Research Of Shoeprint Feature Extraction And Retrieval Algorithm

Shoeprints,as the four biological evidences of the Ministry of Public Security,play an important role in the detection of cases and the identification of criminals,and the retrieval of pattern images of footprints is a key link in the analysis of footprints.On the one hand,the search for footprint images can be used for similar case cases to investigate together;on the other hand,information can be searched for suspects and even used as an intermediate bridge to connect different sites.Therefore,the efficient retrieval of footprint images has become one of the key issues that need to be solved in the analysis of footprints.Due to the influence of the surrounding environment and the extraction process,footprint images have a lot of noise and interference,such as uneven edges,distortion of image textures,and large areas of image defects.In addition,the commonly used local features are more sensitive to noise and do not represent the texture and shape information of the live pattern image.Therefore,when the image retrieval algorithm based on the traditional bag of visual words model is applied to the retrieval of footprint pattern images,this paper has combined the characteristics and existing problems of the footprint pattern image to improve the algorithm and proposed a new local features.the main work is as follows:(1)A new local feature based on Sobel operator is proposedAccording to the characteristics of the footprint pattern image itself,this paper presents a new local feature: edge direction features.The edge direction feature can effectively extract the gradient information of the texture image based on the Sobel operator,calculate the gradient direction information of the local area by the least square method,and perform histogram statistics,finally obtain the local feature.Experiments show that this local feature can effectively improve the experimental results compared to the previously used SIFT features,and so on.(2)Footprint pattern image retrieval algorithm based on bag of visual word modelAiming at a large number of image defects in the footprint images of the site,this paper proposes an algorithm that can effectively solve the problem of excessive calculation error of the incomplete image similarity,incomplete template algorithm.The degree of picture incompleteness is judged according to the duty ratio of the feature.When the global feature is encoded,an incomplete template is generated.In the similarity calculation,the incomplete template can be used to effectively reduce the influence of the incomplete image on the retrieval result.The traditional image retrieval algorithm based on the visual word bag model has a wide range of applications.In this paper,the retrieval algorithm based on the visual bag model is applied to the retrieval of footprint pattern images,and the algorithm is improved according to the characteristics of the footprint pattern image itself..In this paper,two traditional visual word bag-based image retrieval algorithms are successfully applied to footprint pattern images.It is sparse constrained pyramid matching(SCSPM)algorithm.Compared to other footprint pattern image retrieval algorithms,the experimental results have been improved.