Skip to Main Content
Wavelet moment features of image can reflect the image's part and whole characteristics and have strong anti-jamming ability. We use wavelet moments extracted from Paper-cut patterns to get multi-scale features. Combined with the paper-cut images' characteristics, the different mean and standard deviation of eigenvector are used to compute resolution and produce N class model feature selection. Finally, the eigenvectors are sent to nearest neighbor classifier for recognition. Experiments show that this method is effective in distinguishing paper cut-cut patterns with noise contamination or geometric deformation.