Robot vision
A survey of thresholding techniques
Computer Vision, Graphics, and Image Processing
| Hi-index | 0.00 |
The present contribution presents a novel stand-alone surveillance system based on detection and tracking of moving objects using CMOS imagers. The system fuses the acquisition and processing task, so that a flexible, low-cost, and highly efficient image processing system for home and industrial applications can be realised. The surveillance system consists of a CMOS camera mounted on a multi-functional module (MfM) using a PCMCIA 10 Mbit Ethernet card. The acquired images can be transmitted directly via Internet protocol to the requesting workstations. The surveillance camera system is capable of suppressing extraneous illumination effects caused for example by day light variations. The detection of moving objects in the image sequence is carried out using a novel adaptive image processing algorithm. The required parameters are adapted automatically to the observed scene. The apparent shapes of independently moving objects are extracted with high accuracy even for objects that are stationary for certain time periods. The approach also takes into account slowly changing illuminations, and does not detect them as moving objects. It works very well with severely image sequences corrupted by noise. The presented simulation results highlight the performance of the proposed method using synthetic and real video data.