Research On The Method Of Tree Image Division Based On Matlab

Abstract: It is difficult for us to segment the tree image because of its background complexity. The paper focused on research on the methods of tree image segmentation. quickly and accurately so as to lay the foundation for precious toward-target spraying and plant-protection machinery design in the forest. Static color tree images were regarded as samples and 3 methods were employed to segment the tree images based on Matlab. Because the value of excess-green factor of green plant is usually much larger than that of the artificial objects in the background, the excess-green factor was choosed as the threshold of segmentation. Experiments showed that the tree image could be segmented with the method based on excess green factor on the color space of RGB. On the HSV color space, nearly equal good results could be obtained by directly setting the green chromaticity bandwidth combined with the saturation and brightness of the tree image. On the basis of researches, the fractal theory was applied in the segmentation of tree image combined with texture features of the image. The tree images were segmented with the threshold method after box-counting fractional dimensions of obtained gray scale image based on excess-green were calculated and fractional gray scale images were formed. Experiments showed that the method is practical in the segmentation of tree image. At last, 3 methods were compared in both the precision and the real time of tree image segmentation. The results showed that the method of excess green feature applied in segmentation of tree image is suitable for precise pesticide application and plant protection machinery design.