Vacant Parking Space Detection Based on a Multilayer Inference Framework

In a practical environment, the viewing angle and height of a video surveillance camera are uncontrollable. This may cause severe interobject occlusion and complicate the detection problem. In this paper, we proposed a novel inference framework with multiple layers for vacant parking space detection. The framework consists of an image layer, a patch layer, a space layer, and a lot layer. In the image layer, image patches were selected based on the 3D parking lot structure. We found that the occlusion pattern within each patch reveals cues of the parking status. Thus, our system extracted lighting-invariant features of patches and trained weak classifiers for the recognition of the occlusion pattern in the patch layer. The outputs of the classifiers, presenting the types of interobject occlusion, were treated as the mid-level features and inputted to the space layer. Next, a boosted space classifier was trained to recognize the mid-level features and output the status of a three-space unit in a probability fashion. In the lot layer, we regarded the local status decision of three-space units as high-level evidence and proposed a Markov random field to refine the parking status. In addition, we extended the framework to bridge multiple cameras and integrate the complementary information for vacant space detection. Our results show that the proposed framework can overcome the interobject occlusion and achieve better status inference in many environmental variations and under different weather conditions. We also presented a real-time system to demonstrate the computing efficiency and the system robustness.