Microlens array (MLA) errors in plenoptic cameras can cause the confusion or mismatching of 4D spatio-angular information in the image space, significantly affecting the accuracy and efficiency of target reconstruction. In this paper, we present a high-accuracy correction method for light fields distorted by MLA errors. Subpixel feature points are extracted from the microlens subimages of a raw image to obtain correction matrices and perform registration of the corresponding subimages at a subpixel level. The proposed method is applied for correcting MLA errors of two different categories in light-field images, namely form errors and orientation errors. Experimental results show that the proposed method can rectify the geometric and intensity distortions of raw images accurately and improve the quality of light-field refocusing. Qualitative and quantitative comparisons between images before and after correction verify the performance of our method in terms of accuracy, stability, and adaptability.
Keywords: error correction; light-field image processing; microlens array; plenoptic camera.