Finite impulse response low-pass digital filter based on particle swarm optimisation for image denoising

Abstract

Noise has very important influence on image quality. Finite Impulse Response (FIR) digital filter is an effective technique for removing image noise. The effectiveness of FIR digital filters designed by traditional methods in image denoising needs to be improved further because of their demerits such as lower accuracy. Therefore, to suppress noise and improve image quality, FIR low-pass digital filter designed by the improved particle swarm optimisation algorithm based on refraction principle (refrPSO) is employed. In the method, firstly, coefficients of FIR low-pass digital filter are mapped to a candidate solution of refrPSO, and then refrPSO is invoked to optimise it. Finally, the designed FIR low-pass digital filter is employed to denoise images. Experimental results of two images with different sizes show that the FIR digital filter designed by refrPSO has better image denoising effect compared with ones designed by window function method, best equivalent ripple approximation method, and PSO.