Super-Resolution Reconstruction Of Array Image

Super-resolution image reconstruction is a hot issue both in the computer vision field and in the image processing field. The array camera, composed of multi-array lens, has become a hot topic with the development of virtual reality, high-speed cameras. But the array camera is not easy to carry for it’s big size, and the microarray camera becomes the hot issue in the study of the camera. This paper studied Super-resolution image reconstruction based on the microarray camera, which is applied on the handheld image capture device. The main work and innovation of this paper including:1) Image calibration. The mathematical limit principle is used for image calibration. It divided the image into blocks, which can be very small by the mathematical limit principle, to be calibrated.2) The improved regularization method is used for the super-resolution image reconstruction after calibrating the array. In the image reconstruction the image blur is introduced as the weight of each image, which can determined the influence of each image on the reconstruction. And the influence on the reconstruction of the fuzzy images is weakened so the quality of the reconstructed result is improved.3) Sharpness enhancement. The final result of super-resolution reconstruction is out by using the geometric locally adaptive sharpening nuclear enhancing.The simulation experiment verified that the image parallax caused by the position angle can be eliminated using a plurality of calibration matrix based on the principles of the mathematical limit. The subjective visual of the image reconstructed by the BTV method with the blur as images’ weight is good than the traditional BTV method according to the characteristic of the array image. And the image definition evaluation operator is also higher. The experiment also got a good result.