JPH04184107A - Visual inspection device - Google Patents

Abstract

PURPOSE:To dispense with the need of a picture-image memory, and to enable high- speed processing by detecting the start point of an object, and in the respective scannings thereafter, by providing this start point with an offset value for producing an inspection area, and on the basis of this, by performing characteristic detection. CONSTITUTION:The picture-image signal from a television camera 1 is converted into binary digits by an A/D converting part 2, and the change point of this binary- digitized image from 0 to 1, or from 1 to 0 is detected by a change-point detecting part 3. The coordinate values of respective change points are successively added, and the total integrated value is latched by a coordinate latch part 8. The data in the latch part 8 are given to a coordinate shift part 9, and the average value of the middle points of all the coordinates of change points is obtained. To a X-offset setting part 10, start-point coordinates are given from the shift part 9, while an offset value is set, and this is added to the coordinate values. When the coordinate values reach a prescribed value, a start signal is produced in an area producing part 12. A Y- directional offset value is added by a Y counting part 15, and when the coordinates reach a prescribed value, a start signal is output to the producing part 12. A characteristic extraction part 13 extracts a characteristic value within the inspection area.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention uses a television camera or the like to extract an image of an object as an electrical signal, detects the external shape of the object from the signal, and detects a predetermined shape according to the imaging signal. The present invention relates to an appearance inspection apparatus that generates a region, extracts feature quantities, and performs an appearance inspection of an object.

[Conventional technology]

Conventionally, a predetermined inspection area is set according to the external shape of the object for the imaging signal from a television camera, etc., and when extracting the characteristics of the object within this area, a method is used to specifically deal with changes in the position of the object. In order to do this, the object image is temporarily stored in memory, its outline is detected, and then the image is retrieved from the memory and an inspection area corresponding to the detected object outline position is generated. , it is common to extract features within that region (
For example, see Japanese Patent Publication No. 63-12310).

[Problem to be solved by the invention]

The above method not only requires memory to store the image, but also requires two steps of processing: inputting the image to memory, outputting the image data from memory, and processing, which is time consuming. There are also problems.

Therefore, the problem to be solved by the present invention is to provide an appearance inspection apparatus capable of high-speed processing.

[Means to solve the problem]

In order to solve such problems, the present invention provides a binarization means that binarizes an imaging signal obtained by imaging an object while performing raster scanning, and a 0-1 change point and a 1 change point of the binarized signal. -Position detection means for detecting each coordinate position of the 0 change point; and detection means for detecting the averaged midpoint of all change points by sequentially adding the coordinate values of each change point and shifting the result by a predetermined number of bits. and calculate the averaged midpoint from a preset number of change point coordinate values, generate a predetermined inspection area at a preset position starting from this midpoint in each subsequent scan, and It is characterized by extracting features related to the object to be inspected within the area.

[Effect]

The position near the tip of the object (starting point) is detected in the first few scans, and in each subsequent scan, a predetermined inspection area is generated at a position that has a certain positional relationship with the detected starting point. Executing feature extraction within a region eliminates the need for image memory and enables high-speed processing.

〔Example〕

FIG. 1 is a block diagram showing an embodiment of the invention, and FIG. 2 is an explanatory diagram for specifically explaining the processing according to the invention.

In FIG. 1, 1 is a television camera as an imaging device;
2 is an analog/digital (A/D) converter, 3 is a change point detector (DET), and 4 is a change point counter (CNT).
, 5 is the X (horizontal) coordinate generator (XGE), 6 is the Y (vertical) coordinate generator (YGE), and 7 is the X coordinate adder (ADD).
), 8 is the coordinate latch section (LCH), 9 is the coordinate shift section (SFT), 10 is the X offset setting section (ADJ)
, 11 is the X address comparison unit (XCMP), 12 is the area generation unit (AGE), and 13 is the feature extraction unit (FET).
, 14 is a microprocessor (MP), 15 is a ΔY counter (ΔYCNT), and 16 is a bus.

The operation will be explained below with reference also to FIG.

First, the image signal from the television camera 1 is sent to the A/D converter 2.
200 million, and this binarized image's 0 → 1,
The change point from 1 to 0 is detected by the change point detection section 3.

