Simulation Analysis of Feature Extraction and Recognition for Oil Paintings

摘要

This paper analyzes and introduces global features and local feature extraction of characterized painter's artistic style. Then it integrates features and trains recognizer to construct a model for oil painting feature extraction and recognition. Since any one kind of feature cannot singly and completely express image content, data fusion will be firstly performed, the next one is decision fusion and finally these two fusions will be combined to form the recognized framework. The experiments show principal component-leveled fusion analysis of characterized data promotes images to be expressed by characterization. On the other hand, it reduces feature dimension to eliminate redundant features. Compared to single classifier, it shows higher recognition rate.