JPH01189549A - Detecting device of defect of glass - Google Patents

Abstract

PURPOSE:To enable the detection of a defect of glass with high resolution and at a high speed, by a method wherein a light from a light source provided linearly is projected into the glass through a slit. CONSTITUTION:Glass 2 is irradiated by a light of a light source, e.g. a halogen lamp 7, through a slit 6. In the case when a defect exist inside or one the surface of the glass 2, a normal part thereof is bright and a defect part is dark when the light is applied to the glass 2. The intensity of the light therefrom is sensed by a one- dimensional camera 1, converted into an electric signal and transmitted to a prepara tory processing element 11. The glass 2 is carried at a prescribed speed by a carrying device 5 driven by a motor 3, and an information on the position of the glass 2 is detected by a rotary encoder 4 fitted to a driving system and sent likewise to the preparatory processing element 11. In the preparatory processing element 11, a video signal from the camera 1 is transformed into one-dimensional binary-coded data and further it is subjected to run-length coding and transmitted to an after processing element 12. In the after processing element 12, one-dimensional signal is converted into a two-dimensional signal on the basis of the run-length data and a position signal from a carrying control element 10, and thereafter the dimensions of the defect part are calculated to determine the quality of the glass 2.

Description

[Detailed description of the invention] [Industrial application fields] The present invention relates to a glass defect detection device.

[Prior Art] Conventionally, a glass defect apparatus that captures a one-dimensional camera signal, converts it into a two-dimensional signal, and processes the image has been known, but it has been insufficient in that it cannot perform image processing in real time.

[Problem 8 to be solved by the invention] Some conventional glass defect detection systems use two-dimensional cameras, but the resolution per camera is small, so if you try to increase the detection area, the number of cameras will increase. The problem is that the number of image processing devices increases, the size of the image processing device increases, and the price becomes high. Further, in the case of using a one-dimensional camera, since the 1.1-dimensional signal is processed as it is, there is a problem that the defect image (two-dimensional image) cannot be accurately recognized.

Furthermore, image processing devices that use technology to capture one-dimensional camera signals and convert them into two-dimensional signals use a method that stores all data in memory without compressing the one-dimensional signal. If you try to increase the resolution, the amount of data will increase, the amount of memory required will increase, the amount of data that can be captured will be limited, and the processing speed will slow down.

Therefore, when the conventional device is applied to a glass defect detection system, there is a problem that processing cannot be performed in real time.

[Means for Solving the Problems] The present invention has been made to solve the above-mentioned problems, and includes a mechanism for controlling the conveyance of glass, and a mechanism for transmitting light from a linearly arranged light source to the glass through a slit. By projecting light, a signal detection unit converts the internal defects and surface defects of the glass into electrical signals using a one-dimensional camera as shadows of light, and after analog/digital conversion of the electrical signals from the one-dimensional camera, filtering, A preprocessing unit that binarizes, converts the binary data into a run-length code, and transmits the run-length code, and converts the one-dimensional data into two-dimensional data from the run-length code and measures the defect shape in real time. The present invention provides a glass defect detection device comprising a post-processing section that determines the quality of glass.

Preferred embodiments of the glass defect apparatus according to the present invention will be described in detail below with reference to the drawings. FIG. 1 shows the case of transmission as the light irradiation method. The glass 2 is illuminated by light from a light source such as a halogen lamp 7 that has passed through the slit 6 . When light is irradiated onto the glass 2, if there is a defect inside or on the surface of the glass, the normal area will be bright and the defective area will be dark.
It receives the light, converts it into an electrical signal (video signal), and transmits it to the preprocessing section.

The glass 2 is transported at a constant speed by a transport device 5 driven by a motor 3, and position information of the glass 2 is detected by a rotary encoder 4 attached to the drive system and similarly transmitted to the preprocessing section. The pre-processing section converts the video signal from the -dimensional camera 1 into one-dimensional binary data, further performs run-length encoding on this data, and transmits it to the post-processing section. The post-processing section converts the one-dimensional signal into a two-dimensional signal based on the run length data and the position signal from the conveyance control section, calculates the size of the defective part, and determines the quality of the glass 2.

Next, the preprocessing section will be explained in detail using FIG. 2.

The video signal received from the signal detection section is A/D converted.
The signal is converted into a digital signal (multi-level) in the ``filter'' section, and then noise is removed in the ``filter'' section, and then in the ``binarization'' section, bright areas in the normal area are converted to ``white'' and dark areas in the defective area are converted to ``white''. It is converted to binary data with the part as "black". Furthermore, in the "run length code" section, it is converted into run length data representing the length of "white" and "black", and stored in the "run length memory" via the run length memory changeover switch 8. be exposed.

