Stephan Simon

Detection of close cut-in and overtaking vehicles for driver assistance based on planar parallax

Image processing is widely considered an essential part of future driver assistance systems. This paper presents a motion-based vision approach to initial detection of static and moving objects observed by a monocular camera attached to a moving observer. The underlying principle is based on parallax flow induced by all non-planar static or moving object of a 3D scene that is determined from optical flow measurements. Initial object hypotheses are created in regions containing significant parallax flow. The significance is determined from planar parallax decomposition automatically. Furthermore, we propose a separation of detected image motion into three hypotheses classes, namely coplanar, static and moving regions. To achieve a high degree of robustness and accuracy in real traffic situations some key processing steps are supported by the data of inertial sensors rigidly attached to our vehicle. The proposed method serves as a visual short-range surveillance module providing instantaneous object candidates to a driver assistance system. Our experiments and simulations confirm the feasibility and robustness of the detection method even in complex urban environment.

On entropy coded and entropy constrained lattice vector quantization

Two lattice vector quantization methods are compared. The first, classical method uses Dirichlet domains of lattice points as quantization cells and assigns them reconstruction vectors minimizing the distortion. The second method uses the lattice as codebook but modifies the shapes of the quantization cells by searching for each input vector the mapping onto either the nearest lattice points or one of its neighbors, such that an error criterion subject to an entropy constraint is minimized. Both methods use entropy coding for the codevector indices and an entropy constrained global optimization scheme to find the best lattice scale. Examples demonstrate that the rate-distortion performances of both methods are nearly identical. Hence, especially for non-stationary sources the selection of one of these methods should be based on other criteria: while the first method requires the transmission of the new codebook, the decoder for the second method can permanently adapt itself without side information. The drawback is a higher search complexity for encoding.