Research On Segmentation And Thinning Algorithm Of Finger Vein Image

As an emerging identification technology in biometric identification,finger vein recognition has the characteristics of high security and strong confidentiality,and has good development prospects and application market.However,due to the high complexity of the feature information in the finger vein image,the acquisition process is easily interfered by factors such as illumination and strength,and the identification work cannot be directly performed.In this context,this paper focuses on finger vein image segmentation,finger vein image refinement and feature extraction in the process of finger vein recognition.The main work is as follows.(1)Finger vein image pretreatment.The principle of collecting finger vein images is briefly explained,and the key factors affecting image quality are explained.Grayscale normalization,enhancement and other processing are performed on the collected finger vein images,and the finger vein structure feature information is further highlighted in the image,which initially satisfies the standard requirements for subsequent segmentation work,and provides a good foundation for subsequent work.(2)Finger vein image segmentation method based on improved Niblack algorithm.Finger vein information is more complicated,and there are a large number of bifurcation structures.These features make the segmentation algorithm difficult to achieve the desired effect.Aiming at this problem,the advantages and disadvantages of Niblack algorithm are analyzed emphatically.According to the distribution characteristics and details of finger veins,a finger vein image segmentation method based on improved Niblack algorithm is proposed.The automatic adaptation method of correction coefficient k is adopted to improve the accuracy of threshold.Calculating and analyzing the change information of the gray of the global finger vein image,and combining the expected information between the respective local pixel points,calculating the value of the correction coefficient k and the image segmentation process by calculating the expected value and the standard deviation in the r×r templateWhile retaining finger vein information,reduce redundant background information and suppress noise interference.(3)Finger vein image refinement algorithm based on improved Hilditch algorithm.Based on the principle that the Hilditch refinement algorithm can't delete outliers and incomplete refinement results,based on the principle of the classical Hilditch refinement algorithm,the consideration of deleting pixels is expanded from its eight neighborhoods to include the current pixel points.The eight neighborhoods and the neighboring pixels of the eight neighborhoods total twenty-five pixel regions,and the decision constraints are re-submitted according to the distribution of pixel points in the region and the distribution characteristics of the vein information,and the refinement work is performed according to the new constraint conditions.(4)Feature extraction and matching method based on the shape of the finger static skeleton.The number of pixels in each finger vein skeleton is different,and it is impossible to make one-to-one correspondence with very accurate and error-free.The distribution of each pixel in the neighborhood of each pixel on the finger vein skeleton is defined as the skeleton feature of the finger vein,and the similarity between the two finger vein skeletons is defined and calculated,and the two finger vein skeletons are As the difference between the two,the distance between the two skeletons is obtained by inversion,and the value is used as a matching basis.(5)Finger vein feature extraction method based on direction detection.The traditional method mainly extracts features based on high-dimensional data in image feature space,but ignores the determination of pixel point lines,which leads to the problem of low feature extraction accuracy.A finger vein feature extraction method based on direction detection is proposed.According to the original signal and the discrete signal of the finger vein image,the image is wavelet transformed,and the wavelet transform image is reconstructed by the low-pass and high-pass filter of the wavelet function,and the image signal and the quantized image threshold are hierarchically calculated.The image is calculated by using the gray value and the number of directions of the image to determine the pattern of the image pixel.The feature point of the image pattern is calculated by using the detection point to realize the feature extraction of thefinger vein image.The finger vein image segmentation method based on the improved Niblack algorithm,the finger vein image refinement algorithm based on the improved Hilditch algorithm and the finger vein feature extraction method based on direction detection are tested in the development environment of MATLAB,and compared with the existing algorithms..The results show that the finger vein image segmentation method based on the improved Niblack algorithm can achieve the segmentation of the finger vein image,and the running time of about0.6 seconds can meet the requirements of rapid stability.The finger vein image refinement algorithm based on the improved Hilditch algorithm performs a good processing on the redundancy of the pixels in the image,and the number of forks is small,which can basically maintain good consistency with the original image structure,0.18 seconds.The running time of the left and right is completely satisfactory.The finger vein feature extraction method based on direction detection can effectively extract the features of the image,and the accuracy is always between 50% and 90%.