Panthera Tigris Altaica Individual Identification Based On GLCM And Hu's Invariant Moments

Panthera tigris altaica individual identification has important significance for Panthera tigris altaica protection and field training.At present,the number of the wild Panthera tigris altaica all over the world is about 500.Panthera tigris altaica were listed as one of the world's top ten endangered species by the International Wildlife Fund.So it is necessary to study Panthera tigris altaica individual identification.The paper used the image processing technology and pattern recognition theory,adopted the image texture features and statistics to identify Panthera tigris altaica individual identification.Gray level co-occurrence matrix (GLCM) is the classic and sophisticated method to extract statistical features of texture,and Hu's invariant moments are statistical characteristic too,being used to describe the shape of the region.So,the paper offered a new method combined by GLCM and Hu's invariant moments to describe characteristic parameters of Panthera tigris altaica.In order to confirm the parameters of Panthera tigris altaica individual recognition,the paper analyzing the changing rule of(GLCM) feature parameters along with its three building factor including the making step d,the gray-level of image N_g and making directionθ,the way of building GLCM suitable to describe Panthera tigris altaica texture was established combined with its characteristics.The rule was that d equaled to 4, N_g equaled to 256,andθwas took as 0°,45°,90°,135°four directions.The texture parameters value took the average of four directions to form rotation invariants.Calculating 7 different feature parameters on the basis of it,combining 10 shape feature formulae of Hu's invariant moments,choosing 7 features of the 17 by means of correlative modulus analysis as Panthera tigris altaica individual identification last feature parameters.Designed BP neural network suitable for Panthera tigris altaica individual recognition.It's input layer was 7,middle layer was 6 and output layer was 5.At last,simulating and identifying to the BP neural network,recognition rate was 82.5%.It proved the method based on GLCM and Hu's invariant was feasible.