A Novel Texture Classification Method Based on Neutrosophic Truth

Abstract

Texture analysis is one of the basic procedures used in solving problems in computer vision and image processing. In this study, we propose a new local binary pattern (LBP) method based on neutrosophic set. The proposed method is named as the NZ - LBP. In the proposed NZ - LBP method, the texture image is converted into a neutrosophic set and the texture image is expressed by truth membership set. The local binary pattern features are calculated, by using the neutrosophic truth set instead of the original input image. The neutrosophic membership sets are more resistant to noise than the original input image. The neutrosophic set suppresses noise components, so that edge information can be calculated more accurately. Thus, utilization of the neutrosophic truth set instead of the original image has provided more effective local binary pattern features. The proposed method is able to achieve high classification accuracy with low feature size, reasonable computational cost. Experimental results show that the proposed method increases the accuracy of the local binary pattern method to the classification by approximately 11% without increasing the feature dimension. The obtained results reveal that the proposed method is applicable for real-time applications.