In the two components of hybrid image segmentation algorithm, the initial segmentation cannot form the over-segmentation region sets with low wrong segmentation rate, while region merging lacks the label selection mechanism for region merging and the method of determining region merging stopping moment in this component commonly does not meet the scenario requirements. To solve the above problems, a Multi-level Region Information fusion based Hybrid image Segmentation algorithm (MRIHS) was proposed. Firstly, the improved Markov model was used to smooth the superpixel blocks, so as to form initial segmentation regions. Then, the designed region label selection mechanism was used to select the labels of the merged regions after measuring the similarity of the initial segmentation regions and selecting the region pairs to be merged. Finally, an optimal merging state was defined to determine region merging stopping moment. To verify MRIHS performance, comparison experiments between this algorithm with Multi-dimensional Feature fusion based Hybrid image Segmentation algorithm (MFHS), Improved FCM image segmentation algorithm based on Region Merging (IFRM), Inter-segment and Boundary Homogeneities based Hybrid image Segmentation algorithm (IBHHS), Multi-dimensional Color transform and Consensus based Hybrid image Segmentation algorithm (MCCHS) were carried out on Visual Object Classes (VOC), Cambridge-driving labeled Video database (CamVid) and the self-built river and lake inspection (rli) datasets. The results show that on VOC and rli datasets, the Boundary Recall (BR), Achievable Segmentation Accuracy (ASA), recall and dice of MRIHS are at least increased by 0.43 percentage points, 0.35 percentage points, 0.41 percentage points, 0.84 percentage points respectively and the Under-segmentation Error (UE) of MRIHS is at least decreased by 0.65 percentage points compared with those of other algorithms; on CamVid dataset, the recall and dice of MRIHS are at least improved by 1.11 percentage points, 2.48 percentage points respectively compared with those of other algorithms.

An Ultra WideBand (UWB)/ Pedestrian Dead Reckoning (PDR) fusion positioning algorithm with adaptive coefficient adjustment based on UWB error prediction was proposed in order to improve the UWB performance and reduce the PDR accumulative errors in the indoor Non-Line-Of-Sight (NLOS) positioning scenes and solve the UWB performance degradation caused by environmental factors. On the basis of the creative proposal of predicting the UWB positioning errors in complex environment by Support Vector Machine (SVM) regression model, UWB/PDR fusion positioning performance was improved by adding adaptive adjusted parameters to the conventional Extended Kalman Filter (EKF) algorithm. The experimental results show that the proposed algorithm can effectively predict the current UWB positioning errors in the complex UWB environment, and increase the accuracy by adaptively adjusting the fusion parameters, which makes the positioning error reduced by 18.2% in general areas and reduced by 48.7% in the areas with poor UWB accuracy compared with those of the conventional EKF algorithm, so as to decrease the environmental impact on the UWB performance. In complex scenes of both Line-Of-Sight (LOS) and NLOS including UWB, the positioning error per 100 meters is reduced from meter scale to decimeter scale, which reduces the PDR errors in NLOS scenes.

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.

Aiming at the problem that the existing 3D model information hiding algorithms cannot effectively resist uneven compression, a multi-carrier low-density information hiding algorithm based on multiple fusion states was proposed. Firstly, multiple 3D models were positioned, oriented and stereotyped by translation and scaling. Secondly, the 3D models were rotated at different angles and merged by using the center point as merging point to obtain multiple fusion states. Thirdly, local height and Mean Shift clustering analysis were used to divide the energy of the vertices of the fusion state model, obtaining the vertices with different energies. Finally, by changing the vertex coordinates, the secret information changed by Arnold scrambling was quickly hidden in multiple fusion states and 3D models. Experimental results show that the proposed algorithm is robust against uneven compression attacks and has high invisibility.

The existing hyperspectral image compression algorithms that aim at high compression quality generally have problems such as high computational complexity, off-line processing, and difficulty in implementing an embedded platform. They are difficult to be implemented in practical applications at present. To resolve the above problems, a real-time compression method for embedded hyperspectral images based on Karhunen-Loeve Transform (KLT) and HEVC (High Efficiency Video Coding) was designed. Firstly, the inter-spectral correlation was reduced by KLT. Then, the spatial correlation was removed by HEVC. Finally, the process of quantization and entropy coding was accomplished by HEVC. Based on NVIDIA Jetson TX1 platform, a heterogeneous parallel compression system which utilizes both the CPU and GPU was designed and implemented. Using real data sets, the performance of the designed algorithm and the practicability of the implemented platform were verified. The experimental results show that compared with the Discrete Wavelet Transform (DWT)+JPEG2000 algorithm, the reconstruction accuracy is improved significantly under the same compression ratio. The Peak Signal-to-Noise Ratio (PSNR) is increased by 1.36 dB on average; at the same time, compared with CPU, performing KLT calculations on GPU can also reduce the runtime by 33% at most.

