Research on Multi-feature Fusion for Support Vector Machine Image Classification Algorithm

Considering that the traditional machine learning algorithm mainly relies on manual feature extraction for image classification, the classification results are often not ideal. This paper proposes a front-end optimization method based on multi-feature fusion. Extracting image features uses fusion mode of Scale-Invariant Feature Transform (SIFT) and Histogram of Oriented Gradient (HOG) and forms a new feature set HOG-SIFT. Classification results are obtained by training with Support Vector Machine (SVM). Results show that the fusion of new features has a higher precision and recall than single feature extraction.