METHOD FOR ELIMINATING PERIODIC NOISE IN LIGHT FIELD RECONSTRUCTED IMAGE BASED ON FREQUENCY DOMAIN

摘要

Disclosed is a method for eliminating periodic noise in a light field reconstructed image based on a frequency domain. Specific steps are as follows: S1 collecting a light field imaged graph of a sample; S2 collecting a sample free light center position graph; S3 calibrating a center of each micro lens image, and performing rendering reconstruction on the light field imaged graph; S4 transforming a light field reconstructed image to a frequency domain to generate an image spectrum; S5 preprocessing the image spectrum; S6 generating a low pass filter; S7 multiplying the low pass filter with a pre-processed image spectrum, and zeroing a spectrum value of a low-frequency component; S8 performing binarization processing on a spectrum of the light field reconstructed image to obtain an image mask; S9 filtering out a high-frequency periodic noise component in an original spectrum of the light field reconstructed image; and S10 transforming the filtered spectrum of the light field reconstructed image back to a spatial domain to obtain a light field reconstructed image without periodic noise. The present invention can effectively eliminate the periodic noise in the light field reconstructed image, eliminate spliced edge traces, and optimize the reconstruction result.