Stray Light Of Focal Plane Array And Stray Light Reduction Using Image Restoration Algorithm

Stray light is the light other than the intended light in the imaging system. With theincrease in performance demand of optical instrument in various fields, the problem ofstray light has become prominent. Existing studies mainly focus on stray light in opticaland optomechanical systems. However, the stray light inside Focal Plane Array (FPA)detector is an unnegligible source of stray light in imaging systems. In this thesis, we firstresearch on the generation mechanism of stray light. Then, based on the results, wepropose two approaches to reduce stray light inside FPA of imaging systems. Moreover, forthe acquired images already contained by stray light we design a novel algorithm to reducethe stray light of images.In the first chapter, the signification and research background of this issue isintroduced and reviewed.In the second chapter, we first build models of FPA. Both front-illuminated andback-illuminated Charge Coupled Device (CCD) is investigated. Then we analyze thegeneration mechanism of stray light in FPA and find the stray light of FPA is visible underpowerful incident light. Finally, the conclusions are verified through experiments.In the third chapter, we propose two hardware methods, i.e., anti-reflective coatingmethod and fluid-filled method, to reduce the stray light in FPA. Their effectiveness isdemonstrated via extensive simulation and analysis.In the fourth chapter, based on the research on characteristics of stray light, ashifted-variant degradation model of imaging optical system is developed. Moreover,evaluation metrics for images with stray light are proposed.In the fifth chapter, we present a segmentation processing algorithm for image with stray light. The segmentation algorithm is combined respectively with three existing imagerestoration algorithms, i.e., inverse filtering algorithm, Richardson-Lucy iterativealgorithm, and Van-Cittert iterative algorithm. We evaluate the performance in removingstray light of these improved algorithms and compare the metrics of images post-processedby these algorithms.In the last chapter, we conclude the thesis and prospect the further work of ourresearch.