【神经网络在图像编码中的应用】王卫.pdf
Dissertation Submitted to Huazhong UniversityofScienceandTechnologyfortheDegree ofDoctorofPhilosophyinEngineering TheApplicationsofNeuralNetworkstoImageCoding Ph.D.Candidate:WangWei Advisor Prof.WanFaguan Ass.Advisor Prof.CaiDejun March,1994 DepartmentofElectronic.
预测矢量量化编码,设计了神经网络非线性内插器和神经网络矢量预测器,提 出了独到的见解,此外,还提出了一种高压缩比图像编码方案,进一步证明了混 合模型法在图像编码领域应用的潜力.本文的所有算法已做成软件包,可在Windows环境下运行.关键词:神经网络,图像编码,图像数据压缩,模型,学习算法.*国家自然科学基金资助项自,国家模式识别重点实验室资助项目。
Theblockingeffectisreducedsignificantly.InChapter 7,wepresentafastcodebook search algorithm and an algorithmfor edgc-preserved codebook design,which are aimedatimprovingtheedgedegradationandthe time-consumingproblemsduringthe VQ coding process,thenpropose a novelcorrelation vector quantization(CORVQ