A Modified Fuzzy Color Histogram Using Vision Perception Difference of Pixels Location

Semantic enhanced deep learning for image classification

“…The main idea is to model the image feature vector and text annotation containing semantic information as a nonparametric Gaussian kernel model problem. Zhao et al proposed the combining color histogram and texture histogram as image visual features and then applied the fuzzy mean multi‐class method to realize the mapping from visual features to semantic features. Considering the large number of branches in a binary tree construction, a linear discriminant analysis is utilized to optimize the features before clustering, which improves the accuracy of the mapping method.…”