Aiming at the problem that "with the increase of the prediction distance, the prediction of travel time becomes more and more difficult", a comprehensive prediction model of Long Short Term Memory (LSTM) and Artificial Neural Network (ANN) based on spatiotemporal eigenvectors was proposed. Firstly, 24 hours were segmented into 288 time slices to generate time eigenvectors. Secondly, the LSTM time window model was established based on the time slices. This model was able to solve the window movement problem of long-time prediction. Thirdly, the bus line was divided into multiple space slices and the average velocity of the current space slice was used as the instantaneous velocity. At the same time, the predicted time of each space slice would be used as the spatial eigenvector and sent to the new hybrid neural network model named LSTM-A (Long Short Term Memory Artificial neural network). This model combined with the advantages of the two prediction models and solved the problem of bus travel time prediction. Finally, based on the experimental dataset, experiments and tests were carried out:the prediction problem between bus stations was divided into sub-problems of line slice prediction, and the concept of real-time calculation was introduced to each related sub-problem, so as to avoid the prediction error caused by complex road conditions. Experimental results show that the proposed algorithm is superior to single neural network models in both accuracy and applicability. In conclusion, the proposed new hybrid neural network model LSTM-A can realize the long-distance arrival time prediction from the dimension of time feature and the short-distance arrival time prediction from the dimension of spatial feature, thus effectively solving the problem of urban bus travel time prediction and avoiding the remote dependency and error accumulation of buses.

To solve the problem of data sparsity and cold start which is prevalent in recommender system, a new social recommendation algorithm was proposed, which integrates rating and trust relation. Firstly, the initial trust value of the new user in the network was reasonably assigned, which solves the problem of cold start of the new user. Since the user's preferences were affected by his friends, the user's own feature vector was modified by the trust matrix between friends, which solves the problem of user's feature vector construction and trust transition. The experimental results show that the proposed algorithm has a significant performance improvement over the traditional social network recommendation algorithm.

Focusing on the issue that the Signal-to-Clutter-and-Noise Ratio (SCNR) of echo signal is low when cognitive radar detects extended target, a waveform design method based on SCNR was proposed. Firstly, the relation between the SCNR of cognitive radar echo signal and the Energy Spectral Density (ESD) of transmitted signal, was gotten by establishing extended target detection model other than previous point target model; secondly, according to the maximum SCNR criterion, the global optimal solution of the transmitted signal ESD was deduced; finally, in order to get a meaningful time-domain signal, ESD was synthesized to be a constant amplitude based on phase-modulation after combining with the Minimum Mean-Square Error (MMSE) and iterative algorithm, which met the emission requirements of radar. In the simulation, the amplitude of time-domain synthesized signal is uniform, and its SCNR at the output of the matched filter is 19.133 dB, only 0.005 dB less than the ideal value. The results show that not only can the time-domain waveform meet the requirement of constant amplitude, but also the SCNR obtained at receiver output can achieve the best approximation to the ideal value, and it improves the performance of the extended target detection.

In the existing reversible watermarking algorithm based on mean-adjustable integer transform, there are following defects such as non-adaptive threshold selecting, incomplete location map building strategy which may lead to poor compression performance and compulsive partition strategy for embedded vectors which may lead to a failure embedding even if embedding capacity is enough. To address these problems, an iterative adaptive reversible image watermarking algorithm combined with mean-adjustable integer transform was proposed. Firstly, according to Peak Signal-to-Noise Ratio (PSNR) affected by the payload data size and integer vector, an iterative adaptive algorithm was used in selecting mean-adjustable offsets to balance the watermarking embedding capacity and the visual quality of embedded carrier; Secondly, based on the strategy that adjacent pixels have similar pixel values, a complete location map generating strategy was proposed to improve location map compression performance; Finally, to avoid failure embedding, the proposed reversible watermarking algorithm adopted hierarchical order embedding strategy to embed payload data in order from the first least significant bits to the third least significant bits. The experimental results show that the proposed algorithm has a big embedding capacity and does not need to preset threshold. Location map building strategy has a better performance in making location map data in smaller size and increasing the capacity indirectly compared with the reversible watermarking algorithm based on mean-adjustable integer transform, and the PSNR increases by 14.4% averagely in experimental sample.

In conventional permutation and confusion based image encryption algorithm, there usually exists some problems such as inefficient permutation and difficult to resist known or chosen plaintext attack. To solve these problems, an image encryption algorithm based on maze permutation and Logistic mapping was proposed, where Depth First Search (DFS) maze permutation was used to product permutation efficiently. In order to resist known or chosen plaintext attack, the plaintext image Message Digest Algorithm 5 (MD5) digest was bound with the user key to generate maze starting coordinates, Logistic chaotic map parameters and initial values which drive Logistic maps to generate random numbers. These random numbers were used to determine maze node probing directions and participate in image confusion to make all encryption stages tight coupled with the plaintext image. Experiments show the proposed algorithm has better performance in encryption quality and it can resist known or chosen plaintext attack with high security.

Concerning data sparsity and malicious behavior of traditional collaborative filtering algorithm, a new social recommendation method combining trust and matrix factorization was proposed in this paper. Firstly, the incredible nodes in the network were founded by computing their prestige value and bias value, and then the weight of their evaluation would be weakened. Finally, the collaborative recommendation was conducted under the social network circumstance by combining the user-item matrix and trust matrix. The experimental results show that the proposed algorithm reduces the importance of not credible node to weaken the negative influence the false or malicious score brings to recommendation system, the data sparsity and malicious behavior problems can be alleviated, and a higher prediction accuracy than that of the traditional collaborative filtering algorithms can be achieved.

Traditional outlier detection algorithms fall short in efficiency for their holistic nearest neighboring search mechanism and need to be improved. This paper proposed a new outlier detection method using attribute reduction techniques which enabled the algorithm to focus its detecting scope only on the most meaningful attributes of the data space. Under the reduced set of attributes, a concept of neighborhood-based outlier factor was defined for the algorithm to judge data's abnormity. The combined strategy can reduce the searching complexity significantly and find more reasonable outliers in dataset. The results of experiments also demonstrate promising adaptability and effectiveness of the proposed approach.

A keyframe merging-based video clip retrieval method was presented. Videos were segmented to sub-shots by features joint distribution histogram and the sub-shots were represented by keyframes.Finding the similar keyframes to any keyframes of an example video clip when retrieving,and all the similar keyframes were merged into clips.An efficient similarity model of video clips was put forward. The experimental results show that the proposed method is efficient.