Incorporating fractional calculus in echo-cardiographic image denoising

Speckle noise is an inherent property of echo-cardiographic images. It badly affects the diagnostic value of the image. This research is focused on mitigating speckle noise from echo-cardiographic images. Proposed denoising filter is based on fractional calculus, which is an emerging research topic in the field of image processing. The proposed methodology first divides pixels of noisy image into various regions i.e. homogeneous, texture and edge using a region classifier. This region classifier works on the basis of large eigenvalue of hessian matrix. Fractional calculus based filter mask is applied on each pixel of noisy image. Order of this mask is selected adaptively according to the membership of current pixel with image region. Simulations are carried out on real echo-cardiographic and standard images. Results of the proposed technique outperforms other recently proposed methods when compared on qualitative and quantitative basis using metrics, peak signal to noise ratio (PSNR), correlation coefficient (r), mean square error (MSE), edge preservation index and structural similarity (SSIM).