This Letter introduces an efficient human detection method in thermal images, using a center-symmetric local binary pattern (CS-LBP) with a luminance saliency map and a random forest (RF) classifier scheme. After detecting a candidate human region, we crop only the head and shoulder region, which has a higher thermal spectrum than the legs or trunk. The CS-LBP feature is then extracted from the luminance saliency map of a hotspot and applied to the RF classifier, which is an ensemble of randomized decision trees. We demonstrate that our detection method is more robust than conventional feature descriptors and classifiers in thermal images.

Abstract. We propose a novel human detection approach that combines three types of center symmetric local binary patterns (CS-LBP) descriptors with a cascade of random forests (RFs). To detect human regions in a lowdimensional feature space, we first extract three types of CS-LBP features from the scanning window of a downsampled saliency texture map and two wavelet-transformed subimages. The extracted CS-LBP descriptors are applied to a three-level cascade of RFs, which combines a series of RF classifiers as a filter chain. The three-level cascade of RFs with CS-LBPs delivers rapid human detection with higher detection accuracy, as compared with combinations of other features and classifiers. The proposed algorithm is successfully applied to various human and nonhuman images from the INRIA dataset, and it performs better than other related algorithms.

No abstract

No abstract