A MATLAB-Based Ghost Detection Algorithm Development in the Context of Stray Light Test

Abstract

High dynamic range (HDR) video cameras with modern, non-linear CMOS image sensors can provide not only better visual expression, but also a new and powerful driving force for the development of advanced driver assistance systems (ADAS).However, mechanical defects, impurities, and imperfections in image sensors or lenses are inevitable in actual volume production.As a result, light refl ections can cause disturbing ghost artifacts in received images. Eff ectively detecting these artifacts in dynam-ic scenes, camera jitter, inaccurate image registration, and large displacements of moving objects is a key problem. Therefore, a comprehensive understanding of the currently unknown characterization of ghost artifacts must be improved. Currently, there is no automated method based on image processing tools to identify ghost artifacts in HDR images obtained from camera systems. In this paper, an algorithm to detect ghost artifacts has been developed and evaluated for future automated manufacturing test systems using Matlab. The result of the fi rst module of the algorithm indicates that not only the existence of ghost artifacts can be detected,but also the type of ghost can be determined in single testing images with an acceptable reliability of image processing methods.Furthermore, a typical feature of ghost artifacts can be extracted by identifying the signifi cant portion of the ghost as a “Ghost Ring” from the image with normal ghost within a reasonable level of accuracy and success rate.