To solve the problems of requiring a large number of samples, not associating with time information, and not using global navigation information in motion planning for autonomous driving based on machine learning, a motion planning method for autonomous driving with directional navigation based on deep spatio-temporal Q-network was proposed. Firstly, in order to extract the spatial features in images and the temporal information between continuous frames for autonomous driving, a new deep spatio-temporal Q-network was proposed based on the original deep Q-network and combined with the long short-term memory network. Then, to make full use of the global navigation information of autonomous driving, directional navigation was realized by adding the guide signal into the images for extracting environment information. Finally, based on the proposed deep spatio-temporal Q-network, a learning strategy oriented to autonomous driving motion planning model was designed to achieve the end-to-end motion planning, where the data of steering wheel angle, accelerator and brake were predicted from the input sequential images. The experimental results of training and testing results in the driving simulator named Carla show that in the four test roads, the average deviation of this algorithm is less than 0.7 m, and the stability performance of this algorithm is better than that of four comparison algorithms. It is proved that the proposed method has better learning performance, stability performance and real-time performance to realize the motion planning for autonomous driving with global navigation route.

In order to improve the detection level of bridge cracks,and solve the time-consuming and laborious problem in manual detection and the parameters to be set manually in traditional image processing methods,an improved bridge crack detection algorithm was proposed based on GoogLeNet. Firstly,a large-scale bridge crack Retinex-Laplace-Histogram equalization(RLH)dataset was constructed for model training and testing. Secondly,based on the original GoogLeNet model,the inception module was improved by using the normalized convolution kernel,three improved schemes were used to modify the beginning of the network,the seventh and later inception layers were removed,and a bridge crack feature image classification system was established. Finally,the sliding window was used to accurately locate the cracks and the lengths and widths of the cracks were calculated by the skeleton extraction algorithm. The experimental results show that compared with the original GoogLeNet network,the improve-GoogLeNet network increased the recognition accuracy by 3. 13%, and decreased the training time to the 64. 6% of the original one. In addition,the skeleton extraction algorithm can consider the trend of the crack,calculate the width more accurately,and the maximum width and the average width can be calculated. In summary,the classification and measurement method proposed in this paper have the characteristics of high accuracy,fast speed,accurate positioning and accurate measurement.

To solve the problems such as data sharing difficulty, data privacy disclosure of data sharing between medical institutions, a blockchain-based Electronic Health Record (EHR) sharing scheme was proposed. Firstly, based on the blockchain characteristics of non-tampering, decentralization and distributed storage, a blockchain-based EHR data sharing model was designed. The blockchain network and distributed database were used to jointly store the encrypted EHR and the related access control policies, preventing the modification and leakage of EHR data. Secondly, the Distributed Key Generation (DKG) and Identity-Based Proxy Re-Encryption (IBPRE) were combined to design a data secure sharing protocol. The Delegated Proof of Stake (DPOS) algorithm was used in this protocol to select the proxy node, which re-encrypted the EHR to achieve the data sharing between single pair of users. The safety analyses show that the proposed scheme can resist the fake identity and the replay attack. Simulation experiments and comparative analyses show that DPOS algorithm has the efficiency higher than Proof of Work (POW) algorithm, and slightly lower than the Practical Byzantine Fault Tolerance (PBFT) algorithm, but the proposed scheme is more decentralized and costs less computing power.

The existing Improved Multi-label Lazy Learning Approach (IMLLA) has the problem that the influence of similarity information is ignored with only the neighbor label correlation information considered when the neighbor labels were used, which may reduce the robustness of the approach. To solve this problem, with firefly method introduced and the combination of similarity information with label information, a Multi-label Lazy Learning Approach based on FireFly method (FF-MLLA) was proposed. Firstly, Minkowski distance was used to measure the similarity between samples to find the neighbor point. Secondly, the label count vector was improved by combining the neighbor point and firefly method. Finally, Singular Value Decomposition (SVD) and kernel Extreme Learning Machine (ELM) were used to realize linear classification. The robustness of the approach was improved due to considering both label information and similarity information. The experimental results demonstrate that the proposed approach improves the classification performance to a great extent compared to other multi-label learning approaches. And the statistical hypothesis testing and stability analysis are used to further illustrate the rationality and effectiveness of the proposed approach.

