期刊信息
- 刊名： 河北师范大学学报（自然科学版）Journal of Hebei Normal University (Natural Science)
- 主办： 河北师范大学
- ISSN： 1000-5854
- CN： 13-1061/N
- 中国科技核心期刊
- 中国期刊方阵入选期刊
- 中国高校优秀科技期刊
- 华北优秀期刊
- 河北省优秀科技期刊
Optical Flow Computation for Regional Matching Based on Normalization
摘要/Abstract
提出了对传统区域匹配计算光流场的改进方法,通过归一化匹配误差和强化光流场的一致性2个方面改善了对序列图像的光流计算.归一化图像区域匹配计算能够强化图像区域分析能力,它充分考虑到了输入图像存在干扰和亮度过低时的情况;通过强化光流场一致性特性,建立了一种可信度评测标准,将光流信息由高可信度区域向低可信度区域扩展,而不是根据距离的远近而统一进行衰减而扩展.实验表明,该方法具有计算简单,能够满足实时性计算,对运动图像的变化具有较快的反应能力.
The method is presented here to improved way of optical flow computing for traditional regional matching.The computing of the optical flow of image sequences through normalizing the matching error and strengthening optical flow consistency is improved.Normalized computing of image regional matching can strengthen the ability of image regional analysis and fully consider the situation that the input image has been interfered and its brightness is rather low.The evaluating standard of reliability was established by strengthening optical flow consistency.The optical flow information was extended from the dist rict of the high reliability to the dist rict of the low reliability, rather than the past method that it was extended by uniformly attenuating according to distance.The method has the advantages such as simple computation,the adaptability for the rea-ltime computation,as well as the quickly response faculty for the change of motive images.