The Research On Identification Of Weeds Based On Machine Vision

Weed infestation to crop is a question which people pay attention to all the time, at present main weeding method is applying herbicide, but for over-applying herbicide, which not only increases the cost of production but also causes pollution to environment. To adapt to the request and need of precision agriculture, the paper based on computer machine vision system, the main purpose of this study was to identify weed from wheat by using digital image processing and pattern recognition. Specific study was carried on under normal sunlight condition by using digital camera photo in order to identify weed from wheat. The weed identification system had good effective on identifying weed.First, the paper applying color feature increased the contrast between green plants and soil, separated the plants from complex background and changed real color image to gray-level image. The paper compared many algorithms during image pretreatment, applying neighborhood filtering method and fast median filtering method to filter images and gray-level image and eliminated effects of all kinds of noise to images. Secondly some algorithms of image threshold segmentation were researched and selected a suitable segmentation method, in the paper the author put forward an improved threshold segmentation, which is based on the Otsu threshold selecting method. The algorithm could transfer gray-level image into binary image with stabilization and no distortion. And mathematical morphological method was used to deal with the noise in binary images. Several weeds commonly found in the wheat field are used to test the system, according to the features of wheat and weed, the author studied four nondimensional shape features, including area, perimeter, compactness, centroid. The results showed that compactness is the best feature to identify broad-leaved weeds and angustifoliate weeds, and the system has good effective for different complex background weed images.The aim of the research is to provide essential theoretical evidence and technological basis for the further development of herbicide use which is valuable in environment protection. The results of this paper attach great importance to reduce the gap between China and developed country in the area of agricultural information automation technology. It accelerates the application of computer image processing technology in agriculture engineering field.