Image classification based on improved random forest algorithm
Weishi Man, Yuanyuan Ji, Zhiyu Zhang · 2018
An improved random forest node splitting algorithm is proposed in this paper for improving the accuracy of image classification. By recombining the mode of attribute splitting in random forests ID3 and CART that a new splitting rule is obtained. Then throughing adaptive parameter selection, the optimal selection of attributes can be obtained and should be used for image classification. Based on the Bag of Word, the Spatial Pyramid Model is introduced to extract the image features and quantify them into visual words. Finally, the improved algorithm is applied to image classification on Spark. The experimental results show that by selecting the optimal coefficient of the combined algorithm, the algorithm can effectively improve the accuracy of image classification and ensure the efficiency of the algorithm.