Application Of Wavelet Transform On Multimedia Image Retrieval

Content-Based Image Retrieval (CBIR) is a kind of technique for retrieval images on the basis of automatically extracting visual features such as color, texture, and shape etc. It has been an active research area in recent years. In this paper, we extensively studied the national and international materials on CBIR retrieval systems, discuss the research status and trend in content based image problems, and presented a method that multi-resolution wavelet transform application in image retrieval. At first we made partitions to image in multi-resolutions, we can find the low frequency image, we can figure out the energy and wavelet entropy of the image, then image retrieval by two steps. One step is using low frequence of the image to probably retrieval,another step is Compare the entropy of the sample image with the entropy of the retrieval image, then figure out the similitude space distance, Value can enactmented to classify by the image. In order to improve the performance of the retrieval arithmetic, In this paper, we have two types of image databases, one is standard face BMP image, which includes 1000 pictures of face image, another is transfer JPEG into BMP image, which includes 100 BMP pictures of cars, 100 BMP pictures of architecture and 150 BMP pictures of scenery. Provides arithmetic aims at different types of image databases, gains the high retrieval purpose.Because feature vectors are all high linked, so, in the retrieval based on content, to calculate the distance will consume a large amount of time. In order to reduce the dimension in the feature vectors, we provided wavelet analyze to reduce dimension of the image in detail in this paper with the amount of calculation reduced. In the similar degree of tolerance of the picture, we adopt the method of the comprehensive tolerance under the different resolution ratio and different wavelet, we develop the image retrieval arithmetic to prove its validity and exactness. We have done a large number of experiments and obtained the better experimental result.In the last part, after epitomizing the CBIR technique and pointing out my work, the paper analyze the complexity of the CBIR and provide CBIR's prospect.