Underwater Images Enhancement Using Multi-Wavelet Transform and Median Filter

Autonomous underwater vehicles (AUV) are usually equipped with vision sensors. However, the underwater images captured by AUV often suffer from effects such as diffusion, scatter and caustics. So image enhancement methods are necessary to increase visual quality. A Median filter de-noising approach based on multi-wavelet transform was proposed to remove the impulse noise viewed as random noise from the blurred underwater image. Biorthogonal muti-wavelet has two scaling functions that may generate different multiresolution analysis, so it was chosen as the basic wavelet for underwater image two-layer decomposition and reconstruction. On this basis, the blurred underwater image was decomposed and reconstructed adopting Biorthogonal and the Median filter was applied for removing the impulse noise form the decomposition images of each layer. Four indexes were involved to evaluate the performance of denoising. The results show that the proposed approach provides superior results compared to other denoising method.