Research On Potato Image Acquisition System And Detection Methods Based On Machine Vision

The research of the potato external quality classification based on machine vision technology is mainly in domestic, so it is unable to meet the needs of on-line detection at present. This paper designed suitable for dynamic image acquisition potato machine vision systems. The main contents of this paper are as follows:1. Delivery device designUnder the potatoes without secondary damage, potato belt conveyor with difference in height is designed. According to take advantage of height difference in the fall of potatoes, the image of potatoes can be captured.2. Machine vision hardware designMachine vision hardware that can capture three images of potato by one shot is designed. The light source of this system is LED strip light, and its background is white. In order to realize the full range of shooting, two mirrors are installed at the bottom of the box.3. Image processing and defect detectionIn the paper, the images captured by camera are pre-treated according to several means, such as B-channel grayscale,3×3 median filter and Otsu. The budded potatoes are also tested based on Euclidean distance algorithm.4. Software DesignMatlab7.0 is used in image processing programming. The captured images are pre-treated and their surface defects are detected by that software.