Research Of Image Retrival Base On Multi-Features Fusion

In recent years, how to retrieve the data from the vast amout of informations which we were interesting in, has become a major concern. And the informations that be contained in image were rich and intuitive. Therefore, image retrieval technology in recent years has become a new hotspot.For the color features, we discuss the image important point firstly, around the different color space, test and analysis the retrieval results that base on important point color feature. Then discuss the Fourier method, to convert the image date to frequency domain to extract the texture feature. After that, we design a arithmetic that could reflect a target image profile to describe the image shape feature. Then make this method to compare with the Hu method and analyzed. Finally, using the idea of information fusion to discuss the image characteristics normalization method. Then, through the weight which the users made, to solve the issue of informations fusion. Experimental results show that it is a more effective image retrieval method than normal method.