FPC product classification methods are many, according to the FPC layer can be divided into: single-sided, double-sided, multi-layer board and soft and hard combination board. FPC as a common type of circuit board, its market share with the electronic products to miniaturization, light development continues to rise. But FPC in the processing, feeding, placement and other production processes may be broken, short circuit, line width and other defects. In view of this, this paper mainly analyzes the FPC defect detection method.
 
Different production process to make it have many unique features:
(1) high assembly density, reducing the connection between the parts;
(2) light weight, thin thickness, can effectively reduce the weight of the product, easy to carry;
(3) can be folded, can bend any bending.
 
FPC has the advantages of good reliability, strong heat dissipation, easy installation and low comprehensive cost, and is convenient for the high integration and high performance of electronic products.
FPC product classification methods are many, according to the FPC layer can be divided into: single-sided, double-sided, multi-layer board and soft and hard combination board.
 
Development Status of FPC Defect Detection Technology
The existing FPC defect detection algorithm is mostly derived from the PCB detection algorithm, but by its own unique limitations, FPC defect accuracy is higher, the detection model size is bigger, the template imaging is easy to deform, so that the PCB board defect detection algorithm can not directly apply FPC detection algorithm, according to the actual characteristics of FPC to develop appropriate detection algorithm.
In order to solve the problem that the traditional template matching algorithm is slow and the accuracy is low, the FPC line defect is divided into two parts: global defect and local defect. The eight-connected domain area method and the histogram matching method are used to capture the global defect of the image. Based on this, the local defects in the image are identified by projection matching and correlation coefficient method. The method is faster and more accurate than the traditional detection algorithm, but the classification of the relevant defect category is not enough.
 
Analysis of FPC Defect Detection Method
Aiming at the problem of template matching in the global scope, FPC defect detection is implemented considering the local range template matching method. Considering the contour-based template matching method, the template should have significant contour characteristics. Although the conventional lines on the FPC go to the rules, there is no significant shape feature. Moreover, the conventional lines are distributed in the whole image, and the template matching time Too slow, is not conducive to line detection.
Shaped lines are irregular shape on the FPC, generally including LED lights, S-shaped round; due to the shape of the peripheral, the wiring is generally related to the style of the FPC to be tested. For this type of line, consider the template matching method to implement the test: first through the template matching method to steer the various contour lines on the whole FPC position, access to contour lines; and then based on morphological theory for defect detection.
 
Guidelines for FPC Defect Detection Methods
(1) load the registration template related data information, including the registration template area, registration template contour information.
(2) using the area information carried by the registration template, locate the template search space, and search the template instance in the search image based on the normalized mutual correlation coefficient (NCC) metric principle.
(3) cut the contour area. The size of the profiled area is obtained by finding the smallest circumscribed rectangle of the shape template instance, since the shape template is a rectangular region, so the area obtained by the minimum circumscribed rectangle is the matching alien region. Any template matching can not make the two images perfectly aligned in space, there is a permissible range of matching. FPC local range can be ignored, the shape of the template matching template space matching due to narrow space, shape template matching accuracy, the overall matching error of about 1/5 line width (3 pixels) range, the defect can be determined Allowable range.
(4) Considering the accuracy error of shape matching, before the standard template image and the profiled area image are directly measured, the small-size structural elements are used to etch the alien-shaped area to remove the boundary difference effect. When the image is defective, the corrosion operation at this size can not completely corrode the defective line, so it does not affect the final judgment result.
(5) the template image and the contour area of ​​the regional difference set operation, access to the difference between the two areas. When there is no defect, the difference is 0; when there is a defect, the area difference set returns a binary image that is not zero;
(6) the difference between the image area of ​​the domain marked, split the different defect blocks;
(7) Calculate the area of ​​each connected domain, and use the geometric center of the largest defective block as the center of the profiled area to output the defect information.
 
The FPC defect detection method is studied from the global and local scope, and the experimental results are drawn:
(1) due to FPC imaging deformation, based on the global scope of the template matching line dislocation, can not correctly locate the image on the defect information.
(2) based on the local scope of the template matching, the first cut off the FPC on the conventional line with significant shape characteristics of the line area, as a template source; in the search image corresponding to the template area near the instance detection, compression search space; And then based on morphological theory to detect line defect information.