This paper presents unsupervised vision-based system for parking lot occupancy detection. The proposed method exhibit low computation complexity and use just a few frames per minute. Method is based on three main processing stages. In the first section, raw image acquired by camera system is preprocessed. Shadows in the image are significantly attenuated or completely removed. The image distortion is corrected subsequently. In the following step optimal correspondences between one or more stationary cameras and visible parking places are established. During this process occlusions are taken in account. Finally, a parking place status is evaluated. Acquired information about parking lot occupancy can be served to another system, e.g. intelligent transportation system.Experimental results from both artificial and real scenes show promising results even on quite challenging conditions. Nonetheless a scope for improvement of presented method is given.

No abstract

No abstract

No abstract

No abstract