Research Of Tobacco Leaf Image Segmentation Based On Morphological Algorithm

Image segmentation is a basic and vital technology in computer vision and image processing fields. It is also a key step from image processing to image analysis. Since 1960, thousands of image segmentation algorithms have been invented, for example, the algorithms based on image gray statistics information, the algorithms based on pixels classification and the algorithms based on edge detection. The research of tobacco leaf image segmentation based on morphological algorithm belongs to image segmentation by edge detection. The basic idea is detecting marginal pixels of a digital image by morphological algorithm firstly, then link up them to outline by some strategies, the image segmentation is made by this way. In allusion to deficiencies of traditional edge detection based on grade, combined with tobacco leaf image characteristic, contents about this were studied:1. Based on the research of digital image collection and different kinds of digital image files, using CCD digital camera, combined with the application of the research, tobacco leaf vision processing system was worked out.2. Deficiencies of traditional edge detection based on grade were analyzed, and then image border detection algorithm based on morphology including two value mathematical algorithm, grayscale morphological algorithm and fuzzy mathematical algorithm was designed and realized. The results of the experiments were compared in noise and no noise conditions. The validity of tobacco leaf image edge detection based on fuzzy mathematical algorithm was proved.3. After researching the theory of algorithms, with the advantage of oriented object RAD tools in Windows platform, making use of functions and components of Delphi, algorithms was realized and the tobacco leaf image vision processing software was developed.