Line Filter-Based Parking Slot Detection for Intelligent Parking Assistance System

摘要

A computer vision-based system is proposed to detect four types of commonly used parking slots for intelligent parking assistance system (IPAS). It consists of two submodules, i.e., around view construction and parking slot detection. A line filter is designed to abstract central points of parking lines. The designed line filter makes full use of the feature of parking lines that are brighter than its neighborhood. Geometry constraints of parking slots are used to localize parking slots. The proposed system has been validated on a model car equipped with SH7766 embedded platform and a Nissan vehicle. The accurate detection rate is 87.5 %, and the average position error of four corners of the parking slot is 10 cm. It is robust to light variation, strong shadow, and partial occlusion of parking lines.