Aiming at the problem that some 3D model-based information hiding algorithms are incompetent against combined attacks, a new strategy based on feature point labeling and clustering was proposed. Firstly, edge folding was adopted to achieve mesh simplification and all the vertexes were labeled in order by their energy level. Secondly, the ordered vertexes were clustered and re-ordered by using local height theory and Mean Shift clustering analysis. Lastly, hidden information and cover model carrier information were optimized, matched and modified by Logistic chaos mapping scrambling and genetic algorithm, completing the final hiding. The data in hiding areas were labeled and screened locally and globally according to the energy weight, which is good for the robustness and transparency of the algorithm. The experimental results show that, compared with 3D information hiding algorithms based on inscribed sphere and outer skeleton, the robustness of the proposed algorithm against single or joint attacks is significantly improved, and it also has the same degree of invisibility.

Aiming at these problems that the traffic sign has a small proportion in the natural scene, the extracted features are insufficient and the recognition accuracy is low, an improved Scale Dependent Pooling (SDP) model was proposed for the recognition of small-scale traffic images. Firstly, because the deep convolution layer of neural network has better contour information and class characteristics, Supplementary Deep convolution layer characteristic Scale-Dependent Pooling (SD-SDP) model for deep convolution layer characteristic was used to extract features based on the feature information of shallow convolution by SDP model, enriching feature information. Secondly, the Multi-scale Sliding window Pooling (MSP) was used to make up the edge information of the target object, instead of the single-layer spatial pyramid method in the original SDP algorithm. Finally, the improved SDP model was applied to the recognition of traffic signs. The experimental result show that, compared to SDP algorithms, the extracted feature dimension increases and the accuracy of small scale traffic image recognition is improved.

In the traditional indoor pedestrian positioning algorithm based on dead reckoning and Kalman filtering, there is a problem of cumulative error in the heading angle, which makes the positional error continue to accumulate continuously. To solve this problem, a pedestrian heading particle filter algorithm with indoor environment constraints was proposed to correct direction error. Firstly, the indoor map information was abstracted into a structure represented by line segments, and the map data was dynamically integrated into the mechanism of particle compensation and weight allocation. Then, the heading self-correction mechanism was constructed through the correlation map data and the sample to be calibrated. Finally, the distance weighting mechanism was constructed through correlation map data and particle placement. In addition, the particle filter model was simplified, and heading was used as the only state variable to optimize. And while improving the positioning accuracy, the dimension of state vector was reduced, thereby the complexity of data analysis and processing was reduced. Through the integration of indoor environmental information, the proposed algorithm can effectively suppress the continuous accumulation of directional errors. The experimental results show that, compared with the traditional Kalman filter algorithm, the proposed algorithm can significantly improve the pedestrian positioning accuracy and stability. In the two-dimensional walking experiment with a distance of 435 m, the heading angle error is reduced from 15.3° to 0.9°, and the absolute error at the end position is reduced from 5.50 m to 0.87 m.

When directly applying the existing fog algorithms to the traffic image, the transitional region is obvious and the color cast is serious, which can not meet the requirement of subsequent traffic sign detection. In order to solve this problem, a modified single traffic image dehazing algorithm based on dark channel prior image was proposed. Firstly, the modified mean shift algorithm was used to segment the sky region of traffic image; then the histogram equalization algorithm was used to defog the partitioned sky region, and the dark channel prior algorithm based on efficient bilateral filter was used to defog the non-sky region. At last, the final image dehazing was finished by image fusion. Experimental result shows that compared with the typical image dehazing algorithms, the proposed algorithm has better effect in transitional region of the sky, the color cast is not serious, and its processing speed is faster; the quantitative analysis result indicates that the proposed algorithm has better effect in dehazing, and can meet the requirement of subsequent traffic sign dectection system.

