Object detection in images with low light condition

Abstract

Images acquired by computer vision systems under low light conditions are characterized by the existence of noises. As a rule, it results in decreasing object detection rate. To increase the object detection rate, the proper image preprocessing algorithm is needed. The paper presents the image denoising method based on bilateral filtering and wavelet thresholding. The boosting method for object detection that uses the modified Haar-like features which include Haar-like features and symmetrical local binary patterns are proposed. The proposed algorithm allows increasing object detection rate in comparison with Viola-Jones method for a case of face detection task. The algorithm was tested on the two image sets, Yale B and the proprietary – VNTU-458.