CAJ | 학술논문

图像分形编码压缩率大且质量较高，但实时性不好，因此难以推广应用。提出使用GPU对编码进行加速的方案。以图像FW算法为基础，在CUDA规范下，使用GPU并行地对图像中所有的待编码子树同时进行最优父树搜索。实验表明，该文编码方案在保持原有算法图像解码质量的情况下，可将编码时间缩短至毫秒级，满足了实时性的要求。
도상분형편마압축솔대차질량교고，단실시성불호，인차난이추엄응용。제출사용GPU대편마진행가속적방안。이도상FW산법위기출，재CUDA규범하，사용GPU병행지대도상중소유적대편마자수동시진행최우부수수색。실험표명，해문편마방안재보지원유산법도상해마질량적정황하，가장편마시간축단지호초급，만족료실시성적요구。
Fractal image encoding can provide a high compression ratio with a high reconstructed image quality, but it suffers from long encoding time, so it is difficult to promote. A new scheme based on image-fractal ( FW) encoding was proposed. Under the CUDA specification, GPU was used to search the best match parent-tree for each sub-tree parallel. Experiment results proved that the scheme could magnificently reduce encoding time to milliseconds-level with maintaining the reconstructed quality and the requirements of real-time was satisfied.