An Improved LBP Feature Extraction Algorithm

In this paper, an improved LBP (Local Binary Patterns) feature extraction algorithm is proposed. According to the traditional LBP algorithm for noise sensitivity, an improved median filtering is adopted. For the LBP operator, the neighboring neighbor points are ignored when extracting feature information. The shortcomings of the relationship, the LBP algorithm is improved, firstly according to the distance between the neighborhood point and the central pixel point, the new field pixel value is obtained, and then the central pixel is compared with the new neighborhood pixel gray value. The operation takes its absolute value, and finally finds the average of all absolute values as an introduced threshold. The threshold improves the parameters of LBP encoding, and uses the improved algorithm to extract the railway fasteners features. Experiments show that the improved LBP algorithm extracts the railway fasteners. The effect is better.