Aiming at the characteristics of rail surface images such as uneven illumination, limited discernible features, low contrast and changeable reflection characteristics, a background differential rail surface defect detection method based on defect proportion limitation was proposed. The method mainly includes five steps:pre-processing of rail surface images, background modeling and difference, defect proportion limitation filtering, maximum entropy threshold segmentation of defect proportion limitation and connected area labeling. Firstly, the column grayscale mean and median of the rail surface image were combined to perform the rapid background modeling, and the difference operation was carried out to the pre-processed image and the background image. Secondly, the feature with low defect proportion in the rail surface image was used to truncate the upper threshold limit of the defect proportion in order to enhance the contrast of the difference image. Thirdly, the maximum entropy threshold segmentation was improved by using this feature, the global variable weighting of the target entropy was carried out by using the adaptive weighting factor, and an appropriate threshold was selected to maximize the entropy value, so as to reduce the interference of noises such as shadow and rust while retaining the real defects. Finally, the connected area labeling method was used to perform the statistics of the defect areas in the segmented binary image, and the area with defect area lower than the rail damage standard was determined as the noise and removed, so as to realize the rail surface defect detection. Simulation results show that the new method can detect rail surface defects well, and its results have the recall rate, precision rate and weighted harmonic mean of 94.19%, 88.34% and 92.96% respectively, and the average mis-classification error of 0.006 4, so that the method has certain practical value.

A new occluded object tracking algorithm based on mirror image and Mean Shift was proposed to solve the problem that the track object is not accurate, even lost during full occlusion in this paper. The algorithm included three steps: Firstly, when the object was uncovered (Bhattacharyya coefficient matching degree of adjacent frames was greater than or equal to 80%), color features and contour features were used to locate the target, and size adaptive adjustment was realized by sandbag kernel window based on partition. Secondly, when the object is occluded (Bhattacharyya coefficient matching degree of adjacent frames was less than 80%), the location and the size of the target was predicted by using prior training classifier and mirror principle.Thirdly, When target left the occlusion area (Bhattacharyya coefficient matching degree of adjacent frames was greater than or equal to 80% again), Mean Shift algorithm was used to track the target. The experimental results show that when the object is fully occluded, the proposed algorithm is more accurate and robust to better solve the occlusion problem than sub-regional on-line Boosting algorithm and multi-view object tracking algorithm combining modified fusion feature with dynamic occlusion threshold, and meets the real-timer requirements.

Aiming at the problem of the high computation and application difficulty in traditional no-reference image assessment methods, a simple and direct no-reference Gaussian image quality assessment algorithm based on wavelet high frequency Structural SIMilarity (SSIM) was proposed. The proposed algorithm took into account the similarity among the natural images high frequency in the same scale which would be reduced with the distortion deepening. Three different directional sub-bands of high frequencies were obtained by the wavelet transform firstly，and then the Peak Signal-to-Noise Ratio (PSNR) and SSIM were combined for calculating the differences of sub-bands respectively as the last objective assessment index. The simulation results show that the proposed method has good consistence with the subjective assessment on three common image databases，in addition，the algorithm only needs about 0.2s for evaluating an image, which has good practicality.

The traditional Binary Particle Swarm Optimization (BPSO) algorithm does not make full use of the historical position information for its iterative optimization, which impedes further improvement on the efficiency of the algorithm. To deal with the problem, an improved BPSO algorithm with the experience factor was proposed. The new algorithm exploited the experience factor, which could reflect the historical information of particle's position, to influence the speed update of particles and therefore improved the optimization process. In order to avoid the excessive dependence on the historical experience information of particles, the algorithm regulated the historical information through the reward and punishment mechanism and a history-forgotten coefficient, and in the end, empirical weights were used to determine the final effect on the experience factor. Compared with the classic BPSO and related improved algorithm, the experimental results show that the new algorithm can achieve better effects both in convergence speed and global search ability.