Research On Image Classification Algorithm Based On Multiple Features

With the development of information technology, image classification has been widely used in various fields. Traditional image classification techniques are based on single-feature classification, which affect the classification performance. The current research trend is using multiple-feature information. This dissertation made a thorough research of how to integrate multiple-feature information and proposed an image classification algorithm based on multiple feature channels.Image classification includes three main processes:extracting features, representing image by pattern vector and classification. One of the key factors affecting the performance of image classification is feature extraction. Comparing to global features, local features are more adaptive when the image has blur background, partial occlusion or illumination changes, local features have raised in recent years. Many feature extraction algorithms have been proposed. However, different algorithms performance differently in different circumstances, there is no one performance excellent in any situation. Traditional image classification methods only use single feature for classification, which brings limitations. This dissertation made a thorough research of image classification methods based on feature combination. This dissertation also proposed an image classification algorithm based on multiple feature channels, this algorithm represents image by multiple features, then classifies image by using each feature, and the final classification result will be decided according to the classification information of each channel. Experience has shown that the algorithm performance better than the single-feature-based algorithm.At last, this dissertation summarized the presented work and drawn up a plan of future work.