Monday, November 13, 2006

Development of Yield and Quality Estimation Methods for Tea-plant by Image Processing.

Daitaro Ishikawa1, Ishiguro Etsuji2, Shin-ichi Sekioka3, Tsuyoshi Okamoto4, Atsushi Nesumi4, and Kazuhiko Oba4. (1) Kagoshima Univ, 1-21-24 Kohrimoto, Kagoshima, Japan, (2) Japan, 1-21-24 Korimoto, Kagoshima, 890-0065, Japan, (3) Kagoshima Univ., 1-21-24 Korimoto, Kagoshima, Japan, (4) National Agric. and Bio-Oriented Res. Organization, Kagoshima, Japan

This study was conducted to evaluate growth condition of tea plants and to estimate quality of tea leaves at each growing stage by spectral image analysis. The experimental plots were designed for five levels of nitrogen application levels. The spectral characteristics of their canopies were measured by a handheld spectroradiometer from 400 nm to 1,000 nm with 3 nm intervals, in the growth season of new leaves. Those canopies were recorded by a video camera with several band pass filters. Surface temperature of tea leaves were measured with thermo-radiometer and thermo-couples. At the same time, chlorophyll contents in the leaves were measured with a chlorophyll meter. And the picked new leaves were analyzed with NIR analyzer for several chemical contents in the leaves. The spectral reflectance curves were differentiated with respect to wavelength on the first and second orders. As a result, several characteristic peaks were identified. These peaks ranges with +/-10 nm were named characterized waveband, CW. Several indices consisted of these CW’s were proposed and compared with ground truth data. It was demonstrated that the index, ND77,66 =(R770–R660)/(R770+R660), strongly coincided with the results of chlorophyll meter and nitrogen content of NIR analysis. It was shown that the ND77,66, ND77,52, and ND54,52 were in concord with the quality components of live new tea leaves observed by NIR products analyzer. These results suggest that the spectral reflectance measured out of contact with a plant non-destructively is effective in evaluating the leaf growing condition, estimating the leaf quality and determining the most suitable leaf picking day in terms of leaf quality. It can be an alternative to conventional methods using a chlorophyll meter. Moreover, the spectral images taken by a video camera with band pass filters will make it possible to analyze differences of growth stages of plants.

Handout (.ppt format, 3422.0 kb)