Skip to Main Content
In this paper, a novel relevance feedback algorithm based on SVM is proposed for 3d model retrieval. It aims to enhance retrieval accuracy in 3D model database systems. During the retrieval process, the system learns from the related samples marked by the user after each feedback, and update the training sample set with the previous returns. Thus an SVM classifier model is established and improved iteratively to retrieve. This method has strong generalization ability when the number of samples is small. In addition, this paper compares the performance of SVM with different kernel functions and the performance of SVM with the same kernel function using different low-level features.