무안경식 입체 다시점 모니터의 엘리어싱 감소를 위한 필터링 기법

Abstract

요 약본논문에서는무안경식다시점입체모니터에서다시점 차원영상을시청할때발생하는시점엘리어싱 및시점3 (view aliasing)간엘리어싱 을제거하기위해변위정보에적응적인필터를제안한다일반적으로다시점영상의해상도가(inter-perspective aliasing) .다시점모니터의해상도보다크면입력되는다시점영상의부표본화 로인하여시점엘리어싱이발생하게된다또한무(subsampling) .안경식다시점모니터에서관찰자가영상을시청할때시점이동시불연속적인구간으로인해발생하는시점간엘리어싱때문에눈의피로를느끼게된다이러한엘리어싱현상은입체영상을시청할때에변위정보가클수록더욱강하게나타난다따라서본논. . 문에서는중간영상생성개수와변위정보의강도에따라필터의강도를조절하는저역통과필터링기법을제안하였고제안한필, 터링에의해엘리어싱의영향이줄어드는효과를DSCQS (double stimulus continuous quality scale method) 방식을이용하여실험을통해검증하였다.AbstractIn this paper, we propose a disparity-adaptive filtering method to reduce view-aliasing and inter-perspective aliasing onauto-multiscopic 3D display. View aliasing by the subsampled multi-view images could happen if the resolution of anauto-multiscopic monitor is lower than that of the original multi-view images. Furthermore, multi-view images on auto-multiscopicmonitor usually cause eye strain and fatigue because of inter-perspective aliasing by discontinuity that exists between viewpoints.The greater disparity value becomes, the stronger these types of aliasing could be. Thus, we design a lowpass filter whose cut-offfrequency is determined adaptively to the number of intermediate reconstructed views and their disparity's strength. Thoughexperimental results, we show that the proposed filtering algorithm could reduce the aliasing effect very efficiently by usingDSCQS (double stimulus continuous quality scale method).Key Word : auto-multiscopic monitor, view aliasing, inter-perspective aliasing, disparity, low-pass filter