The position signal detection unit 1, which receives the position signal from the transport control unit, sends a scan signal that causes the -dimensional camera to scan to the camera drive circuit unit. Also, based on the position signal, a run length data transfer signal is generated to control the run length memory changeover switch 8 and the "run length memory". The reason for using two sets of "run-length memories" and switching between them is to take into consideration the time required to transfer data from the "run-length memory" to the post-processing section. This mechanism prevents data from being lost and enables real-time continuous processing. The operation of this mechanism will be explained below. ■
Let us consider a state in which data is stored and connected to changeover switch 8, rNo., 1 run length memory. (2) In response to the run-length data transfer signal, the selector switch 8 is first switched to "r No. 2 run length memory" and data is stored in "r No. 2 run length memory".

The switching at this time is done in the timer between the data sent from the "run-length encoding" section, so there is no loss of data by switching the "run-length memory". . ■The selector switch 8 is set to r
After switching to "No. 2 run length memory", rN
The data in the ``1 run length memory'' is transferred to the ``communication boat'' via the bus under the control of the microprocessor.
and then transmitted to the post-processing section. By making the transmission time used at this time shorter than the time (measuring time) to capture the amount of data stored in the "run-length memory" on one side, the two "run-length memories" can be sequentially switched, allowing real-time measurement. , run length data can be transmitted to the post-processing section.

Next, the post-processing section will be explained using FIG. The transmitted run length data is captured every l scans and compared with the previous scan to check the continuity of dark areas. As a result of this check, if it is understood that a dark area or a new beginning has started (Fig. 3, No, x+1 scan, y+6 to y+10
areas) are numbered to distinguish between areas (
For example, N). At this time, at the starting position of the dark part (x+l,
y+6), end position (x+1.

y+IO), ffl(5) is determined from the run length data and stored. If the dark area is understood to be continuous from the previous scan (No. 3 in Figure 3).

x+2 to x+7 scan dark area), scan the start position with No.
is updated as it is to the smaller bit number only, the end position is updated to the larger bit number along with the scan number, and the length is added to the previous length to calculate the area. Reread and memorize these values. If it is understood that the dark part is isolated, (Fig. 3 NO, x + 8
scanning) dark area N is the starting point coordinates (Fig. 3 5=(X+1
．． y+1)), end point coordinates (Fig. 3 E= (x+17
, y+lo) ), area (Fig. 3 Number of dark pyrocells =
47), and based on these, the effective diameter is determined by the following procedure.

(1) Multiplying the diagonal distance of S to E by a conversion coefficient determined in advance is obtained as a distance in the physical unit system.

(2) Multiply the area (47 in the case of FIG. 3) by a conversion factor determined in advance to obtain the area in the physical unit system.

(3) An ellipse whose major axis length is the obtained diagonal distance is regarded as a defect, and the short axis length of the ellipse is determined by assuming that the area matches the value obtained in (2).

(4) The effective diameter of the defect is determined as 1/2 of the sum of the major axis length and minor axis length determined in (3).

(5) Based on the effective diameter determined in (4), classify the size of the defect or determine whether it is good or bad according to predetermined criteria, and output the results to the outside.

By completing the above arithmetic processing within one scanning period of the signal detection section, real-time processing becomes possible as a defect detection system.

3 In the present invention, minute changes in the refracted light and reflected light that occur on the glass surface or at the interface with internal defects are considered as changes in the amount of light within the camera field of view, but linear white light is projected. This has the effect of increasing the S/N ratio by reducing the influence of light from sources other than the defective portion.

[Effects of the Invention] Since the present invention is configured as described above, defects in glass can be detected with high resolution and at high speed, and minute defects that occur in various types of glass can be detected in real time. be. Moreover, defects can be detected in terms of size, so if a standard is set in advance, the judgment can be made by comparing with this standard. Therefore, by adjusting this standard in accordance with the required level of detection in consideration of the type of glass, etc., defects can be detected at an appropriate level.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the glass defect detection device according to the present invention, FIG. 2 is a block diagram of the pre-processing section shown in FIG. 1, and FIG. 3 is a processing method of the post-processing section shown in FIG. 1. FIG.

Claims (1)

[Claims] (1) By using a mechanism that controls the transport of glass and projecting light from a linearly arranged light source onto the glass through a slit, a one-dimensional camera can detect the internal defects and surface defects of the glass as shadows of light. After converting the signal detection unit into an electrical signal and the electrical signal from the one-dimensional camera from analog to digital,
A preprocessing unit that filters, binarizes, converts the binary data into a run-length code, and transmits the run-length code, and converts one-dimensional data into two-dimensional data from the run-length code, and detects defect shapes in real time. A glass defect detection device comprising a post-processing section that measures and determines the quality of glass.