Study On The Algorithms Of Searching Of Content Based Image Of Fish Retrieval And Development Of Retrieval System

Taking ill fish image as a specific study object, the fundamental theory and practicable method of Content based image retrieval has been studied. Combining digital image processing technique, pattern recognition and machine vision technique, a demo system had been presented using the technique of Content based image retrieval.At the present time the using the technique of Content based image retrieval is mainly focused on the features of the target in the image. The features include the color of the target, the texture of the target, the shape of the target and their combination. Considering the fact of the ill fish image, the paper researched and presented the algorithm of content based image retrieval that was according to the represent shape, color or vein in image. This method can either present statistical features or spatial information in the image.A new algorithm was presented to gain the shape feature of the image in the paper ,the new algorithm can greatly improve retrieval efficiency. The paper used six angle of the fish body and the two ratio as the shape eigenvalue.This retrieval was the main retrieval method of the demo system.Based on the HIS color model, the paper selected five color eigenvalue (kurtosis, energy). Using grey level co-occurrence matrix, picked up CON,ASM,ENT and COR as the vein eigenvalue.In the Windows XP environment, the image retrieval demo system, which is based on the image characteristic, are implemented with Visual C++6.0.