Image analysis of grains from a falling stream

As a engineering requirements we initiated a study using an on-line computer system to analyse grains in a gravitational now, i.e. a falling stream of particles. Compared with images from a moving conveyor belt, in gravity flow both grains and background appears more and less afflicted by overlaps. This, apart from alleviating the problems of overlapping, implies that the size and shape of grains can be measured more accurately, but one problem is how to resolve blurring. To improve the machine efficiency of the particle detection and classification system, the samples must be examined dynamically. The dynamic detection leads to an unavoidable result of image motion blur, which greatly degrades the quality of the image. So we build up the mathematics model of motion-blurred image, estimating Point Spread Function, introduce a high speed and efficient way to solute this problem and restore the blurred image.