Welcome to the InfoSci Platform
Reference Hub3
Multidirectional Gradient Feature With Shape Index for Effective Texture Classification

Multidirectional Gradient Feature With Shape Index for Effective Texture Classification

Xi Chen, Jiangmei Li, Yun Fei Zhang
Copyright: © 2022 |Volume: 18 |Issue: 1 |Pages: 19
ISSN: 1552-6283|EISSN: 1552-6291|EISBN13: 9781799893967|DOI: 10.4018/IJSWIS.312183
Cite Article Cite Article

MLA

Chen, Xi, et al. "Multidirectional Gradient Feature With Shape Index for Effective Texture Classification." IJSWIS vol.18, no.1 2022: pp.1-19. http://doi.org/10.4018/IJSWIS.312183

APA

Chen, X., Li, J., & Zhang, Y. F. (2022). Multidirectional Gradient Feature With Shape Index for Effective Texture Classification. International Journal on Semantic Web and Information Systems (IJSWIS), 18(1), 1-19. http://doi.org/10.4018/IJSWIS.312183

Chicago

Chen, Xi, Jiangmei Li, and Yun Fei Zhang. "Multidirectional Gradient Feature With Shape Index for Effective Texture Classification," International Journal on Semantic Web and Information Systems (IJSWIS) 18, no.1: 1-19. http://doi.org/10.4018/IJSWIS.312183

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

Recently, local gradient microstructure of image textures has become an important field of texture classification, but it is generally to investigate the multiscale local microstructures of image gradient, and rarely consider the multidirectional and multiscale local microstructure of image gradient. The proposed algorithm first extracts the two-order gradient feature of the image from different orthogonal directions and further constructs the multiple shape index of the image, and then calculates the histogram feature vectors of the shape index on different orthogonal directions and scales, and finally connects all histogram feature vectors on different orthogonal directions and scales to obtain the final matching feature vector of the image. To further enhance the discriminant ability of feature vector generated by multidirectional shape index schemes, the weight of each block of images is also considered. Experiments on two texture databases and one palmprint database have fully confirmed the effective of proposed algorithm.