Stamen movement detection of Commelina purpurea based on feature matching

Abstract

The stamen movement is very important to reproduction of plant. This paper presents a feature-matching-based method to detect stamen movement of Commelina Purpurea using image processing. The regions of stamen were firstly extracted by image segmentation using thresholding. Then, the area discrimination factor of each stamen region was calculated to judge whether these stamens were adhered or not, if such adhesion did exist, the adhered stamens were segmented by watershed algorithm. After that, feature detection and matching were carried out between two adjacent frames in the image sequence in order. Then, the slope and distance statistics between the matching feature points are maded to further optimize the matching results and figure out the matching relation of different stamens in the two frames. Finally, through the matching result of different stamens in consecutive frame images, the moving trajectory of the stamens was obtained. The empirical results show that, this method can accurately segment the adhered stamens and acquire stamen trajectory of Commelina Purpurea.