Video stabilization is important in many application fields, such as visual surveillance. Video stabilization and completion based on a single camera have been well studied in recent years, but it remains a very challenging problem. In this paper, we propose a novel framework to produce a stable high-resolution video for visual surveillance by using two cameras, in which one static camera serves to capture low-resolution wide-view-angle images, and the other is a pan-tilt-zoom camera to capture high-resolution images. Different with using a single camera, the interesting target can be detected and tracked more effectively and much more high-resolution information can be utilized for the stabilization and completion by using two videos from two cameras. A three-step stabilization approach is designed to deal with the resolution's discrepancy between two synchro videos and a four-stage completion strategy is taken to utilize more high-resolution information. Experimental results show that the proposed algorithm has a satisfying performance.

Depth information is a critical cue for scene understanding which is very important for video surveillance. Traditional depth estimation approaches based on stereo vision have been well studied in past decades. However, little research is conducted with active cameras. In this correspondence, we discuss the depth estimation problem by employing dual-PTZ (pan/tilt/zoom)-camera system. The contributions of this correspondence includes the following three aspects: 1) we analyze the effect on depth's precision with different PTZ settings; 2) we propose a coarse-to-fine framework to deal with depth estimation problem for complex region; 3) we offer a method to generate mosaic of depth maps to combine depth information from different visual angles and resolutions for large scope. These contributions will widen the usage of dual-PTZ-camera systems to a greater extent. Real-data experimental results show that the proposed approach works well.

In this paper we proposed a novel stereo rectification method for dual-PTZ-camera system, which is called spherical rectification. This method can be divided into two parts, offline pre-settings and online rectification. The offline pre-settings include camera calibration and building the spherical coordinate system. The online rectification only requires the pan, tilt and zoom values and does not need any image information such as feature points. So, compared with traditional rectification approaches, our method is more convenient and time-saving.

Utilizing more information other than minutiae is much helpful for large-scale fingerprint recognition applications. In this paper, we proposed a polynomial model to approximate the density map of fingerprints and used the model's parameters as a novel kind of feature for fingerprint representation. Thus, the density information can be utilized into the matching stage with a low additional storage cost. A decision-level fusion scheme is further used to combine the density map matching with conventional minutiae-based matching and experimental results showed a much better performance than using single minutiae-based matching.