关键词: 目标识别, 主从分析法, 仿射不变矩, 图像识别

Abstract: For the problem of real-time of target recognition algorithm based on moment is not high, and analyzes the application of the principal component analysis(PCA) in fast target recognition algorithm, this paper proposes the method based on the Jan Flusser affine invariant moments and PCA of fast recognition. By using the PCA in the Jan Flusser affine invariant moment feature space are optimized for dimensionality reduction, the method reduces the amount of calculation, and then some mainstream recognition algorithms are used to verify the real-time and accuracy of the method. The experimental results show that the method of real-time in Euclidean distance is increased 23.68%, the probabilistic neural networks is improved by 8.7%, the support vector machine is improved the accuracy of 21.01%, and only a small amount of reduction in accuracy, it does not change the recognition algorithm of the process. Meanwhile, it suits for three-dimensional object recognition under small angle change.

Key words: object recognition, PCA, affine moment invariants, image recognition