X-ray bio-image denoising using directional-weighted-mean filtering and block matching approach

An X-ray bio-image might suffer interference from salt-and-pepper (SAP) noise during transmission or capture, thus reducing image quality. This paper proposes a three-stage method to cope with this problem. A directional-weighted-mean (DWM) filter is used to remove the corruption noise in the first stage. In the second stage, extreme pixel (255 or 0 for an 8-bit gray level bio-image) confirmation is performed to restore the X-ray bio-images. In the final stage, block matching identifies blocks with similar textures in a local region. The center pixels of these similar blocks are then averaged to refine the gray value of the restored pixel, thus allowing improvement to the quality of the restored X-ray image through consideration of the texture properties in neighbor pixels over a large size window. Experimental results show that the proposed approach can effectively remove background noise from a SAP noise corrupted bio-image for various noise densities. The reconstructed bio-image does not incur blurring even under heavy noise corruption.