Stream Data Processing Systems with Progressive Quality Improvement

Abstract

AbstractOne of the main applications of stream data processing systems is detecting objects in video data streams. In such stream data processing systems, each camera device sends its recorded video data to the remote processing computer for the detections in cases that the computational resources of the camera devices are insufficient for the detections. To improve the performances of the stream processing systems for object detections, most methods reduce the communication traffic between the cameras and the processing computers. Although object detection processes do not always require the predetermined original data quality, the processing computers of conventional systems always receive original quality data. By considering the necessity of original quality data, we can reduce redundant communication traffic in some situations and can improve the performance indexes of the stream data processing systems. Hence, in this book chapter, we explain a Progressive Quality Improvement (PQI) approach to further improve the performances. In the PQI approach, for only the case that the data for higher qualities are needed, the processing computer progressively collects them from the data sources. Moreover, we implement a video surveillance system with the approach.