Fail output algorithm of vision sensing

Vision sensor has become a recent adoption as a potentially effective means for external sensing for automobiles. In the field of external sensing, a number of options for lane detection, distance recognition and other parameters have been proposed. Among others, the most important and yet difficult challenge is to establish a failure detection method for each recognition algorithm in order to prevent an error in system's detection under actual conditions where the sensor is used. This paper reports on one version for detecting failure in the vision sensor that recognizes obstacles in front of the vehicle in motion. The study was based on an innovative algorithm that judges the limit of vehicle recognition using a stereo vision camera under adverse weather conditions, in particular in the rain.