Research On Image Quantum Median Filtering And Tracking Algorithm

Computer vision and image processing is considered to be one of the most promising subjects,It has an indispensable importance in the realization of artificial intelligence.Before the processes of images and video sequences,it is necessary to have some pretreatment process,so the median filtering algorithms are studied at first and an adaptive dual qubit state median filtering method is proposed in this paper,after that in order to improve the tracking performance of the continuously adaptive Mean Shift tracking algorithm in the complex background,an automatic selection and tracking method of the motion target is proposed based on the saliency hue level.The main research contents of this thesis areas follows:In order to improve the adaptive ability of the quantum median filtering method to the illumination variations,a novel median filtering method based on the dual qubit state is proposed.The distribution probability of the pixels in the image is determined according to the gray scale distribution information.And the normalized distribution probability is determined as the probability function of the quantum bit states.In this way,the filtering method has good adaptive ability to the illumination variations,and the filtering performance can be improved.The method is used to filter noise in the images under the normal,low and high illumination.Simulation result show that the method can get better filtering result to the salt and pepper noise than the other median filtering methods,and the method is also superior to others on some objective evaluation criteria,such as the peak signal to noise ratio,the normalized mean square error,and the structural similarity.In order to improve the tracking performance of the continuously adaptive Mean Shift tracking algorithm in the complex background,an automatic selection and tracking method of the motion target is proposed based on the saliency hue level.Gaussian mixture model is used to determine the target template,and the saliency hue level can be determined by comparing the hue histograms of the target template and the background.The area composed of pixels with the saliency hue level is selected as the tracking target.The probability distribution image of the tracking image can be got according to the hue histogram of the tracking target by the back projection.The continuously adaptive Meanshift tracking algorithm is used to perform the target tracking.Simulation results show that the method can extract the saliency hue level of target,and it is easy to distinguish the target from the background and restrain the information from the background to disturb the target tracking.Thus,the tracking performance can be improved under the complex background.The average tracking time of single frame is less than 15ms,so the method can meet the real-time requirement of the tracking system.Experimental results show that the method proposed in this paper can effectively remove the salt and pepper noise in images,and can realize the stable tracking of motion target in complex environments.