corrected so that all the radiation scanning lines pass through, the effect of the positional deviation of the container can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a hardware configuration as one embodiment of the present invention.

FIG. 2 is a diagram showing a high-luminance portion on the inner surface of a circular container.

FIG. 3 is a diagram illustrating a change in density on a radial scanning line on the inner surface of a circular container and a conventional window division.

FIG. 4 is an explanatory diagram of a method for determining a container processing area according to the present invention.

FIG. 5 is a block diagram showing a configuration as one embodiment of a failure detection circuit 12 of FIG. 1;

FIG. 6 is a diagram showing an embodiment of setting a threshold value on a radiation scanning line using the threshold value table of FIG. 5;

FIG. 7 is a graph for comparing the absolute value of the density difference on the radiation scanning line with the threshold value to supplement FIG. 6;

FIG. 8 is a block diagram showing a configuration of a circuit replacing the processing area determination circuit of FIG. 1 as an embodiment of the invention according to claim 1;

FIG. 9 is a diagram showing an embodiment of a method for determining a radiation scanning line in FIG. 8;

FIG. 10 is a diagram showing an embodiment of a method for determining a radiation scanning line different from that in FIG. 8;

FIG. 11 is an explanatory diagram of a circular container center detecting method as an embodiment of the invention according to claim 2;

FIG. 12 is a principle diagram of a valley detection binarization method for supplementing FIG. 12;

FIG. 13 is a block diagram showing an embodiment of a configuration in which emission addresses are generated by software in FIG. 1;

[Explanation of symbols]

 PO Multilevel grayscale image signal 1a Multilevel grayscale image signal 1 Frame memory 2 Window memory 3 Raster address generator 4 Radiation address generator 5 Raster address generator 6 Radiation address generator 7 Window gate circuit 8 Image edge detection circuit 9 Roundness Determination circuit 10 Failure detection circuit 11 Failure detection determination circuit 12 Failure detection circuit 13 Failure detection determination circuit 14 Image edge detection circuit 15 Position shift amount determination circuit 16 X projection circuit 17 Y projection circuit 18 Processing area determination circuit 18A Processing area determination circuit 19 Comprehensive judgment circuit 20 Output circuit 21-1 Fifo 21-2 Fifo 21-3 Fifo 23 Threshold table 24 Subtractor 25 Absolute value conversion circuit 26 Comparator 27 Counter THP (THP1 to THP4) Threshold 31 Fixed binarization Circuit 32 Image change point detection times Road 33 First rising point memory 34 Final falling point memory 35 Scanning function determination circuit L (L1 to L3) Radiation scanning line 41 CPU 51 Normal part attention point 52 Normal part front background point 53 Normal part rear background point 54 Valley part attention Point 55 valley front background point 56 valley rear background point POD valley binarized image signal 101 ring illuminator 102 container 103 mouth high brightness section 104 bottom high brightness section 105 dent 110 bottom section 111 bottom low brightness section 112 Y side Position detection window 113 X-side position detection window 114 Position reference point O Container center 201 Fixed binarized image of container 202 Processing area 203 X-direction projection amount 204 Y-direction projection amount

Claims (2)

(57) [Claims] 1. An inner surface side of an axially symmetric circular container is illuminated from the axial direction of the circular container, and an illumination surface of the circular container is imaged from the axial direction via a TV camera. In a circular container inner surface inspection device to analyze and inspect the deformation and dirt of the circular container, a plurality of radial scanning lines for scanning an image of the inner surface of the circular container, passing through the center of the image, and, Means for obtaining the outermost periphery of the image as both ends, and sequentially scanning the image of the inner surface of the circular container with the radial scanning line, and the pixels on the radial scanning line arranged at one or a predetermined plurality of pixel intervals. The absolute value of the density difference between each corresponding pixel between the density and the density of a pixel at a position corresponding to the pixel on the radial scanning line having a predetermined interval with respect to this scanning line is defined as the coordinates of the corresponding pixel on the scanning line. Determined according to The threshold value is compared with each other, and this comparison is sequentially performed on the other pairs of radial scanning lines having the predetermined interval. When the absolute value of the density difference larger than the threshold value is detected, the inner surface of the container is detected. An inner surface inspection device for a circular container, comprising: means for determining a defect. 2. A method for illuminating the inner surface of the circular container from the axial direction of the axially symmetric circular container, capturing an image of the illumination surface of the circular container from the axial direction via a TV camera, and reconstructing the captured image. A circular container inner surface inspection device that analyzes and inspects the circular container for deformation and dirt; a means for comparing an image of the bottom of the circular container and calculating a center of the image based on an edge of the difference image; The image of the inner surface of the container is sequentially scanned with a plurality of radial scanning lines passing through the center of the image obtained by the calculating means, and the density of pixels on the radial scanning line arranged at one or a predetermined plurality of pixel intervals. The absolute value of the density difference between each corresponding pixel and the density of the pixel at the position corresponding to the pixel on the radial scanning line having a predetermined interval with respect to the scanning line is determined according to the coordinates of the corresponding pixel on the scanning line. Was The threshold value is compared with each other, and this comparison is also sequentially performed on the other pairs of radial scanning lines having the predetermined interval. When an absolute value of the density difference larger than the threshold value is detected, it is determined that the inner surface of the container is defective. A circular container inner surface inspection apparatus, comprising: means for determining that there is a circular container.