动态纹理是计算机视觉中的动态模型之一,在空间范围内具有统计平稳性,在时间维度上具有随机重复性。动态纹理合成的目标是生成与给定纹理在视觉上相似的图像。在进行动态纹理合成时,回归预测误差积累是导致纹理质量下降的一个关键问题...动态纹理是计算机视觉中的动态模型之一,在空间范围内具有统计平稳性,在时间维度上具有随机重复性。动态纹理合成的目标是生成与给定纹理在视觉上相似的图像。在进行动态纹理合成时,回归预测误差积累是导致纹理质量下降的一个关键问题。为此,本文提出一种基于自纠正机制的动态纹理合成模型。利用清晰度、结构相似性、光流等指标来确定优化数据范围,并找到优化极值点。通过自纠正机制,将原始数据替换为优化数据,并将优化数据用于回归预测。最后,利用卷积自编码器将预测数据重构为高维的动态纹理视频帧。在DynTex数据库上进行实验,并与几种典型的动态纹理合成模型进行比较。实验结果表明,用该模型合成的动态纹理视频帧与真实视频帧计算得到的MSE(Mean Square Error)数值更小,PSNR(Peak Signal to Noise Ratio)和SSIM(Structural SIMilarity)的数值更大。它解决了动态纹理合成中出现的残影、模糊、噪声等问题,从而能够生成视觉效果更好并且更长的动态纹理。同时,验证了所提出的建模方法的有效性。展开更多
It is a regular way of constructing quantum error-correcting codes via codes with self-orthogonal property, and whether a classical Bose-Chaudhuri-Hocquenghem (BCH) code is self-orthogonal can be determined by its des...It is a regular way of constructing quantum error-correcting codes via codes with self-orthogonal property, and whether a classical Bose-Chaudhuri-Hocquenghem (BCH) code is self-orthogonal can be determined by its designed distance. In this paper, we give the sufficient and necessary condition for arbitrary classical BCH codes with self-orthogonal property through algorithms. We also give a better upper bound of the designed distance of a classical narrow-sense BCH code which contains its Euclidean dual. Besides these, we also give one algorithm to compute the dimension of these codes. The complexity of all algorithms is analyzed. Then the results can be applied to construct a series of quantum BCH codes via the famous CSS constructions.展开更多
文摘动态纹理是计算机视觉中的动态模型之一,在空间范围内具有统计平稳性,在时间维度上具有随机重复性。动态纹理合成的目标是生成与给定纹理在视觉上相似的图像。在进行动态纹理合成时,回归预测误差积累是导致纹理质量下降的一个关键问题。为此,本文提出一种基于自纠正机制的动态纹理合成模型。利用清晰度、结构相似性、光流等指标来确定优化数据范围,并找到优化极值点。通过自纠正机制,将原始数据替换为优化数据,并将优化数据用于回归预测。最后,利用卷积自编码器将预测数据重构为高维的动态纹理视频帧。在DynTex数据库上进行实验,并与几种典型的动态纹理合成模型进行比较。实验结果表明,用该模型合成的动态纹理视频帧与真实视频帧计算得到的MSE(Mean Square Error)数值更小,PSNR(Peak Signal to Noise Ratio)和SSIM(Structural SIMilarity)的数值更大。它解决了动态纹理合成中出现的残影、模糊、噪声等问题,从而能够生成视觉效果更好并且更长的动态纹理。同时,验证了所提出的建模方法的有效性。
基金Supported by the National Natural Science Foundation of China (No.60403004)the Outstanding Youth Foundation of China (No.0612000500)
文摘It is a regular way of constructing quantum error-correcting codes via codes with self-orthogonal property, and whether a classical Bose-Chaudhuri-Hocquenghem (BCH) code is self-orthogonal can be determined by its designed distance. In this paper, we give the sufficient and necessary condition for arbitrary classical BCH codes with self-orthogonal property through algorithms. We also give a better upper bound of the designed distance of a classical narrow-sense BCH code which contains its Euclidean dual. Besides these, we also give one algorithm to compute the dimension of these codes. The complexity of all algorithms is analyzed. Then the results can be applied to construct a series of quantum BCH codes via the famous CSS constructions.