Research On Tracking Algorithm For The Laser Scattering Rate Of Settling Particles

Tracking of the laser scattering rate of settling particle,or simply “settling particle tracking” for short,is to track the settling particles in the settling equipment.The particle size,which is used to control the material sizein production,process and application,is measured during the tracking process.The settling particle isof great research and application value in material size analysis field.In the research,some of the settling particles can berecognized easily,while the other are hardly distinguishable from the background.Moreover,the settling of particles accompanies the disappearing,merging and splitting of certain settling particles.Otherwise,the present settling particle tracking algorithms are designed to solve the specific problems,which are disabled to settle the problem in this research.Therefore,a particle probabilitybased settling particle detection algorithm is proposed in terms ofthe features of settling particles,and the tracking algorithm is put forward based on the Kalman filter.The main research contents are as follows:(1)Extract and represent the features of the settling particles.Through the observation of particle settling video,the settling particle features are presented in three aspects,including grayscale feature aspect,contourfeature aspect and motion feature aspect.According to the grayscale value change,the centroid and the area,and the displacement and the velocity in the horizontal and vertical direction,the settling particle features are extracted and represented in formulation.The extracted settling particle features is able to describe different particles,distinguishing the settling particles from background.Furthermore,the features make the tacking algorithm capable to track the settling particles of disappearing,merging and splitting.(2)Propose the Haar feature and particle probability based detection algorithm of settling particles is proposed.The algorithm is divided into two steps.At the first step,three Haar features are applied to transform the grayscale values into Haar feature values,so that the signals of the unrecognizable settling particles is enhanced.Secondly,based on the Haar feature values,the probability of each pixel which indicates whether it belongs to a settling particle or not is calculated combined with the contour features.Then,the detection is accomplished after the probability weighted calculation.Haar feature is enabled to highlight the unrecognizablesettling particles,but more noise is introduced after this process.While particle probability is capable to eliminate the noise,advancing the accuracy.(3)Propose the grayscale difference motion vector based settling particle tracking algorithm.According to the detection result,a grayscale difference based motion vector is established combined with the settling particle features.The Kalman filter is applied to predict the position of settling particles,and the Hungarian algorithm is used to match between the predicted particles and real particles.After the match,for the unmatched tracks or particles,the grayscale difference is used to judge the disappearing,merging and splitting state of settling particles.Lastly,the results are corrected and the whole tracks are obtained.The judgment of disappearing,merging and splitting state enhances the reliability of settling particle tracking algorithm.(4)Experiment and analysis of the settling particle detection and tracking algorithm.The validity is verified by the experiments and the comparison with other detection and tracking algorithms.The detection algorithm is compared with Lei Yang's algorithm,the Gaussian mixture model detection algorithm and the Ostu threshold algorithm,the results show that the detection algorithm proposed by this paper is bit worse than Lei Yang's algorithm,but better than the Gaussian mixture model detection algorithm and the Ostu threshold algorithm.Sequentially,the tracking algorithm is compares with the general Kalman filter algorithm and the nearest matching algorithm.The results manifest that the grayscale difference motion vector based algorithm is better than the other two.Therefore,the proposed settling particle detection and tracking algorithms are efficacious.