The effect of image compression on automotive optical flow algorithms

Advantageous interconnections of automotive cameras and electronic control units are possible by using IP based technology within the vehicles. Automotive implementation of this kind of technology enables limited data rates. Camera streams have to be compressed for the transmission over such a medium. The image processing algorithms can be applied after a decompression of the camera streams. In this paper we propose a method to compare the effect of different image compression algorithms on the optical flow. It comprises a metric, based on a nearest neighbour method, to compare optical flow fields.