Optical flow method is an important and effective deformation registration algorithm based on optical flow field model. Aiming at the problem that the feature quality used by the existing optical flow method is not high enough to make the registration result accurate, combining the features of deep convolutional neural network and optical flow method, a deformable medical image registration algorithm based on Deep Convolution Feature Based Optical Flow (DCFOF) was proposed. Firstly, the deep convolution feature of the image block where each pixel in the image was located was densely extracted by using a deep convolutional neural network, and then the optical flow field was solved based on the deep convolution feature difference between the fixed image and the floating image. By extracting more accurate and robust deep learning features of the image, the optical flow field obtained was closer to the real deformation field, and the registration accuracy was improved. Experimental results show that the proposed algorithm can solve the problem of deformable medical image registration effectively, and has the registration accuracy better than those of Demons algorithm, Scale-Invariant Feature Transform（SIFT） Flow algorithm and professional registration software of medical images called Elastix.

In the process of indoor positioning and navigation, a Bayesian network-based floor localization algorithm was proposed for the problem of large error of floor localization when only the pedestrian height displacement considered. Firstly, Extended Kalman Filter (EKF) was adopted to calculate the vertical displacement of the pedestrian by fusing inertial sensor data and barometer data. Then, the acceleration integral features after error compensation was used to detect the corner when the pedestrian went upstairs or downstairs. Finally, Bayesian network was introduced to locate the pedestrian on the most likely floor based on the fusion of walking height and corner information. Experimental results show that, compared with the floor localization algorithm based on height displacement, the proposed algorithm has improved the accuracy of floor localization by 6.81%; and compared with the detection algorithm based on platform, the proposed algorithm has improved the accuracy of floor localization by 14.51%. In addition, the proposed algorithm achieves the accuracy of floor localization by 99.36% in the total 1247 times floor changing experiments.

In order to improve the reliability of a hub and spoke network and maintain the normal operation of the hub and spoke network during the failure of the initial hub, a new hub backup optimization method for the hub and spoke network was proposed, in which a backup hub was selected for each hub point to make the initial cost and the backup cost of the hub and spoke network the best. Firstly, the hub backup variables were introduced into the basic model of a hub and spoke network, and an extension model of nonlinear programming was established. The extended model was linearized by the linearization method of variable substitution, and mathematical solver CPLEX was used to solve the small scale problem of the hub and spoke network hub backup. Then, the scale of hub and spoke network nodes was increased, and a genetic algorithm was designed to solve the problem of large scale hub backup optimization in the hub and spoke network. Finally, in the CPLEX and genetic algorithm, the proportion weights of the initial hub and spoke network cost and backup cost were adjusted, the exact solutions and optimal solutions of initial cost, backup cost, hub location and backup hub were obtained respectively. The optimal values of the initial hub and spoke network, backup hub as well as the objective function were obtained by the example experiments. The experimental results show that, the backup hub of the the proposed method shares the traffic and capacity of the initial hub, and when the initial hub fails, the backup hub can undertake the transportation task of the initial hub and keep the hub and spoke network running. The proposed optimization method of hub backup can be applied to the emergency logistics and security management of logistics network.

For the scheduling problem of no cross-over twin Rail-Mounted Gantry (RMG) and container slot selection, considering the safety distance between the two RMGs and the buffer capacity, a coupled model of twin RMG scheduling and container slot selection was proposed with the goal of minimizing the completion time by setting the twin RMG scheduling as the main line and setting the container slot selection as the auxiliary line. The basic idea of it is to set the decision variable to describe the relationship between the tasks. A Genetic Algorithm-Ant Algorithm (GAAA) was designed for solving the coupled model, and the CPLEX was developed for comparisons by analyzing the efficiency in relay mode and mixed mode. The experimental results show that the efficiency in relay mode is better than that of mixed mode when dealing with 8 to 150 container tasks; in small and medium-large sized experiments, the minimum completion time of GAAA is reduced by about 2.65% and 18.50%, respectively; the running time of GAAA is reduced by 88.6% and 99.19% respectively on average compared with CPLEX, which validates the validity of the model.

