A performance comparison of feature detection methods using wide-angle lens for 360°panoramic image

In this paper we compared and assessed the processing speed of some feature mapping methods regarding 360-degree panoramic image generation. We used a wide-angle lens with a wide FOV to reduce the number of cameras. This results in reduction of hardware complexity and processing time in feature extraction and matching as well as reduction of the region of interest in images. In order to compare feature extraction methods, we analyzed the methods SIFT and SURF, and perspective-SURF. Analysis showed that features extracted by perspective-SURF yielded 1.3 times higher accuracy than other methods, This paper gives emphasis on establishing a 360-degree panoramic system with the fast and accurate feature extraction PSURF approach that reduces cost with fish-eye lenses.