The Research On Algorithm Of Detection For Visible Contaminant Particles In Ampoule Solutions

With the improvement of the theory of computer system and the development of microelectronic hardware system, computer vision gradually becomes the main research method of intelligent industrial automation detection. Many excellent properties, such as the unified standards, accurate calculation performance, scientific and efficient work, more in line with the production requirements that is complex testing environment and efficiency of production demanding. Liquid contaminant Particles detection is a hot research problem in current industrial detection of pharmaceutical industry. As a key link, the algorithm theory of moving target detection and moving target tracking is constantly updated development. However, the detection algorithm research of the target characteristics of visible contaminant particles in ampoule solutions is less. Based on the ampoule contaminant particles detection as the research object, this paper deeply studies detection tracking algorithm for moving target, developed a robust, high efficiency and precision of detection tracking system. The main research content of this paper are as follows:1. Based on the theory of dynamic visual processing framework, design a rational detection process system which accords with the environment of ampoule contaminant particles detection, and analyzes the possible interference sources in the testing environment and solutions.2. After analyzed several kinds of moving target detection algorithm, combined with the local threshold method based on integral figure and the frame difference detects ampoule image target. Using mathematical morphology method to fix the target features and connectivity analysis, to provide the target information for moving target tracking.3. The problem for movingt target tracking is deeply analyzed. The matching description for the ampoule contaminant particles and target motion model which accords with the movement characteristics of contaminant particles are discussed. The target motion is choosed as the judgment information for contaminant particles detection.The hypothesis test algorithm which combines with the nearest neighbor data association method and motion prediction is studied as the application of the ampoule contaminant particles detection.