In order to solve the problem of user's fairness in D2D (Device-to-Device) communication system, firstly, the existing proportional fairness principle was extended to derive an optimization problem relating to weighted sum-rate, and then a KMPF (Kuhn-Munkras Proportional Fair) resource allocation algorithm was proposed to optimize it. The algorithm maximized the user's weighted sum-rate through power control, and allocated the cellular user's resources that could be reused for the D2D users according to maximization of the total weighted sum-rate by Kuhn-Munkras (KM) algorithm. Simulation results show that the fairness index of the proposed algorithm is 0.4 higher than that of the greedy resource allocation algorithm and the throughput of the system is over 95% of its level, and the throughput of proposed algorithm is about 50% higher than that of the random resource allocation algorithms. It is shown that the algorithm can solve the problem of user's fairness while considering the system throughput.

In order to adapt to the the requirement of the Instruction Set Simulator (ISS) simulation speed in embedded system development, an improved ISS technology was put forward.The technology introduced instruction preprocessing, dynamic decode cache structure, multi-thread C function generation and dynamic scheduling technique based on the existing static multi-core simulator to achieve the optimization of the simulator performance. This technique has been applied successfully in forming OPT-ISS, which is based on IME-Diamond multi-core DSP processor. The experimental results show that this technique improves the simulation speed indeed.

Due to the swallow and over-enhancement problems of traditional histogram equalization, an improved histogram equalization algorithm combining scene classification and details preservation was proposed. In this algorithm, images were classified according to their histogram features. The parameter of piecewise histogram equalization was optimized according to the scene classification and the characteristics of image histogram. The complexity of the improved algorithm is only O(L).L is the level of image grayscale, and equals to 256 here. The improved algorithm has the small amount of computation and solves the swallow and over-enhancement problems of traditional histogram equalization. The results from TI (Texas Instruments) DM648 platform show the algorithm can be used for real-time video image enhancement.

To reduce the computation of Gauss mixture model effectively and improve the accuracy of shadow elimination in moving object detection, an algorithm which updated the model selectively and eliminated the shadow by the change of brightness was proposed. Firstly, the weight of the Gauss distribution and the rate of those that did not belong to the background were compared before updating the Gauss distribution, if the former was larger, then did not update it, otherwise, updated it; Secondly, the range of brightness change was chosen to be a threshold factor of shadow detection, so that the threshold could be adjusted adaptively according to the change of brightness. Finally, compared this algorithm with the traditional ones through experiments on indoor and outdoor videos, the experimental results show that the time consumption of the algorithm is about one-third of the traditional ones, the accuracy of shadow eliminating is improved and the efficiency of the algorithm is confirmed.

To avoid the problem of the carton missing in the process of cigarette production, this paper introduced a new method of pattern matching based on machine. Using the method could avoid the effects of the random reflecting light on images. After getting the taboo area of the image, the result of the pattern matching was used to determine whether some cartons miss or not. In addition, the taboo matching method could also adjust the image and get the template image automatically without considering the pattern or color of the carton. The taboo matching would reduce the error detection rate in a real system and provide a way of solving problems of the similar kind.

Detecting records that are approximate duplicates, but not exact duplicates, is one of the key tasks in information integration. Although various algorithms have been presented for detecting duplicated records, strings matching is essential to those algorithms. In self- adaptive information integration algorithm presented by this paper, the hybrid similarity, a comprehensive edit distance and token metric, was used to measure the similar degree between strings. In order to avoid mismatching because of different expressions, the strings in records were partitioned into vocabularies, then were sorted according to their first character. In the process of vocabularies matching, misspellings and abbreviations can be tolerated. The experimental results demonstrate that the self-adaptive approach for information integration achieves higher accuracy than that using Smith-Waterman edit distance and Jaro distance.