IMAGE CLASSIFICATION BY PRINCIPAL COMPONENT ANALYSIS OF MULTI-CHANNEL DEEP FEATURE

Currently, image classification plays an important role in the image understanding. Most of image classification algorithms are only to extract the visual feature of the image, that can only describe a small number of attributes of the images, and the description of the image is not comprehensive, resulting in the classification information of the image is not enough, and the distinction accuracy between images is not high. For this, this paper proposes an image classification technology by principal component analysis of multi-channel deep feature. Combining RGB image and depth image can effectively improve the image classification accuracy. In this method, we extract deep feature from the RGB three-channels and the depth information of image, respectively, that the extracted features are of more degree of differentiation, and then we implement the dimension reduction processing through the principal component analysis. Finally, we make use of the Support Vector Machine (SVM) for image classification. Experiment Results show that the promising performance of our proposed method.