The Study Of Shoeprint Recognition Method Based On Image Processing Technology

Footprint recognition based on image processing technology is an important research direction in the field of biometric identification.In recent years,shoeprint recognition technology has been widely used in the field of criminal investigation.Shoeprint recognition technology is the use of image processing technology to collect,identify and analyze the traces of shoe marks left in the scene of the case,and thus provide the possibility for the detection of the case.This paper mainly applies image processing technology to the recognition of flat shoeprints.The main research work includes:Firstly,the filtering algorithm of shoeprint image based on hybrid filtering is studied.The basic idea is to divide the pixels in the shoeprint image into three categories by setting thresholds:pixels contaminated with salt and pepper noise,pixels contaminated with Gaussian noise,and pixels not contaminated by noise.According to the different characteristics of salt and pepper noise and Gaussian noise,the algorithm selects filter templates of different sizes,and uses the combination of median filtering algorithm and mean filtering algorithm to denoise the noisy shoeprint image.The signal-to-noise ratio improvement factor ISNR is used as the determination index of the filtering performance.Experiment shows that the filtering method has better noise reduction effect than the single-use median and mean filtering in processing the shoeprint image.Secondly,the shoeprint recognition algorithm based on gray level co-occurrence matrix is studied.By studying the definition of gray level co-occurrence matrix and its eigenvalues,it is found that the calculated feature values can reflect the texture characteristics of the shoeprint image,and the feature value can be used as the basis for shoeprint recognition.The algorithm calculates the average value of the four gray level co-occurrence matrix feature values in different directions of the shoeprint image,and uses the calculated average value as the feature of the shoeprint image,and uses the Euclidean distance to complete the shoeprint recognition experiment.The experimental result proves that the shoeprint image recognition based on the gray level co-occurrence matrix is effective.Thirdly,a shoeprint recognition algorithm based on improved LBP is proposed.Aiming at the problem that the traditional LBP algorithm directly extracts the whole image and easily ignores the local detail features,an improved LBP feature extraction algorithm is proposed.The algorithm firstly divides the shoeprint image according to the biometric proportion,then extracts the LBP histogram feature of each shoeprint image,and then superimposes the histogram of each block as the histogram feature of the final shoeprint image,and finally adopts the Papper.The distance is calculated to complete the recognition experiment of the shoeprint.Different LBP models are applied to shoe-print matching,including LBP model,Block LBP model,Uniform LBP model and Block-Uniform LBP model.The differences among different models are compared by calculating recognition rate and recognition time.The experimental result shows that the recognition rate based on the Block-Uniform LBP algorithm can reach 95%,which is 5%higher than the traditional LBP algorithm.