Study of Recognition Method of leguminous Weed Seeds Image

This paper provides a sorting method for the identification of leguminous weed seeds by using clustering. The image processing methods are made use of threshold segmentation, smooth processing and edge detection etc, as well as the image processing algorithm is delivered. According to the stability and genetic characteristics describes in phytotaxonomy for leguminous weed seeds, the feature set includes the aspect ratio between the seed and the umbilical on: the major axis, surface area, perimeter, position and angle between the center points. Although the images are rotated, shifted or zoomed, the visual effect will keep invariance. A tentative auto-recognizing clustering method is adopted, the method may enhance the recognition ratio through learning. Experiments show that the method is effective.