A chest silhouette recognition method based on digital image processing

Abstract

The digital target image, in a machine vision based automatic scoring system, should be recognized for the subsequent automatic scoring and records analysis. However, target images collected from a camera are bound to be distorted, which would pose some difficulties to recognition. In this paper, according to the automatic closed loop controlling idea, a loop algorithm is proposed after the verification of both the intermediate results and the final ones with the utilization of prior knowledge, which is founded on the basic image processing algorithm and employs the technology of border extraction and segmentation as well as the pixel search. And it is considered that the position of the 10.9 ring and the spacing between each two rings are obtained through the recognition of the chest bitmap at a minimum time with the least memory resource with both accuracy and precision guaranteed. The method put forward by this paper has been tested a lot in android automatic scoring system practically, so the results feature reliability, accuracy and accordance with the requirements of the system.