Image Retrieval System of Fish Disease Diagnosis by Using Shape Feature
In order to overcome the inaccuracy in the fish disease diagnostician system which relies on text information of the diseased fish, a retrieval system was developed based on shape feature for fish disease images. The diseased fish was segmented using threshold method and the image edge was extracted by means of the 3 neighbourhood chain code method. The shape characteristics of the fish diseases image were analyzed and six angles of the fish body and the two ratios were taken as eigenvalues to structure the retrieval matrix. The similarity measurement was computed based Mahalanobis distance model after the normalization process was given by means of Gaussian normalization method. The fish disease image retrieval system is designed in development environment Visual C++6.0. Experiment results indicate that shape feature-based retrieval method has a fairly good recognition effect with a precision rate of 62％ and a recall rate of 83％. It will greatly improve the system robustness and meet the demands of fish disease diagnosis when this algorithm is applied to the fish disease diagnostician system.
fish disease diagnosis shape feature image processing retrieval system
Jian Song Xueyan Sun
College of Machinery, Weifang University, Weifang, Shandong, China College of Information and Control Engineering, Weifang University, Weifang, Shandong, China
国际会议
2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)
上海
英文
249-252
2010-12-25（万方平台首次上网日期，不代表论文的发表时间）