Research Of Image Retrieval Based On Shape Analysis Technique

Information retrieval is one of the important research contents in the field of information science.With the rapid development of digital information technology and the widespread popularity of internet communication,many new research topics have generated in the field of multimedia information retrieval.As one of the important form of multimedia,digital image contains rich information that is intuitive to be recognized by human visual system.Nowadays,Image is the most frequently used information form besides text.With the development of image generation equipment and the popularization of online image sharing services,the scale of image information resources grows at an unprecedented speed,accompanying the ceaseless expansion of image database composed of massive image information.At the germination stage,the technique of content-based image retrieval(CBIR)develops rapidly due to the rising tide of digital library.Nowadays,CBIR has attracted growing attention for its wide application domain and high application value.In methods of content-based image retrieval,visual contents are used as image features in the match and retrieval stage.Compare with other perceptual features such as color and texture,shape can characterize the perceptual content of an image more effectively and provide more valuable visual cues to the identification of similar objects.In this thesis the image segmentation and description approaches are discussed around the topic,that including three key techniques:shape segmentation,contour analysis and feature fusion analysis.In the process of shape segmentation,an adaptive edge detection algorithm based on morphology is proposed according to the application needs of shape analysis in image retrieval.The algorithm structures an edge detection operator based on morphological filtering,uses multi-direction structure elements to detect edge information accurately,and adaptively adjusts the weights of different structure elements.And then,reasonably adjust the size of the structures.The results of simulation demonstrate that the adaptive edge detection algorithm gets better effect than classical differential operator and common morphological methods.The theory proposed not only suppresses the interference of noise effectively,but performs better in extracting the edge information of different directions.In the process of contour-based shape analysis,the descriptor of image contour should reflect the information of global shape and local feature.Also the description should be robustness to random noise.In this paper,we propose a new image retrieval method based on contour reconstruction and feature point chord length.First,the contour of the shape is extracted,and in order to reduce the distortion caused by random noise,the contour is reconstructed by analyzing the energy retention rate.Then,on the base of the new defined supportive region,the feature intensity is calculated at each point of the contour to extract the valid feature points.After that,the contour feature function is structured by using the chord length between contour points and corresponding feature points.Finally,the shape descriptors are processed to meet the invariance property.A significant amount of experiments show that,in both normal and noisy sample sets,the proposed method demonstrates better performance compared with other seven techniques.In the process of fusion feature based shape analysis.The description of visual feature is the most important task in specimen recognition.The descriptor should reflects the feature of both global and local regions,and be robustness to random noise.Based on the research of shape feature in specimen recognition,in this paper,a description of specimen using combined visual features is proposed.The improved Hu moment invariants eliminate the data redundancies,and enhance the detailed information.In contour processing,the approximate polygon is used to eliminate the data redundancies.The specimen contour is described by a function based on feature points.Then the description function is analyzed by using Fourier transform,which makes the descriptors more efficient and compact.The experimental results reveal that the proposed approach outperforms other algorithms in both animal specimen database and leaf specimen database.