作者: Kartik Venkataraman , Dan Lelescu , Jacques Duparré , Andrew McMahon , Gabriel Molina
关键词: View synthesis 、 Pixel 、 Computer graphics (images) 、 Computer vision 、 Image restoration 、 Channel (digital image) 、 Parallax 、 Mobile device 、 Light field 、 Camera auto-calibration 、 Computer science 、 Image quality 、 Lens (optics) 、 Artificial intelligence 、 Superresolution 、 Depth map
摘要: We present PiCam (Pelican Imaging Camera-Array), an ultra-thin high performance monolithic camera array, that captures light fields and synthesizes resolution images along with a range image (scene depth) through integrated parallax detection superresolution. The is passive, supporting both stills video, low capable, small enough to be included in the next generation of mobile devices including smartphones. Prior works [Rander et al. 1997; Yang 2002; Zhang Chen 2004; Tanida 2001; 2003; Duparre 2004] arrays have explored multiple facets field capture - from viewpoint synthesis, synthetic refocus, computing images, speed micro-optical aspects system miniaturization. However, none these addressed modifications needed achieve strict form factor quality required make array cameras practical for devices. In our approach, we customize many lenses, pixels, sensors, software algorithms imaging comparable existing phone cameras.Our contributions post-processing include cost function integrates across color channels, regularized restoration (superresolution) process takes into account all degradations adapts conditions. registration uncertainty Maximum-a-Posteriori formulation estimate scene depth. conclude some examples capabilities such as postcapture (still) video view synthesis demonstrate motion parallax, 3D briefly address future work.