Research On Segmentation Technology Of Crop Leaf Image Based On The Multifractal Characteristics

Leaves are one of the most important organs of crop, and their morphology and characterization can express the pests and lack of elements in the crop. At the same time, the changes in the shape is smaller than that in the root and stem, so its texture feature is the ideal object of crop growth status. However, at present there’re not enough effective tools to describe the texture feature of images, which was a problem urgent to be solved. The multi-fractal theory is an important method to describe the texture feature of images, and applied perfectly in the field of image processing. According to the leaf image singularity, this paper was proposed to describe the multi-fractal characteristics of leaf image with the multi-fractal theory, and to study on the crop image segmentations based on above features.The paper aims to lay the theoretical foundation for leaf nutrient deficiency and disease detection system through machine intelligence.1.The paper has realized the extracting method based on local multifractal capacity measuring texture feature of rapeseed diseases and insect pests of the leaf, and has further realized the image segmentation technology. The experiment shows that such segmentation method is very sensitive to the edge, vein and disease area of the leaf, thus it can segment the disease area relatively well. When we segment those different disease images with multifractal spectrum based on different capacity measurement, we get different effects. However in general, the multifractal spectrum segmentation method based on capacity measurement can detect the key disease region accurately.2.the paper proposed a new two-dimensional multifractal detrended fluctuation average analysis (2D MF-DMA) based on detrended fluctuation average analysis. Then we applied the new segmentation method on potassium and magnesium deficiency rape leaves images segmentation to verify the validity of the method.