MATLAB Implementation of Pedestrian Detection Combining HOG and LBP Features

This system implements a comprehensive pedestrian detection solution using MATLAB, with the core algorithm combining HOG and LBP techniques. The HOG algorithm extracts gradient-based features, capturing object edges and corner characteristics by computing gradient orientations in localized image regions. The LBP algorithm analyzes texture patterns by comparing pixel intensities with their neighbors, generating robust texture descriptors. By integrating both feature extraction methods, the system achieves enhanced detection accuracy through complementary feature representation. The implementation includes configurable parameters such as detection window size, classification thresholds, and feature normalization options to adapt to various surveillance scenarios. Key MATLAB functions involved include feature extraction using `extractHOGFeatures()` and custom LBP implementation, SVM classification with `fitcsvm()`, and sliding window detection with optimized scanning algorithms. This robust system supports performance tuning through parameter adjustment interfaces and includes visualization tools for result analysis, making it a practical and valuable solution for computer vision applications in pedestrian detection.