The changing point counting unit 4 counts the number of each changing point, and calculates the number of changing points for the preset number of scanning lines (2"), that is, 2 x 2" (
2"'), it outputs a termination signal. The X coordinate generating section 5' outputs the coordinate value in the X scanning direction according to the screen scan, but the coordinate latch section 8 receives the signal from the change point detecting section 3. Therefore, first, the X coordinate corresponding to the first change point (see XLI in FIG. 2) is latched by the coordinate latch unit 8 via the Every time a signal is given to the coordinate latch section 8,
The operation of adding the previously latched X coordinate to the current X coordinate in the X coordinate adding unit 7 and latching it in the coordinate latch unit 8 is repeated, and an end signal is received from the change point counting unit 4. At this point, the total integrated values of 2n+1 change point coordinates are latched in the coordinate latch unit 8. Upon receiving the end signal from the change point counting section 4, the data of the coordinate latch section 8 is given to the coordinate shift section 9, which converts the data to (n+1).
By shifting by bits, the average value of the midpoints of all the change point coordinates of a preset number is determined.

In FIG. 2, change points XLI and XRI are detected in the first scan, and change points XL2 and XL2 are detected in the second scan. An example is shown in which XR2 is detected and the average value (XS, YS) of the midpoint is determined as the starting point P. Note that Ml is the change point XL1. The Y coordinate ys, which indicates the midpoint of XRI, and the coordinate XS is detected by the Y coordinate generator 6.

The starting point (XS, YS) is read by the microprocessor 14. The X offset setting section 10 includes a coordinate shift section 9.
While the starting point coordinates are given from the microprocessor 14, a predetermined amount (offset amount ΔX) is set, so the
The amount ΔX is output and given to the X address comparator 11. However, this offset amount ΔX is signed,
When it is to the right of XS, it is positive, and when it is to the left, it is negative, and it is determined in advance by an appropriate method. The X address comparison unit 11 outputs a start signal for area generation in the X direction to the area generation unit 12 when the coordinates output from the X coordinate generation unit 5 during X direction scanning become XS+ΔX=XO. .

On the other hand, when the scanning amount after receiving the end signal from the change point detecting section 4 reaches YS+ΔY=Y O, where the Y direction offset amount is ΔY, the ΔY counting section 15 instructs the area generating section 12 in the Y direction. Outputs a start signal for area generation. The area generating section 12 receives these signals and generates an inspection area (inspection area) in accordance with subsequent screen scanning.

The inspection area is indicated by the symbol A in FIG. Note that the offset amount ΔY and the inspection area A are also determined in advance by an appropriate method.

The feature extraction unit 13 extracts the feature amount of the object image including the area within the inspection area A, and the microprocessor 14 reads this and compares it with a predetermined reference value to determine its quality.

By doing this, even if the object position changes, an inspection area can be generated that matches this, and desired features can be extracted within that area.

〔Effect of the invention〕

According to this invention, a starting point position near the tip of the object is detected in response to image scanning, an offset value is given to this starting point position in subsequent scanning to generate an inspection area, and feature extraction is performed based on this. Since this is done, it is not necessary to temporarily store the image of the object to be inspected, and therefore, there is an advantage that stable and accurate inspection can be performed simultaneously with image scanning and regardless of changes in the position of the object.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the invention, and FIG. 2 is an explanatory diagram for specifically explaining the processing according to the invention. 1... Imaging device (television camera), 2... Analog/digital (A/D) conversion section, 3... Change point detection section (DET), 4... Change point counting section (CNT), 5
...X coordinate generation section (XGE), 6.. Y coordinate generation section (YGE), 7.. X coordinate addition section (ADD),
8...Coordinate latch section (LCH), 9...Coordinate shift section (SFT), 10...X offset setting section (AD
J), 11...X address comparison unit (XCMP),
12...Area generation unit (AGE), 13...Feature extraction unit (FET), 14...Microprocessor (
MP), 15...ΔY counting section (ΔYCNT), 16
...Bus, A...Inspection area.

Claims (1)

[Claims] 1) Binarization means that binarizes an imaging signal obtained by imaging an object while raster scanning it, and a 0→1 change point and a 1→0 change point of the binarized signal. position detection means for detecting each coordinate position of the change point, and detection means for sequentially adding the coordinate values of each change point and shifting the result by a predetermined number of bits to detect the averaged midpoint of all change points. , find the averaged midpoint from a preset number of change point coordinate values, generate a predetermined inspection area at a preset position starting from this midpoint in each subsequent scan, and An appearance inspection device characterized by extracting features regarding an object to be inspected.