Digital Signal Processing Reference
In-Depth Information
Figure 5.19. (a) Original “Barbara” image and (b) its noisy version with noise
standard deviation
σ
= 20. (c) Denoised image by the DCTG2 (PSNR =
28.2 dB). (d) Denoised image by the UWT (PSNR = 26.5 dB).
5.7 SUMMARY
In this chapter, we have gone beyond point symmetric or other compact features
to model alignments and curved shapes in images. The ridgelet and curvelet trans-
forms are used for this. We have shown curvelet efficiency in a range of case studies.
Considerable attention has been paid to fast and practical aspects and algorithms.
Concerning the contrast enhancement, our conclusions are as follows:
1. The curvelet enhancement function takes very good account of the image
geometry to enhance the edges, while avoiding noise amplification.
2. As evidenced by the experiments with the curvelet transform, there is better
detection of noisy contours than with other methods.
