Human Detection Using Wavelet-Based CS-LBP and a Cascade of Random Forests

摘要: In this paper, we propose a novel human detection approach combining wavelet-based center symmetric LBP (WCS-LBP) with cascade of random forests. To detect regions, first extract three types WCS-LBP features from scanning window wavelet transformed sub-images to reduce the feature dimension. Then, extracted descriptors are applied forests, which ensembles decision trees. Using forests WCS-LBP, is performed in near real-time, and accuracy also increased, as compared combinations other classifiers. The proposed algorithm successfully various non-human images INRIA dataset, it performs better than related algorithms.