Search

A distortion measure based on conditional differential entropy is presented for image coding. Findings reveal the suitability of the proposed measure for correct representation of the amount of information lost by quantization.

A multiclass image segmentation method developed by using Markov Gauss mixture model (HMGMMs) which could provide better classification of segmentation of aerial images and textures is presented.

A general coding strategy is presented for image compression applications in which the source statistics vary over time or space, or are not known at design time. The strategy is based on a structure comprised of two stages, using a family of codes to replace the conventional single source code.

Two techniques for the reconstruction of lost subband coefficients are presented. These methods can be used for the reconstruction problem that emerges at the decoder during transmission of digital subband-coded images over lossy packet networks. One of the approaches is for low-frequency coefficients while the other is for high-frequency coefficients. The former is based on the hierarchical subband decomposition's inherent properties while the latter involves linear interpolation.

The introduction of a tree-structured posterior estimator into the synergistic Bayes vector quantization (VQ) system improves classification with little change in the compression of two different image sources. The performance of many VQ-based algorithms used to get compression and classification of image sources is analyzed. The best method is a nonparametric method that puts a Bayes risk component in the distortion measure.

Vector quantization (VQ) and subband decomposition provide an effective means of image compression. VQ helps in the conversion of the decomposed signal into bits. The conversion takes advantage of the flexible partitions of the higher dimensional vector spaces, and the inter and intraband correlation. A survey of the methods used since 1988 to study VQ applications to subband/wavelet transform coefficients is presented.

An algorithm for segmenting document images into four classes, is presented.

A Joint Photographic Experts Group (JPEG)-compliant perceptual coding system for gray-scale images that are to be compressed before they are scaled, halftoned and printed is presented. Experimental results revealed that the coding system maintains a near visually lossless image quality at rates under 0.5 b/pixel when used with bilinear interpolation and either ordered dithering or error diffusion. Moreover, it was found to be at least 20% better than a JPEG encoder using suggested JPEG quantization and Huffman tables.

Noise is ubiquitous in real life and changes image acquisition, communication, and processing characteristics in an uncontrolled manner. Gaussian noise and Salt and Pepper noise, in particular, are prevalent in noisy communication channels, camera and scanner sensors, and medical MRI images. It is not unusual for highly sophisticated image processing algorithms developed for clean images to malfunction when used on noisy images. For example, hidden Markov Gauss mixture models (HMGMM) have been shown to perform well in image segmentation applications, but they are quite sensitive to image noise. We propose a modified HMGMM procedure specifically designed to improve performance in the presence of noise. The key feature of the proposed procedure is the adjustment of covariance matrices in Gauss mixture vector quantizer codebooks to minimize an overall minimum discrimination information distortion (MDI). In adjusting covariance matrices, we expand or shrink their elements based on the noisy image. While most results reported in the literature assume a particular noise type, we propose a framework without assuming particular noise characteristics. Without denoising the corrupted source, we apply our method directly to the segmentation of noisy sources. We apply the proposed procedure to the segmentation of aerial images with Salt and Pepper noise and with independent Gaussian noise, and we compare our results with those of the median filter restoration method and the blind deconvolution-based method, respectively. We show that our procedure has better performance than image restoration-based techniques and closely matches to the performance of HMGMM for clean images in terms of both visual segmentation results and error rate. [ABSTRACT FROM AUTHOR]

Presents information on a study which developed an algorithm for segmenting documents into background, photograph, text and graph. Classification features; Formulation of the algorithm; Results and conclusions.