Vision Based Landing Pad Detection for Autonomous Landing of Quadcopter

Downward facing monocular camera (Raspberry Pi camera module) is mounted on the MAV. Algorithm for helipad landing target detection have been implemented on Raspberry pi3 single board computer using OpenCV and Python programming language. In the image preprocessing stage, to preserve sharp edges and to remove noise Bilateral and Median filters have been applied respectively. Next, Adaptive threshold method have been used to segment the foreground image pixels from the background image pixels in the acquired landing pad image. Edges in the landing pad images are detected in the foreground image by detecting strong edge pixels with canny edge detection method. Furthermore, Centroid is calculated using moments of binary image. Experimental result shows that the proposed vision based algorithm is suitable to detect the helipad marks for autonomous landing of MA V.