Search

In this paper, we propose a region based image compression approach for medical applications. We use a vector quantization scheme, combined with regions-of-interest. The image to be compressed is first segmented into regions and a separate codebook is used for each specified region. Codeword size and number of codewords by codebooks may be different according to the diagnostic importance of the corresponding image region. This permits to create appropriate codebooks with representative codewords, and to obtain good reconstruction quality in relevant zones, while reinforcing compression in less important regions. The proposed approach is tested on ultrasound esophagus images and is shown to be very promising.

Vector quantization (VQ) is an effective image compression approach. Among the different existing algorithms, Kohonen's self organizing feature map (SOFM) is one of the well known methods for VQ. It allows efficient codebook design with interesting topological properties. Furthermore, use of VQ for compression gives, in the same process, basic information on the image content. But in order to preserve the diagnostic accuracy in medical applications, the block size is restricted to small values (3/spl times/3, 4/spl times/4), which limits the compression rate. We propose to improve the compression performance by using several codebooks containing codewords of different size, according to the quadtree decomposition of the images. Results are compared to those provided by the standard JPEG image compression algorithm. Finally we explain how it is possible to generate characteristic signature maps of images using compression information. The paper represents an extension of the work presented by G. Cauguel et al. (1997).

This paper investigates the finite state vector quantization signal compression approach. A new coding scheme is proposed which optimizes the performance of the so-called conditional histogram next-state function design. Optimization is performed by determining for every input block the subcodebook size, that minimizes the expected value of the number of bits in the compressed bit-flow. This is done under the constraint to ensure the same reconstruction quality as that of the full-search VQ. Two different state-correction algorithms are studied. The proposed schema for image compression is tested and is shown to give better results than classical FSVQ approaches.

The coding and transmission of medical images over noisy channels are investigated. The proposed coding scheme is based on vector quantization and the codebook is created by Kohonen's neural approach. In conjunction with the ordered codebook, an appropriate QAM (quadrature amplitude modulation) is used for transmission, which results in eliminating a high ratio of transmission errors. Residual erroneous blocks are eliminated by applying a post-processing method on the received image. Based on the visual similarity between each block and its local context, the impulse block noises are detected and corrected. Using the proposed approach, the reconstructed image quality is acceptable even when the channel is extremely noisy.

Image coding using vector quantization is an interesting approach for image compression. Among the different existing algorithms, Kohonen's self-organizing feature maps (SOFMs) are well suited for designing the codebooks because of their specific properties. Moreover, the use of this method for compression gives, in the same process, basic information on the image content, but to preserve the diagnostic accuracy in echoendoscopic images, it is necessary to take small codewords, generally not greater than subimages of 3 by 3 pixels, which limits the compression rate. We propose to improve the compression rate by using four codebooks, with codewords of different size. Image analysis for coding uses a quadtree scheme. Results are compared with those obtained using the standard JPEG image compression algorithm.

The finite state vector quantization (FSVQ) image compression approach is studied. A new compression scheme is proposed which optimizes the performance of the so-called conditional histogram next-state function design. The optimization is performed by determining for every input block the state codebook size, that minimizes the expected value of the number of bits in the compressed data flow. This is done under the constraint to ensure the same reconstruction quality as that of the full-search vector quantizer. The derailment is avoided by transmitting side information to the decoder. The proposed scheme is tested on still medical endoscopic images. Since the system is always adapted to the probability distribution of block occurrences, it provides a lower bit rate than FSVQ that applies state codebooks of fixed size.

International audience; This paper focuses on the conditional histogram (CH) next-state function design used for the finite-state vector quantization (FSVQ) image compression approach. A new coding scheme is proposed which optimizes the performance of CH while ensuring the same reconstruction quality as that of the full-search VQ. The optimization is performed by determining for every input block the subcodebook size that minimizes the expected value of the number of bits in the compressed bit-flow. Two different algorithms are studied in order to ensure the best reconstruction. The proposed scheme is shown to give better results than classical FSVQ approaches. In fact, the proposed approach reveals the relationship between FSVQ and conditional entropy-coded VQ scheme.

International audience; Multispectral remotely sensed images interpretation using fuzzy neural networks