Local color transformation analysis for sudden illumination change detection

• Illumination change detection is carried out without assuming a specific color transformation.
• The method is designed to improve existing standard foreground detection algorithms.
• Backup procedures are designed to recover from detection failures.
• The method is applicable to a wide range of varying illumination conditions.

Sudden illumination changes are a fundamental problem in background modeling applications. Most strategies to solve it are based on determining the particular form of the color transformation which the pixels undergo when an illumination change occurs. Here we present an approach which does not assume any specific form of the color transformation. It is based on a quantitative assessment of the smoothness of the local color transformation from one frame to the background model. In addition to this, an assessment of the obtained illumination states of the pixels is carried out with the help of fuzzy logic. Experimental results are presented, which demonstrate the performance of our approach in a range of situations.

Figure optionsDownload high-quality image (78 K)Download as PowerPoint slide