earticle

영어

Development of the automation system for recognizing diseases of the infected rice is a growing research field in precision agriculture. So, the first and foremost thing we should do is to extract the disease region from rice images. The objective of this paper is to propose an image segmentation method for rice disease spots based on machine vision. The algorithm consists of two main steps: image gray-level transformation and disease region segmentation. Firstly, the color image was transformed into a gray-level image by the color indices 2G-R-B, which making an important contribution for this kind of images. Secondly, the information fusion between the self-adaptation threshold which was obtained by the mean and standard variance of the grey-scale image and the green component distribution features in color image was used to form a new segmentation standard to detect disease region. To test the accuracy and robustness of the proposed algorithm, it was tested with a broad of set of images and compared with the classical approach based on other grey-level convert methods and Otsu’s method. Test result shows that the accuracy of new algorithm appears higher and it can be applied to segment rice disease spots effectively.