Banknote Dirty Degree Identification Method Based on Texture Features of Banknote Images and Multi-layer Support Vector Machines

The dirty degree of banknots determines whether them can continue to circulate to some extent. In this paper, a banknote dirty degree identification method based on texture features of banknote images and multi-layer support vector machines (MLSVMs) was proposed. Based on the contact image sensor, the double-sided reflection images of banknotes in red, green, blue and infrared light, as well as images in green and infrared light transmission are collected. By analyzing the images formed by different dirty grades of banknotes under various light sources, the green reflection positive and negative images, blue reflection positive and negative images, green light transmission images and infrared light transmission images are finally selected for image texture feature extraction. Based on the gray level co-occurrence matrix(GLCM), the texture features such as energy, entropy and inertia were extracted to describe the visual features of banknotes dirty degree. Finally, MLSVMs was used to identify the dirty degree of banknotes, and simulation experiments results show the effectiveness of the proposed method.