Existing weakly illuminated image enhancement algorithms are strongly dependent on Retinex model and require manual adjustment of parameters. To solve those problems, an algorithm based on Convolutional Neural Network (CNN) was proposed to enhance weakly illuminated image. Firstly, four image enhancement techniques were used to process weakly illuminated image to obtain four derivative images, including contrast limited adaptive histogram equalization derivative image, Gamma correction derivative image, logarithmic correction derivative image and bright channel enhancement derivative image. Then, the weakly illuminated image and its four derivative images were input into CNN. Finally, the enhanced image was output after activation by CNN. The proposed algorithm can directly map the weakly illuminated image to the normal illuminated image in end-to-end way without estimating the illumination map or reflection map according to Retinex model nor adjusting any parameters. The proposed algorithm was compared with Naturalness Preserved Enhancement Algorithm for non-uniform illumination images (NPEA), Low-light image enhancement via Illumination Map Estimation (LIME), LightenNet (LNET), etc. In the experiment on synthetic weakly illuminated images, the average Mean Square Error (MSE), Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity index (SSIM) metrics of the proposed algorithm are superior to comparison algorithms. In the real weakly illuminated images experiment, the average Natural Image Quality Evaluator (NIQE) and entropy metric of the proposed algorithm are the best of all comparison algorithms, and the average contrast gain metric ranks the second among all algorithms. Experimental results show that compared with comparison algorithms, the proposed algorithm has better robustness, and the details of the images enhanced by the proposed algorithm are richer, the contrast is higher, and the visual effect and image quality are better.

Traditional feature selection algorithm cannot process streaming feature data, the redundancy calculation is complicated and the description of the instance is not accurate enough. A multi-label Distribution learning Feature Selection with Streaming Data Using Rough Set (FSSRS) was proposed to solve the above problem. Firstly, the online streaming feature selection framework was introduced into multi-label learning. Secondly, the original conditional probability was replaced by the dependency in rough set theory, which made the streaming data feature selection algorithm more efficient and faster than before by only using the information calculation of the data itself. Finally, since each label has a different degree of description for the same instance in real world, to make the description of the instance more accurate, label distribution was used to instead of traditional logical labels. The experimental results show that the proposed algorithm can retain the features with high correlation with the label space, so that the classification accuracy is improved to a certain extent compared with that without feature selection.

Most of traditional text classification algorithms are based on vector space model and hierarchical classification tree model is used for statistical analysis. The model mostly doesn't combine with the semantic information of characteristic items. Therefore it may produce a large number of frequent semantic modes and increase the paths of classification. Combining with the good distinguishment characteristic of essential Emerging Pattern (eEP) in the classification and the model of rough set based on minimum expected risk decision, a Text Semantic Classification algorithm with Threshold Optimization (TSCTO) was presented. Firstly, after obtaining the document feature frequency distribution table, the minimum threshold value was calculated by the rough set combined with distribution density matrix. Then the high frequency words of the semantic intra-class document frequency are obtained by combining semantic analysis and inverse document frequency method. In order to get the simplest model, the eEP pattern was used for classification. Finally, using similarity formula and HowNet semantic relevance degree, the score of text similarity was calculated, and some thresholds were optimized by the three-way decision theory. The experimental results show that the TSCTO algorithm has a certain improvement in the performance of text classification.

