J. Thiem, T. Eisenbach, S. Drüe, and G. Hartmann (Germany)
Image Enhancement, Noise Reduction, Adaptive Filtering
Variations in illumination conditions always raise difficul ties when treating real-life scenes with digital computers. Especially random processes, that come along with the image-capture itself, e. g. read noise or photon shot noise, could lead to incorrect processing results. The human vi sual system, however, shows the advantageous ability to adapt to variations of the perceived scene, e. g. the degra dation of the illumination conditions, and in this way to improve further processing steps. With this contribution we propose a biology-inspired adaptive prefiltering for effective suppression of noise in image data. With the help of the described methods we can establish robust edge-detectors for computer vision tasks. In addition we can extand and improve existing models of the human vision model [9, 10] for extraction of contour features in scenes, which has been non-adaptive till up to now.
Important Links:
Go Back
