The Application Of Color Image Segmentation In Plant Measurement System

Agriculture is the foundation of our national economy, and the improvement in the efficiency of agricultural production and the extent of agricultural automation are the bases of modern agriculture. The technology of computer vision is conducive to the development of modern agriculture. The plant measurement system contains capturing image of plants, processing image, measuring plants and finally display the result on the screen and store data in the database. How to extract the target-plant from image is the main problem to be solved in this paper.Many typical color models are analyzed based on plant image, for instance HIS, HSV, I1I2I3. According to experiment result the H value in HSV is the most effective feature.On the basis of the experiment result feature H in HSV is used to segment the image, but the result shows that the tip part of the leaf can not be segmented correctly. Therefore another 2 methods are proposed. An evaluation experiment is designed to verify the effect of these 3 methods. The outcome of evaluation experiment indicates that 2D threshold is the most effective and stable method and it can apply to the system. Furthermore, 2D threshold algorithm is implemented and this method is working efficiently in the system.In addition, in order to obtain accurate color information, some color correction methods which are used to solve color difference caused by sunlight are tested. According to the experiments result, curve method is the best one.