Implementation of LDP Histogram Calculation Based on B Zhang's Research

This code implements the LDP histogram calculation based on the research by B Zhang, Y Gao, S Zhao and J Liu, titled "Local Derivative Pattern Versus Local Binary Pattern: Face Recognition With High-Order Local Pattern Descriptor," published in IEEE Transactions on Image Processing, February 2010, Volume 19, Issue 2, pages 533-544. The key algorithmic enhancement involves computing higher-order directional derivatives compared to conventional LBP methods, which typically use only first-order comparisons. The implementation specifically handles the computation of local derivative patterns in multiple directions (0°, 45°, 90°, and 135°) to capture more detailed texture information. The code structure includes functions for neighborhood pixel sampling, derivative calculation, pattern encoding, and histogram generation. This approach demonstrates superior performance in facial recognition tasks by capturing more discriminative facial features through higher-order local pattern analysis. Research results indicate that the LDP histogram with high-order local pattern descriptors achieves significantly better performance in face recognition compared to traditional local binary pattern methods.