A Computerized Vision - Based Method for Automatic Detection of Tea Shoot Tips

Abstract

    For tea producers and tea researchers, counting tender shoots in a sampling area is usually done to evaluate productivity, record the growth of tea or to decide the appropriate time for harvest. However, it is a tedious and time consuming task. In this paper, we proposed a computerized vision - based method for automatically detecting and counting the number of tea shoot tips in an image acquired from a tea field. First, we built a parametric model of a tea – shoot tips color distribution in order to roughly separate Regions - of - Interest (ROIs) from a complicated background. For each ROI, we then extracted supportive (local) features with expectations that these features will only appear around an apical bud of tea shoots thanks to two measurements: the density of edge pixels and the statistic of gradient directions. Consequently, the extracted features were put into a mean shift cluster to locate the position of tea shoot tips. The proposed method was evaluated on a set of testing images at different sites of tea fields and different plant ages. The results showed that the system could recognize tea shoot tips with 86% accuracy. It is, therefore, possible to design a counting-assisted tool for supporting tea producers or tea researchers.