Advancing Accurate Depth Estimation and Perception with Angled Stereo Camera
Xiao Da Terrence Fu, Sutthiphong Srigrarom · 2023
This work explores the use of angled stereo vision with low-cost cameras to increase the field of view for an improved situational awareness. While it increases the perception capabilities in computer vision applications, this work also advances the accuracy of depth estimation. Traditional stereo camera setups rely on a pair of parallel optical axes, which consists largely of overlapping regions that results in a wastage of both hardware and processing resources in capturing similar views. By reducing the overlap in the images captured, more data can be captured from the surrounding for analysis. The angled stereo camera is capable of capturing more unique information as compared to the regular stereo camera alternative with the same hardware, improving situational awareness.