Time characterization of video acquisition flow

This paper presents a contribution to time characterization for video data acquisition flow. Video flow is used in many robotics and embedded systems in vehicles. For real time applications, the delay of a video acquisition is very slow compared with other sensors. Indeed, the data capture of an environment consists in sampling this environment at a given moment. The embedded computer will be able to mark temporally these pieces of data when they are available in its internal memory. The time lag, between the moment of sampling in the sensor and the presence of the image in the computer memory is thus unknown. This delay induces a temporal shift on the same scene seen by several sensors of various types (ultrasonic, telemeter, etc). The fusion of data of these various sensors will be then more difficult and induces important degradation over the decision. It is thus important to be able to characterize this latency time of the cameras and to resynchronize more precisely all the data of the various sensors. In this paper, we propose a solution to characterize the latency of the video acquisition system flow. This work was completed in collaboration with Intempora company.