In this paper, we propose a noble line-recognition algorithm of driving parking for intelligent vehicles by using images obtained from a bird’s eye view system. To achieve safe driving and parking of unmanned vehicles, we need to obtain noiseless and effective images around vehicles. In addition, fast image processing is a fundamental requirement for the real time recognition of lanes and obstacles to ensure safety. In fact, the number of sensors equipped with conventional unmanned vehicles is reluctant to their commercialization. To solve this problem, we propose a noble method to detect straight lines and turning curves in the images obtained from a bird’s eye view system. For conventional vehicles equipped with this bird’s eye view system, straight lines and turning curves are detected by using a Hough and trigonometric function. Since parking lines have the form of a rectangle or parallelogram, detection of their vertexs makes it possible to determinate parking areas. In the case of a parking space without parking lines, the parking space is detected by using a stereo algorithm after calculating the area for parking. The experimental results using the proposed algorithm show that, without additional sensors, the lines and the surrounding environment are detected for unmanned driving and auto-parking.