Automatic parking lot mapping for available parking space detection

This paper presents a method to estimate map of parking lot for automatic system of available parking space detection. The parking space detection is important module for the parking guidance system (PGS) that can help drivers to find the parking space efficiently. Instead of masked parking slots by human, we try to mask them by using histograms of spatial features. However, under poor image quality, we face dynamic change of background. Although we have applied adaptive background model for background subtraction, the problem still remains. To overcome this problem, we propose an automatic thresholding for tuning object (car) detection in adaptive background model method. The fine-tuning technique can improve accuracy of the histogram for the parking lot mapping. The experimental results were shown both simulation on the parking model and real parking lot correctly.