Research On Target Automatic Recognition Technology Based On Image Processing

Annual assessment evaluation of the personnel enterprises is a job that the organization department must do every year. At present, annual assessment of enterprises and institutions is accomplished by manual filling paper assessment tables, manual organizing tables and collecting scores. However, workload is heavy and error rate is high. To solve this problem, this thesis named Research on Automatic Target Recognition Based on image processing is proposed. It can recognize and process evaluation information in assessment tables automatically using digital image process.In order to recognize evaluation information quickly and accurately in a whole assessment table, this thesis proposes an algorithm based on digital image process to automatically recognize feature objects in tables. First, by studying preprocessing technology on image of assessment tables, this thesis develops a new preprocess operation to get clear image. The new preprocess operation consists of grayscale image, binarization, smoothing, tilt correction, image thinning and image normalization. Second, a new cell location algorithm is developed in this thesis after analyzing layout features of assessment tables, which takes use of Connected Component and Region of Interest detection methods. Third, this thesis studies how to automatically recognize character information using Optical Character Recognition(OCR).Finally, according to the feature of objects(ticks), the thesis studies to automatically recognize ticks using statistics pixels and corner detection methods.This thesis is the first issue that applies digital image processing to cadres assessment evaluation in enterprises in domestic. The correctness and feasibility of the algorithm is verified and the running status of the system is stable.