Motion compensation for obstacle detection based on homography and odometric data with virtual camera perspectives

Abstract

In this paper we present a method to compensate the image motion of a monocular camera on a moving vehicle in order to detect obstacles. Due to the camera motion, the road surface induces a characteristic image motion between two camera shots. The motion of the camera is determined by the use of odometric data received from the CAN-bus, and the position and orientation of the road is continuously estimated with camera self-calibration. This all leads to a motion field which is predicted based on homography. To prevent the drawbacks of the real camera perspective, different virtual camera perspectives are presented in combination with motion compensation. Possible virtual perspectives are the bird's eye view and image rectification. In addition, a non-linear camera model is used which does not limit the range of obstacle detection to a certain distance and efficiently uses the available image information.

A single camera based rear obstacle detection system

“…INVERSE PERSPECTIVE MAPPING BASED COARSE DETECTION Stereo IPM has achieved good performance for obstacle detection [6] [7].The stereo IPM can keep the non-plane objects and remove the plane objects such as lane-markings, shadows by comparing the differences between the left and right remapped image [6]. For a single camera use IPM for obstacle detection, temporal frame difference is used to simulate the effects of stereo cameras [5][8][9] [10]. Single camera based IPM methods usually need car ego motion parameter to compensate the displacements between two frames to eliminate false positives [5] [10].…”

“…For a single camera use IPM for obstacle detection, temporal frame difference is used to simulate the effects of stereo cameras [5][8][9] [10]. Single camera based IPM methods usually need car ego motion parameter to compensate the displacements between two frames to eliminate false positives [5] [10]. However, it is a complicated task to compute the car ego motion parameter accurately.…”

“…The property is: the prolongation of the edges of obstacles should pass the Focus F on IPM image [5]. We use F as midpoint to set an interval by using a threshold Th (10). We check the prolongations of the principal axis of each ellipse.…”

Robust ground plane induced homography estimation for wide angle fisheye cameras

“…RELATED WORK Estimating motion with respect to the ground plane from monocular image sequences is a task frequently occurring in computer vision applications. For example, Kitt et al [7] use the relative motion with respect to the ground plane to compensate for scale drift in bundle adjustment and Miksch et al [8] present an approach for obstacle detection based on the ground plane. Stein et al [11] estimate velocity, pitch, and yaw using a direct image based method in a probabilistic framework.…”

Simultaneous estimation of ego-motion and vehicle distance by using a monocular camera

“…As an important problem in FCW systems, vision-based vehicle detection for driver assistance has received considerable attention over the last 15 years [10]. And most recently, obstacle detection algorithms [11] and real applications, e.g. the car named Lexus LS430 1) , have made much headway along with the exponential growth in processor speeds and the advances on computer vision research.…”