AN ESTIMATING FLOOR REGION ALGORITHM BASED ON IMAGE SEGMENTATION USING FPGA FOR SMART ROVERS
Chi‐Chia Sun, Nian-Jyun Yang · 2019
By far most of the floor regions detection algorithms are still dependent on the depth or other additional sensors that consume more power relatively and increase the total weigh for smart land rovers.In this paper, a new framework that enable to estimate floor regions in single image for Unmanned Ground Vehicle (UGV) robots is presented.In general case, the cluttered indoor surroundings such as patterned floors, shadows and reflections, those surroundings are very difficult to differentiate floor regions.The proposed algorithm combines extracting surface texture characteristic with specific geometric area is able to find out from object boundary, and through SVM to distinguish between floor and non-floor regions.In experimental results, public MIT Scene dataset and indoor database were selected to verify the detection accuracy.The proposed algorithm accuracy can reach up to 94.72% in average without any other sensors for assistant.