Mobile-Based Flower Recognition System

Conventional flower or leaf recognition studies have some restrictions and limitations. These include a sharp drop in recognition rate due to the varying positions and number of relevant objects in the original object image. Hence, this paper suggests and implements a mobile-based flower recognition system using difference image entropy (DIE) and contour features of the flower from the original image with multi-flower objects. In system framework includes (1) WiBro Net.-based transmission and designation module of the relevant flower object by drawing the flower region of the user´s interest, (2) contour feature extraction module, and (3) DIE-based flower recognition module. The system was evaluated using ten species of flowers with each ten samples. Experimental results achieved an optimum average recognition rate of 95% and average response run-time of 9,033 ms, for a set of ten images per species.