Improved G-K fuzzy clustering segmentation algorithm for rice damaged-spots infested by Rice Leaf Roller

Abstract

Abstract: Image segmentation of crops damaged-spots infested by insect pests under natural conditions is very important to realize the precision spray. Due to the influence of uneven lighting and random noise, the traditional method of image segmentation is difficult to achieve the ideal results. In order to overcome the complications mentioned above, an image segmentation algorithm based on OSTU binarization algorithm and improved Gustafson-Kessell (GK) Fuzzy Cluster for the rice damaged-spots infested by the Rice Leaf Roller (Cnaphalocrocis medinalis Guenee) was proposed in this paper. Firstly, the ultra-green equation and OSTU was utilized for image preprocessing. Secondly, take the S component of color space HSI (Hue, Saturation, Intensity) which transferred from the target image, and then the improved Gustafson-Kessell Fuzzy Cluster algorithm and Morphological Filtering were utilized to obtain the target area which the rice damaged-spots infested by Rice Leaf Roller. Experimental results showed that the accuracy rate of the proposed segmentation algorithm reached 82.4%. In order to test the effects of segmentation results in classification and recognition, three features, skewness of color feature B and R component, average of S component, were selected. The distinguished effect of each features mentioned above were showed good performance. The classification accuracy rate based on the above three features reached 94%. Efficient results were achieved by using the mentioned above method for images with the influence of uneven lighting, random noise and complex background under natural conditions. Keywords: paddy field, precision spraying, rice leaf roller, image segmentation, Gustafson-Kessell fuzzy cluster DOI:  10.33440/j.ijpaa.20190202.51.  Citation: Wang P, Jia G H, Zhou Z Y.  Improved G-K fuzzy clustering segmentation algorithm for rice damaged-spots infested by Rice Leaf Roller.  Int J Precis Agric Aviat, 2019; 2(2): 62–66.