Showing papers on "Pixel published in 1970"

Enhanced MRA Images Quality Using Structure Adaptive Noise Filter And Edge Sharpening Methods

Abstract: MR imaging is an emerging and fast growing medical imaging technique which gives high quality images of the soft tissues. There are certain kinds of noise which contaminates these images and thus makes their interpretation difficult for both human and machine. Filtering is a mathematical technique in which intensities of each pixel of the input image are combined with the intensities of its neighboring pixels, to remove the noise and smooth the image. Filtering could be used with MR images for noise removal. The ordinary image filters blur the image and also remove important structural information like lines and edges. This loss of structural information could be dangerous in a clinical environment and could leads to incorrect diagnosis. To address this problem, a structure preserving noise filter is required. When images are processed for human vision, it is also desirable to make them pleasing by sharpening their edges. Such a structure adaptive noise (SAN) filter and an edge sharpening method is designed and implemented on our MRI Visualization toolkit. Our results show that the methods are effective in removing noise, preserving structure, and sharpening edges.

Blind Scheme with Histogram Peak Estimation for Histogram Modication-Based Lossless Information Embedding

Abstract: In this paper, a simple scheme for histogram modification-based lossless information embedding (HM-LIE) is proposed. The proposed scheme is free from memorizing side information, i.e., blind. A HMLIE scheme modifies particular pixel values in an image in order to embed information in it on the basis of its histogram, i.e., tonal distribution. The scheme recovers the original image as well as extracts embedded information from a distorted image conveying embedded information. Most HM-LIE schemes should memorize a set of image-dependent side information per image. The proposed scheme does not have to memorize such information to avoid costly identification of the distorted image carrying embedded information because of the introduction of two mechanisms. One is estimating side information on the basis of a simple statistic, and the other is concealing not only main information but also a part of the side information in the image. These approaches make the proposed scheme superior to the conventional blind schemes in terms of the quality of images conveying embedded information.