Reconstruction Of Super-resolution Depth Image Based On Compressive Sensing And Development Of Camera Array

This paper addresses the problem of reconstructing a super-resolution depth image based on compressive sensing theory and the degradation model of imaging optical system. Reconstruction of super-resolution (SR) image is the image processing technology of generating higher resolution images from the original low-resolution images. The principle of compressive sensing (CS) ensures that the sparse representation can be correctly recovered from the down sampled signal under mild conditions.The problem consists of designing a stable measurement matrix, a transform matrix, a reconstruction algorithm to recover the signal, and the degradation model. The algorithm is mainly used accompanied with TOF cameras which are a depth acquisition system with fixed resolution. We illustrate the strong points of our approach for objects with different geometric features compared with interpolation algorithm. This paper indicates how to select parameters according to the simulation results.This paper also presents an intelligent IP camera array with global synchronization function. One of the cameras works in master mode and generates the synchronization trigger signal for other cameras in slave mode, which ensures the synchronous capture of every camera. Every intelligent IP camera has a high-resolution image capture capabilities and can transfer the commands and images via Ethernet. The camera array could process image in real-time, and form a distributed computing architecture with the center computer.