Investigates Image data fusion techniques that combine image and track data from multiple sensors to achieve improved accuracies and more specific inferences than could be achieved by the use of a single sensor alone. Our aim is to explore the state-of-the-art image processing algorithms for achieving effective data fusion as in:
1. Introduction[Go Top]
- Under fog weather conditions, images of outdoor scenes suffer from poor contrast. According to physics, "Air-light" caused by scattering of environmental illumination by particles in the atmosphere, that is fog, mist or etc. The air-light makes the image low-contrast and low-saturation. Therefore, most outdoor vision systems, especially surveillance system, require enhancing foggy image.
2. Main algorithm and principle[Go Top]
- "Air-light" is estimated by finding minimum value of difference between standard deviation function of normalized brightness and transformed normalized brightness. The estimated value is very useful to make luminance of image higher contrast than original foggy-image. Also, in order to correct color component, it is proposed the function that present the relation between the estimated value (air-light) and saturation value of image.
3. Demo System[Go Top]
- Defog system is implemeted by Visual C++ 2008 with OpenCV library. Now, using 30-fps(frames-per-second) video input from MATROX frame grabber, the system has about 20-fps performance in 648*480(VGA) color-mode.