Multidirectional Gradient Feature With Shape Index for Effective Texture Classification

1. Introduction

Computer vision and machine learning are critical for image retrieval (Liao et al., 2018), fault recognition, automatic control, automatic driving, ensemble learning (Liu et al., 2018), block chain, vehicle communications, target detection, and tracking (Akhilesh Mohan Srivastava, 2022; Seddik et al., 2022; Thanh, 2022). Image texture has played an important role in computer vision and machine learning since the 1960s (Julesz,1962). It has been widely studied and applied in many fields, such as image segmentation (Li et al., 2018), content-based image retrieval (Zheng et al., 2018), target detection and recognition (Marszalek et al., 2007; Oyallon & Mallat, 2015), biometrics recognition (Ding et al., 2016; Vu, 2003; Xi & Zhang, 2011; Zhang et al., 2009; Zhao & Pietikainen, 2007), computer graphics, and image texture synthesis (Gatys et al., 2016).