Determining parking lot occupancy from digital camera images

Parking lot occupancy is determined from digital images and includes a set-up procedure that receives a layout of a parking lot 201and estimates parking space volume 202 for at least one viewing angle and determines probability that an observed pixel belongs to the parking space volume. One or more images frames of the parking lot are acquired 203 from at least one digital camera and pixels are classified 204 using a vehicle detector to determine a likelihood that a pixel belongs to a vehicle. The probability that a parking space is occupied 205 by a vehicle is computed based on a spatially varying membership probability density function and a likelihood of vehicle pixels within a region of interest. Parking lot vacancy is determined 206 by comparison of the computed probability that a parking space is occupied by a vehicle to a pre-determined threshold. Pixel classification may be performed by a support vector machine classifier that uses rotation invariant local binary patterns or a TextonBoost classifier.