An Image Retrieval Algorithm Based On Shape Feature Matrix

With the research and development of extensive data and its related technologies, it is increasingly demanding for information storage, retrieval. Image retrieval is an important area of information retrieval, how to retrieve image flexible, efficiently and accurately has become one of the current problems to be solved. In order to address this problem, the research of content-based image retrieval emerged. Nowadays, content-based image retrieval technology in the current has become a hot topic. Many research institutions at home and abroad are engaged in the research of this subject, and have made certain achievements.Content-based image retrieval technology firm mainly automatically extract the image object features such as color, shape, texture or a combination of these characteristics, calculate the image features and the target image, the similarity between the feedback out of the search results. Image retrieval based on shape feature in image retrieval based on content occupies an important position, and has been widely research and application of shape feature contains abundant structure information and semantic information, more in line with the characteristics of people’s vision, and the description of the shape characteristics, extraction, etc. It is difficult.Summary of image retrieval algorithm based on the research results at the same time, puts forward a kind of image retrieval algorithm based on shape feature moment: first, studies and analyzes the two kinds of image description method based on shape feature, respectively is based on the description of the boundary and based on the description of the area. Secondly, key research based on the method to describe the shape of the moment, using Hu moment invariant to image translation, rotation and scale transform invariance, respectively to extract the sample images and shape feature vector of each image in the gallery. In the process of put forward a characteristic vector, due to the irregularity of shape features, need to be normalized processing, so as to extract the relatively accurate feature vector. Then, using the improved distance formula to calculate the sample image and gallery shape characteristic vector of each image in the distance, puts forward a can adjust the size of the high and low order moment weight value, according to different experimental results to adjust algorithm, to ensure the accuracy of the algorithm, to meet user demand for the results. Finally, using the simulation algorithm for the geometric transformation invariance and image retrieval performance analysis, the experimental results show that the algorithm has better retrieval effect.