The Recognition Technology Of Go Chess Manual Based On Video Image Processing

Go originated in China and has been a very popular game in China since ancient times.With the development of the times,not only is it widely popular in Asia,but also it has become a popular sport all over the world.Especially because of the emergence of ALPHOGO,Go has attracted much attention not only in the sports field but also in the computer field.For many beginners,it is very meaningful to observe the game process of others.However,in many cases,the game process of Go is realized by manual notation,which is not only time-consuming and laborious,but also the correct rate is not guaranteed.Therefore,in order to solve this problem,this paper uses image processing technology in the computer field to process the video of Go game,and designs a set of processing algorithm to recognize the video of Go game and record and output the information of chess spectrum.The processing algorithm proposed in this paper is mainly divided into two parts.The first part of the research is mainly to obtain the accurate coordinates of the 361 intersections of the board in the image.First,the video is taken every twenty frames to obtain a continuous sequence of pictures.Select a clear picture of the board and perform grayscale and image segmentation.In the image segmentation,different threshold segmentation methods and Canny edge detection are used to process and compare the analysis separately.Finally,the adaptive threshold method is used to binarize the image to obtain a clear checkerboard contour.Then,the line information and the corner point information of the contour image are obtained by Hough line detection and corner point detection respectively.The interference line is removed by the positional relationship between the line and the corner point,and the line is classified by the design clustering method,and then the corresponding corner points are determined by corresponding categories to fit the straight lines of chessboard trend.Finally,the excess line is deleted by the relationship between the lines and the missing line is complemented.The 38 lines of the checkerboard are successfully found,and the simultaneous equations calculate the exact coordinates of 361 intersections.The second part of the study is to obtain complete chess spectrum information.The chessboard area is divided into 361 sub-images by using the intersection coordinates obtained in the first part.In each sub-image,the falling pieces are judged by Hough circle detection combined with edge point counting method.The counting method judges the falling pieces,and the picking pieces respectively use the logic of the chess piece to judge according to thesituation and compare whether the occlusion and the quilt are separated by comparing the picture stream.Then,according to the above method,a set of process algorithms for identifying the notation is designed.The speed of the processing algorithm is optimized during the experiment.Finally,the occlusion processing is analyzed.A large number of experiments show that the processing algorithm designed in this paper can effectively recognize and record the correct chess spectrum information for normal game video.