Computer Science · Computer Vision and Pattern Recognition
Wavelet-based deconvolution algorithms applied to ultrasound images
Simona Maggio, Nicola Testoni, Luca De Marchi, N. Speciale, G. Masetti
See on the mapAbstract
Since deconvolution is a recurring theme in a wide variety of signal and image processing applications, many algorithms have been proposed to address this problem. In particular, in ultrasound imaging, deconvolution is often applied as a fundamental step either for contrast enhancement or as preprocessing in segmentation procedures. In this work we present a comparative study between two wavelet-based deconvolution algorithms as tools for processing ultrasound images, one based on a minimization of an error energy term, the other performing a two-step regularization procedure on both the Fourier and Wavelet domain. The comparison is made in terms of Mean Square Error (MSE) and Signal to Noise Ratio (SNR) calculated on synthetic signals. Moreover, we estimate the computational cost and we provide processed B-mode images through which background noise smoothing and edge sharpness enhancement could be qualitatively evaluated.