Chip Surface Character Detection And Recognition System Based On Machine Method

As the direct identification of component model,the character on the surface of components plays an important role in automatic quality inspection and defect detection.In the current industrial environment,most of the traditional manual methods are used to detect the characters on the surface of components,which can not guarantee the accuracy and is inefficient.Based on the industrial environment,this paper studies a method to detect and recognize the characters on the surface of components,which solves the above problems on the basis of automatic optical detection.Compared with the traditional template matching method,the deep learning algorithm is simple and efficient in the process of detection and recognition.This paper summarizes the text detection and recognition algorithms at home and abroad,and makes a deep understanding of the development of text detection and recognition in natural scenes.Firstly,this paper compares the effect of template matching and craft in character detection,and solves the problem of character adhesion in template matching by threshold cutting method.To solve the problem of unstable detection effect,the real loss and correlation coefficient are optimized to improve the stability of the result of craft algorithm.Then,in the aspect of character recognition,a similarity detection method is proposed,which solves the interference of similar characters and tainted characters on the recognition results,improves the accuracy of character detection using template matching,and achieves good recognition effect through Tesseract training on its own data set.Then through the research and comparison of the methods in this paper,it is found that in the process of character recognition,the reflection problem is an important factor leading to the reduction of recognition rate.Through high dynamic range imaging,the problem of low recognition rate caused by the reflection problem in the recognition process is solved.The adaptive mapping and bilateral filtering mapping are compared,and the bilateral filtering mapping algorithm with better reflection suppression effect is selected.Finally,the software structure and hardware composition of the whole automatic optical detection system are introduced,and the experimental data and result analysis are given.Based on the automatic optical detection system for surface characters of components,the common problems of surface character detection and recognition in common industrial environment are solved by using threshold cutting,optimized craft algorithm,similarity comparison and high dynamic range imaging.