Zixing Cai

Improved Single Image Dehazing Using Dark Channel Prior and Multi-scale Retinex

In this paper, we propose an improved image dehazing algorithm using dark channel prior and Multi-Scale Retinex. Main improvement lies in automatic and fast acquisition of transmission map of the scene. We implement the Multi-scale Retinex algorithm on the luminance component in YCbCr space, obtain a pseudo transmission map whose function is similar to the transmission map in original approach. Combining with the haze image model and the dark channel prior, we can recover a high quality haze-free image. Compared with the original method, our algorithm has two main advantages: (i) no user interaction is needed, and (ii) restoring the image much faster while maintaining comparable dehazing performance.

Automatic Image Haze Removal Based on Luminance Component

In this paper, we propose a simple but effective method for visibility restoration from a single image. The main advantage of the proposed algorithm is no user interaction is needed, this allows our algorithm to be applied for practical applications, such as surveillance, intelligent vehicle, etc. Another advantage compared with others is its speed, since the cost of obtaining transmission map is greatly cut down by using Retinex algorithm on luminance component. A comparative study and quantitative evaluation is proposed with the main present-day methods which demonstrate that similar or better quality results are obtained.

Chinese Traffic Police Gesture Recognition in Complex Scene

We propose a method to recognize Chinese traffic police gesture in complex scene for intelligent vehicle. The gesture recognition is made by integrated nonparametric background modeling with human pose estimation. Firstly, dark channel prior and kernel density estimation are used to extract polices torso and arms from complex traffic environment as foreground region. Then, the coordinates of pixels in the upper and lower arms are determined by using max-covering scheme, which is based on a key observation that body part tiles maximally cover the foreground region and satisfy body plan. Finally, some typical police gestures can be recognized by rotation joint angle. The experimental results show that this method can obtain favorable results on a number of gesture sequences of traffic police.

Universal strategy for surveillance video defogging

Abstract. We present a new approach to remove haze from surveillance video sequences. This approach extracts the background image through the frame differential method, uses the dark channel prior to estimate the atmospheric light, and then calculates a universal transmission map on the intensity component of the background image through a process of multiscale retinex, parameter adjustment, bilateral filtering, and total variation denoising filtering. Finally, it renders haze-free video according to the haze image model. The main advantage of the proposed approach is its speed as this approach adopts a “universal strategy” that applies the same atmospheric light and a universal pseudo-transmission map to a series of video frames. Experiments demonstrate that our method produces visually pleasing defogging results and tends to preserve main details better than previous techniques. A comparative study and quantitative evaluation show the efficiency of the proposed method.

Provision and Analysis of QoS for Distributed Grid Applications

Grid computing provides the infrastructure necessary to access and use distributed resources as part of virtual organizations. When used in this way, Grid computing makes it possible for users to participate in collaborative and distributed applications. This class of applications must adhere to stringent real-time constraints and Quality-of-Service (QoS) requirements. We discuss the design and a prototype implementation of a QoS system, and demonstrate how we enable Grid applications to become QoS compliant. Internet was partially simulated in our laboratory setup of Grid by assuming intermediate routers to be DiffServ enabled. Under such a scenario strict reservation of data paths for flows is not possible and without any dynamic traffic control services reduce to best effort. Experimental results included in this work demonstrate the impact of the architecture with an OGSA-enabled QoS service in the improvement in reliability and throughput of multimedia collaboration between participating nodes in the grid environment.

An objective assessment method for image defogging effects

Methods to identify image defogging effects can help evaluate and optimise image defogging algorithms. However, such methods that assess the performance of defogging algorithms or compare them with other algorithms are lacking, whereas existing assessment methods are mostly inadequate. Two new methods to assess defogging effects are therefore proposed in this study. The first method generates synthetic foggy images by using the transmission map of an image degradation model in a full-reference manner. The second method develops an assessment system from the perspective of human visual perception in a no-reference manner. Experimental results from a comparison of defogging algorithms demonstrate the effectiveness and reliability of our proposed methods. Compared with other existing methods, our proposed methods efficiently assess defogging effects from generated synthetic images and human visual perceptions. These methods represent novel approaches to measure defogging effects.

Fast Haze Removal Algorithm for Surveillance Video

In this paper, we present a new approach to remove haze from surveillance video sequences. The approach first extracts the background image through frames differential method, and then estimates the atmospheric light and transmission map from the background image, finally renders haze-free video according to the haze image model. The main advantage of the proposed approach is its speed since this approach adopts a ‘universal strategy’ that is applying the same atmospheric light and a single pseudo-transmission map to a series of video frames. Experiments and performance analysis demonstrate that a good haze-free video can be produced effectively and efficiently.