Concerning the disadvantages such as the complexity of the system structure, low compatibility and expansibility, time consuming in development and deployment, low security and anti-interference in the smart lighting system based on Power Line Communication (PLC), a new smart lighting system based on IPv6 over Low power Wireless Personal Area Network (6LoWPAN) was proposed in this paper. An example of implementing 6LoWPAN in smart lighting system by replacing PLC was presented, the PLC nodes were replaced by 6LoWPAN nodes, the central controller was replaced by border router, and the Constrained Application Protocol (CoAP) and Internet Protocol for Smart Objects (IPSO) application framework was applied in the application layer. Compared with the smart lighting system based on PLC, the new smart lighting system with 6LoWPAN technology is simpler in system architecture, it has higher compatibility and expansibility, the development and deployment time is reduced more than 50%, the network security and anti-interference is better.

Error propagation seriously degenerates the selection diversity of decode-and-forward-based opportunistic cooperative system. Addressing this problem, a Log Likelihood Ratio (LLR)-based adaptive switching scheme was proposed, which aimed at exploiting relay channel more efficiently through dynamic cooperation selection according to the LLR comparison with Bit Error Ratio (BER)-based threshold at relay. Moreover, the closed-form expression of the average BER was derived, and the threshold was optimized accordingly. Monte-Carlo simulations validate the analysis, and the results show that the proposed algorithm achieves 1.2dB power gain at BER of 0.001, compared to the conventional scheme.

In order to improve image retrieval performance, this paper proposed a new image retrieval algorithm based on motif and color features. The color image edge gradient was detected, and by means of edge gradient image transform, a motif image was obtained. Adopting the gravity center of motif image as the datum point, the distances of all points were calculated to the datum point to get the motif center distance histogram. The all motifs of the motif image were projected in four different directions to get motif projective histogram. Color image was uniformly quantized into 64-color space from RGB space to obtain the color histogram. The above three histograms described image features for image retrieval. The experimental results show that the algorithm has high precision and recall.

During the restoration of actual motion blurred images based on Wiener filtering, restoration results get affected by serious ringing effect and unsatisfactory local restoration. Its main reasons were found through theoretical analysis, experimental comparisons and the study of the characteristics of the actual motion blurring process. It was proposed that the artificial boundary compensation and block-restoration were used to restrain ringing effect and local unsatisfactory restoration. The relations of blur parameters, space positions and speeds, even the standard of blocking partition were given. The experimental results verify that the proposed method of the boundary compensation and the block-restoration can effectively reduce ringing effect and maintain the consistency of the overall image restoration effect.

A new method of filtering mixed noise based on limited grayscale and Pulse Coupled Neural Network (PCNN) was proposed for an image contaminated by salt and pepper noise and Gaussian noise. First, salt and pepper noise was identified according to the limited grayscale in a detecting window. Then the noise was filtered via mean filter in a filtering window. Subsequently, Gaussian noise was identified by using the time matrix of PCNN. Finally the Gaussian noise was filtered by some different filters based on variable step. The experimental results show that the proposed method has more advantages not only in filtering effects but also in objective evaluation indexes of Peak Signal-to-Noise Ratio (PSNR) and Improved Signal-to-Noise Ratio (ISNR) compared to some traditional methods.

A novel text zero-watermarking algorithm based on chaotic mapping was brought forward.The features of presented method are as follows: First,there are no modifications to the text document among the process of watermarking embedding,thus the watermarking is perceptually invisible.Second,the watermarking is robust against copying,cutting and formatted modifying.Third,the watermarking can be detected only by using a secret key and doesn’t need the original text.

A simple,easy but efficient approach,based on modified and piecewise fuzzy Gibbs random fields was introduced.A fine initial classification was got by modified FCM(Fuzzy C-Means),then combined with the two classes fuzzy method,a complex segmentation was got efficiently and precisely.Experiments show that our approach is more reliable and effective than classical methods in multi-class image segmentation.

A new approach to finding the core point of fingerprint was given. Based on the location of core point, fingerprint images were classified by using structural classification and were matched by using graphical matching. The core point method was combined with neighborhood matching approach and a new compound matching approach based on fuzzy discrimination was presented. In this approach, core point seeking once more based on the characteristics of fingerprint images was presented. Experiments were done on 1000 fingerprint images (20 percent of the images of low quality). 100 percent of the images were classified rightly and 98.7 percent of the images were matched rightly.