In this dissertation, several key techniques of CBIR are deeply analyzed and discussed, such as the low-level feature descriptions including color, shape, and texture, the similarity measure between images, and the relevance feedback, and tried to apply in power equipment images. According to the characteristics of power equipment images, the important feature on power equipment images should be the shape feature. This dissertation proposed an approach of image retrieval in power equipment management system, the power equipment images pre-processing and shape feature parameter's extraction have been completed. The power equipment image retrieval system structure based on moment invariants is introduced. The experiment indicates that the method can distinguish different sorts of equipment image better. Finally the summary and prospect of the work was made. |