A Study on High-Quality Layred Transmission of Images on the Basis of Multiresolution Representations

这项研究的一个更广泛的兴趣是针对图像的移动电信。研究的目的是开发一种高质量的图像高速分层传输系统。基于小波的多分辨率表示为图像表示提供了平台，并认为并行组合扩频技术是一种很有前途的无线通信系统。主要成果总结如下四点。/1/为了在图像质量和空间分辨率上提供广泛的可扩展性，研究了一种分层图像编码方案，该编码方案包括小波直接积展开、带死区非线性量化、多级零游程编码和算术编码。/2/针对移动通信中的高质量图像传输，将并行组合扩频系统与纠错解码相结合进行了研究，在典型的多径衰落信道中，其误码率性能得到了显著提高。/3/在实际的无线通信环境中，通过在分层编码数据流和附加纠错码之间进行可协商的权衡，解码后的图像质量得到了很大的改善。/4/作为接收机的一种选择，开发了一种新的图像增强方法。它是一种基于矢量量化码本中描述的统计知识的局部特征估计，该码本由基于小波的多分辨率空间中跨尺度的边缘和轮廓的典型行为组成。

A broader interest in this research is targeted at the mobile telecommunications for imagery. The purpose of study is to develop a quality high-speed layred transmission system for images. A platform for image representations is provided by the wavelet-based multiresolution representation, and the parallel combinatorial spread apectrum technology is assumed to be a promising wireless telecommunication system. Major results are summarized in four points as follows./1/A layred image coding scheme is studied to provide wide-range scalability with respect to both image quality and spatial resolution, and the coding scheme comprises a direct-product wavelet expansion, nonlinear quantization with dead bands, multi-stage zero runlength coding and arithmetic coding./2/For quality image transmission in mobile telecommunications, parallel combinatorial spread apectrum systems are investigated in conjunction with error correcting decoding, and also their bit error rate performances have been found to be significantly improved in typical multipath fading channels./3/Decoded image quality has been very improved in realistic wireless telecommunication environments by negotiable trade-off between a layred encoded data stream and additional error correcting codes./4/As an option at receivers, a new image enhancement method has been developed. It is a local feature-based estimation based on statistical knowledge described in a vector quantization code book, and the code book consists of typical behaviors of edges and outlines across scales in a wavelet-based multiresolution space.