Video Stabilization and Image Processing for Agriculture Applications using UAVs

Abstract

With the advent of technology, Unmanned Aerial Vehicles (UAVs) are emerged as the helping hand for taking care of large fruit cultivation land which involves a substantial number of workers for plucking the fruits, checking their condition of the fruits in the tree. The present work focus on counting the number of ripened fruit in the farm using captured video through UAVs. The methodology adopted is first to stabilize the video, which is obtained using the UAV to cancel out the undesired mechanical movement in the video. Then stabilized video is used for fruit detection. In fruit detection, the HSV of the fruit to be detected is searched in each frame of the video. Edge detection techniques are used to define a clear-cut edge. The blob detection method is implemented to obtain the number of fruits in the frame. Each frame is then combined back to form a video sequence. The result obtained gives the number of fruits available on the tree and thereby decreasing the manual work. It also provides the size and heath of fruit into consideration. The farmer can visualize the farm and its status from his house and at minimal expense.  The fruit count method using the video obtained from the UAV provides 90% accuracy.