Focusing on the issue that the Road Side Unit (RSU) has inefficient response to the request of the vehicles in Vehicular Ad Hoc Network (VANET), a data scheduling algorithm based on Software Defined Network (SDN) architecture, namely SDDS, was proposed. Firstly, a graph of conflicting policies was generated based on status information of vehicles, and a maximum weighted independent set of the graph was solved to maximize the number of satisfied requests in current cycle. Secondly, the redundancy of data in vehicles was analyzed to figure out the optimum parameter, and a selection mechanism for collaborative vehicles was designed based on geographical position. Finally, the characteristics of handover vehicles and some factors that would affect the multi-RSU cooperation were analyzed, and a multi-RSU cooperation mechanism was put forward based on collision avoidance. In addition, a new evaluation indicator, service efficiency, was proposed to estimate the overall quality of service. Simulation results showed that compared with Most Requests First (MRF) and Cooperative Data Dissemination (CDD) algorithms, the service efficiency of SDDS algorithm was increased up to 15% and 20% respectively. The simulation results prove that SDDS algorithm can observably improve the sevice eficiency and quality of scheduling system.

The limited number and high similarity of learning clauses lead to limited historical information and imbalanced search tree. In order to solve the problems, a dynamic branching strategy of award and punishment was proposed. Firstly, the variables of every unit propagation were punished. Different penalty functions were established according to whether the variable generated a conflict and the conflict interval. Then, in the learning phase, the positive variables for the conflict were found according to the learning clauses, and their activities were nonlinearly increased. Finally, the variable with the maximum activity was chosen as the new branching variable. On the basis of the glucose3.0 algorithm, an improved dynamic algorithm of award and punishment named Award and Punishment 7 (AP7) was completed. The experimental results show that, compared with the glucose3.0 algorithm, the rate of cutting branches of AP7 algorithm is improved by 14.2%-29.3%, and that of a few examples is improved up to 51%. The running time of the improved AP7 algorithm is shortened more than 7% compared with the glucose3.0 algorithm. The branching strategy can efficiently reduce the size of search tree, make the search tree more balanced and reduce the computation time.

Data transmission based on opportunistic communication in mobile crowdsensing may take a long period of time. To address this issue, a new Hub-based multi-Task Assignment (HTA) algorithm was proposed. In this algorithm, some nodes were selected to perform as the hubs which could help the requester node to deliver the tasks, according to the different characteristics of the social relationship of the nodes in mobile networks. When the task requester encountered a hub node, the hub node itself and its slave nodes were assigned tasks. After that, the hub node would distribute the tasks to the salve nodes, and received the results from them. Simulations were conducted on The ONE simulator. Compared with the oNline Task Assignment (NTA) algorithm, HTA algorithm reduced the time cost by 24.9% on average and improved the task completion ratio by 150% on average. The experimental results demonstrate that HTA algorithm can accelerate the accomplishment speed of the task and reduce the time cost.

In order to reduce the cost and simplify the installation of Variable Refrigerant Volume (VRV) air-conditioning system, a communication system was constructed based on Power Line Carrier Communication (PLCC) technology. First, a solution based on narrowband PLCC technology was proposed according to the communication requirement of the system control. Then, the four-layer communication protocol including physical layer, Medium Access Control (MAC) layer, network layer and application layer was designed mainly from two aspects of channel access control and networking control. Three kinds of channel access algorithms based on CSMA/CA (Carrier Sense Multiple Access with Collsion Avoidance) design idea and a kind of network algorithm for star network topology were put forward. Last, the networking test, system communication test and anti-disturbance test for a 11-node VRV system were given. The test results show that the proposed algorithms and solution can satisfy the real-time control of VRV system and has strong ability of anti-disturbance. In addition, with the openness of the designed communication protocol, it can be modified according to different requirements and be applied to a variety of real-time control areas in short distance without relay such as smart home.

When the packets are broadcasted with the fixed power in Vehicular Ad-Hoc NETwork (VANET), the wireless channel may not be allocated reasonable. In order to solve this problem, a power control mechanism adapted to the variation of vehicle density was proposed. It is adaptive to the variation of vehicle density. The direct neighbor list of each node was constructed and updated in a power control period, the power that used to transmit the vehicle status message was adjusted according to the location of the direct neighbor to cover all the direct neighbors, thus wireless channel could be allocated more reasonable and the performance of router could also be optimized. The validity of the proposed mechanism was proved by the simulation results. It is also found that the proposed mechanism is useful for adjusting the transmission power according to the vehicular density, reducing channel busy ratio and enhancing the performance of packet delivery ratio among direct neighbors, which can ensure the effective transmission of the security information.

In order to fully use the structural information of the data, and compress the image data, the sparse constraints of the subspace (feedback) were applied to the object function of non-negative tensor factorization. Then this algorithm was used to reduce the dimension of the image sets. Finally, image classification was realized. The experimental results on the handwritten digital image database show that the proposed algorithm can effectively improve the accuracy of the image classification.