Robust Feature Detection Using 2D Wavelet Transform Under Low Light Environment

Résumé du livre

The paper presents a local feature detection method for vSLAM-based self-localization of mobile robots. Extraction of strong feature points enables accurate self-localization under various conditions. We first proposed NAST pre-processing filter to enhance low light-level input images. The SIFT algorithm was modified by adopting wavelet transform instead of Gaussian pyramid construction. The wavelet-based pyramid outperformed the original SIFT in the sense of processing time and quality of extracted keypoints. A more efficient local feature detector and a compensation scheme of noise due to the low contrast images are also proposed. The proposed scene recognition method is robust against scale, rotation, and noise in the local feature space.