Performance of a MVE Algorithm for Compound Eye Image Reconstruction Using Lens Diversity

Super-Resolution Image Reconstruction for Diverse Sub-Imager Arrays

“…Simulations of several diverse architectures show that even a small amount of diversity improves the performance as measured by expected mean squared error [6,7]. Similar performance was achieved for a variety of systems, and the improvement does not rely on any precise relationship between the sub-imagers parameters.…”

“…The Panoptes approach to flat camera design is different from these two applications in several respects [6][7][8][9]. The high resolution images produced by the system will be used by automated image analysis systems, so the measure of performance is the information captured in the reconstruction rather than measures based on human perception.…”

“…Several new computational imaging applications have been reported in recent years [1][2][3][4][5][6][7][8][9] using methods similar to those used in radio astronomy [10] and medical image reconstruction applications [11]. If a moving low resolution camera provides a sequence of images of a stationary scene from multiple shifted perspectives, image registration and reconstruction algorithms can be used to create a higher resolution image that looks the same as an image captured by a more expensive camera [2].…”

“…Like the TOMBO system, the low resolution images are acquired in parallel from an array of sub-imagers, but active control of the look direction of individual sub-imagers through micromirror arrays allows variable resolution improvement and variable object range. In addition, diversity in the sub-imager array design is used to reduce the reconstruction error by eliminating the null space due to a uniform sensor array [6,7,9].…”

Artificial compound eye: a survey of the state-of-the-art

“…In Nitta et al (2006), to further improvement of the method, Nitta et al introduced an iterative back projection method (IBP), where the pixel rearrange method was first applied to obtain the initial image and the back projection method was iteratively applied to refine the image. Except for the pixel rearrange method, lots of image reconstruct methods were also discussed in planar artificial compound eye (Kanaev et al 2007;Boussaid 2008, 2009;Horisaki et al 2009;Oberdörster et al 2011;Mendelowitz et al 2013;Wood et al 2004Wood et al , 2005Wood et al , 2006Li et al 2007Chan et al 2006b;Choi and Schulz 2008;Tudela et al 2008b, a). Kanaev et al (2007) presented a scene-independent method, in which the final image with high resolution could be reconstructed only based on the accurate relative pixel shifts.…”