3D model generating method based on multi-views deblurring images

The image of moving target which captured though image acquisition system in mobile platform is contaminated by blur and noise, then it make the 3D(3 dimension) model of target difficult to reconstruct. A new method of 3D Model generating is proposed for this question. Firstly, detect SURF characteristic points of deblurred images and match it. Obtain the camera parameter matrix through solve Kruppa equation according to character points which is matched. Then, estimate the points of partly deblurred images and get the blur kernel of the other images with camera model which obtained in the previous step. Finally, according to the camera parameter matrix make space transformation for the pixels of images. A serials experiment validate high-efficiency of the proposed method.