Sand Jump In The Trajectory Of The Computer To Recognize

This thesis is a part of the project "Theoretical and Experimental Study On the Motion of Near-bed Non-uniform Sediment "sponsored by National Natural Science Foundation. Our objective is to process the image sequences of saltating particles taken by video recorder, and to give the motive curve of the particles for the further study on the collision between particles and sediment, and the model of saltating particles.For each image, based on Histogram Modification, we use Wavelet Transform to eliminate the disturbance of background. Then we use Thresho]ding, Edge Extraction, Hough Transform, and Thinning to get two part-images. One is the area that is above the ruler, the other is the ruler itself. We recognize the particle in the first part-image by template matching. in the meantime. we use Proj ective Positioning to find numbers on the ruler. To identif~~ them, we put their characteristic vectors to the BP Neural Network that we have trained. At last, with the position of the particle in one image and the numbers identified, we can tell the exact position of the saltating particle in the image sequences, and give the motive curve of it.