In order to realize the hardware real-time correction of smear phenomenon based on the interline transfer area CCD. A smear extraction method based on the dark pixel region was put forward and the smear phenomenon in the large area CCD was eliminated. Firstly, the smear signal from the dark pixel region was collected, and a real-time median filter algorithm was designed based on the FPGA to filter the extracted smear signal. The influence of random noise was eliminated. Then, in the radiating experiment, the effective pixel regional saturation threshold value that the regional signal of light leak embodies was determined. By the weighted average compensation method, the smear region whose digital value is larger than the saturation threshold value was compensated. Finally, the smear correction was completed. Experimental result shows that using the real-time median filter algorithm based on the FPGA to filter the extracted smear signal, the system delay time is only 3 lines of pixel reading out time, which meets the requirements of the real-time system fully. The smear removing algorithm which is combined with the weighted average compensation restores image well, and the smear phenomenon is eliminated completely.