Image Recognition Based on Integration of Blur and Affine Invariants of Radial Krawtchouk Moment

According to the information redundancy of non-orthogonal moment invariants, and the discrete error of continuous orthogonal moments, in order to improve the recognition accuracy, a new image recognition method using radial Krawtchouk moment invariants proposed in this paper. The image blur and affine transformation mathematical model is built, then, using Krawtchouk polynomials which has better local characterization capabilities built radial Krawtchouk moments and blur and affine invariants are derived. The experimental results show that, when compared with other mixed invariants based on the orthogonal moment, the blur and affine blend invariants based on Krawtchouk moment can get better invariance under the blend deformation, thus the information redundancy will not be resulted in and also having good robustness to noise, in addition, such image recognition algorithm has higher recognition rate than other algorithms, especially in the case of images with larger noise.