Constrained Unsharp Masking for Image Enhancement

Abstract

In this paper we present a cost-effective solution for combined de-noising and sharpening of digital images. Our method combines the unsharp masking and sigma filtering techniques through a regularization mechanism thus ensuring effective noise reduction and edge enhancement in the processed image. We describe our method in detail and we analyze the proposed implementation through extensive experiments done in various scenarios. Due to its low computational complexity the proposed method is well suited for mobile implementations.

Combined de-noising and sharpening of color images in DCT domain

“…As expected the unsharp masking technique increases the noise level while sharpening the edges. The regularized unsharp masking method, from [3], shown lower noise amplification and sharper result compared to the standard unsharp masking. The contrast enhancement technique, from [4], presents also a limited noise amplification while enhancing the contrast and the average image intensity.…”

“…In this section we present experimental results comparing the performance of our proposed method with the unsharp masking technique, regularized unsharp masking technique from [3], the approach from [4] and with the DCT based method from [5].…”

“…Image de-noising and sharpening are techniques that aim at reducing the two above mentioned distortions and many solutions have been proposed in the open literature [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12]. In principle image sharpening is done by amplification of the image high frequencies which boosts also the noise component present in the image.…”

“…Basically, such methods can be classified into two main classes: the pixel domain approaches and the transform domain methods. The solutions operating in the pixel domain directly modify the values of the pixels to increase the contrast while reducing the noise component [3], [11], [12]. The transform domain methods firstly transform the input image (using wavelet decomposition, DCT, DFT etc.)…”

Hierarchy of nonlocal means for preferred automatic sharpness enhancement and tone mapping

“…This is because, after smoothing, noise is still present in the fine level and enhancement of the fine level results in enhancement of noise, thus spoiling the visual quality of images. The performance of [16] is constrained by the choice of the training dataset and [28] uses a very simple sigma filtering method to remove noise. Moreover, unlike [28], we add the noise back to the image to preserve the original properties of the image.…”