Lane line detection is an important part of intelligent driving system. The traditional lane line detection method relies heavily on manual selection of features, which requires a large amount of work and has low accuracy when it is interfered by complex scenes such as object occlusion, illumination change and road abrasion. Therefore, designing a robust detection algorithm faces a lot of challenges. In order to overcome these shortcomings, a lane line detection model based on deep learning instance segmentation method was proposed. This model is based on the improved Mask R-CNN model. Firstly, the instance segmentation model was used to segment the lane line image, so as to improve the detection ability of lane line feature information. Then, the cluster model was used to extract the discrete feature information points of lane lines. Finally, an adaptive fitting method was proposed, and two fitting methods, linear and polynomial, were used to fit the feature points in different fields of view, and the optimal lane line parameter equation was generated. The experimental results show that the method improves the detection speed, has better detection accuracy in different scenes, and can achieve robust extraction of lane line information in various complex practical conditions.

Machine learning based image Super-Resolution (SR) has been proved to be a promising single-image SR technology, in which sparseness representation and dictionary learning has become the hotspot. Aiming at the time-consuming dictionary training and low-accuracy SR recovery, an SR algorithm was proposed from the perspective of reducing the inconsistency between Low-Resolution (LR) feature and High-Resolution (HR) feature spaces as far as possible. The authors adopted Iterative Least Squares Dictionary Learning Algorithm (ILS-DLA) to train LR/HR dictionaries and Anchored Neighborhood Regression (ANR) to recover HR images. ILS-DLA was able to train LR/HR dictionaries in relatively short time because of its integral optimization procedure, by adopting the same optimization strategy of ANR, which theoretically reduced the diversity between LR/HR dictionaries effectively. A large number of experiments show that the proposed method achieves superior dictionary learning to K-means Singular Value Decomposition ( K-SVD) and Beta Process Joint Dictionary Learning (BPJDL) algorithms etc., and provides better image restoration results than other state-of-the-art SR algorithms.

The existing general no-reference image quality assessment methods mostly use machine learning method to learn regression models from training images with associated human subjective scores to predict the perceptual quality of testing image. However, such opinion-aware methods expend much time on training, and rely on the distortion types of the training database. These methods have weak generalization capability, hereby limiting their usability in practice. To solve the database dependence, a normalized scale invariance based no-reference image quality assessment method was proposed. In the proposed method, the Natural Scene Statistic (NSS) feature and edge characteristic were combined as the valid features for image quality assessment, and no extra information was required beyond the testing image, then the two feature vectors were used to compute the global difference across scales as the image quality score. The experimental results show that the proposed method has good evaluation for multi-distorted images with low computational complexity. Compared to the state-of-the-art no-reference image quality assessment models, the proposed method has better comprehensive performance, and it is suitable for applications.

Temporal flickering in the video is a key factor of affecting the quality of video. Accurate identification of temporal flickering is required for the automatic analysis and diagnosis of video quality. Moreover, it can be integrated with artifact removal and quality enhancement algorithms to promote the adaptivity of the proposed algorithm. A study of temporal flickering in video surveillance was given to demonstrate that the differential signal of temporal flickering in time domain follows the Laplacian distribution. Motivated by this statistical observation and the idea of small probability events, the proposed method iteratively segmented differential signal of motion in foreground, which affected the identification of temporal flickering. Furthermore, the proposed approach exploited the Just-Noticeable Difference (JND) mechanism of the human visual system to identify the temporal flickering using the flickering frequency and amplitude. The proposed method yielded superior performance to that of the conventional Gaussian Mixture model to achieve more accurate classification of the normal video and temporal flickering video, as verified in the ROC (Receiver Operating Characteristic) curve presented in experimental results. The proposed no-reference algorithm is able to achieve fairly good performance in temporal flickering identification.

In order to solve the problem of auditing the data protected by Secure Sockets Layer/Transport Layer Security (SSL/TLS) protocol on the Internet, a plaintext gathering method for network data protected by SSL/TLS protocol based on the principles of man-in-the-middle was proposed. A data gatherer was connected between the server and the client in series, which was able to get the encryption key by modifying handshake messages during SSL/TLS handshake, so as to decrypt the secure data and then gather its plaintext. Compared with the existing gathering method based on the principles of proxy server, the proposed method has a shorter transmission delay, a larger SSL throughput and a smaller memory occupation. Compared with the existing gathering method in which the gatherer possesses the server's private key, the proposed method has a wider application scope, and also has the advantage of being unaffected by packet losses on the Internet. The experimental results show that the proposed method has a decrease in transmission delay of about 27.5% and an increase in SSL throughput of about 10.4% compared with the method based on the principles of proxy server. The experimental results also show that the SSL throughput of the proposed method approaches the ideal maximum value.

To solve the problems of the ringing, blocking artifacts and the excessive noise amplification in local image enhancement, an object-based multilevel contrast stretching method was proposed. First, segment the image into its constitent objects by using morphological watersheds and regional merging; then, separately stretch the image contrast at inter-object level and intra-object level in different ways. At inter-object level, an approach of stretching between adjacent extrema was adopted to adequately enlarge the local dynamic range of gray levels between objects; at intra-object level, the linear stretching approach was adopted to enhance the textural feature of the object and keep its appearance. Experimental results show, besides enhancing the image structure, the proposed method can effectively avoid ringing, blocking artifacts, restrict excessive noise amplification in smooth areas and preserve the overall brightness of the image, thus can provide the enhanced image with natural appearance.

Data fusion is an important technique to detect and recognize target from images. However, loss of information is unavoidable in fusion process. It is crucial to select fusion algorithm and avoid the loss of useful information. Because images in different levels of image pyramid were self-similar, which was the foundation of fractal, a new image fusion algorithm was presented based on image pyramid and fractal dimension. Different source images were decomposed into image pyramid sequence with different scales. Corresponding level images were merged, with fractal dimension as weight. An experiment was given using a mid-wave and a long-wave infrared image as two source images. Experiment results show the algorithm is feasible.