A new efficient motion estimation algorithm was proposed in this paper, which pould reduce the influence of noise on the registration accuracy and decrease error-matching rate. In the algorithm, the canny operator was applied to extract the contour of current frame to obtain the feature block. Then the reference frame was converted into binary image. The feature block and the reference frame were matched in a new matching criterion. The experimental results showed that the matching accuracy was improved obviously.%为了减少噪声对配准精度的影响,降低误匹配率,提出了一种新的、有效的块匹配运动估计算法.算法利用Canny算子提取当前帧的轮廓,从而得到特征块,并将参考帧进行2值化处理,在新的匹配准则下,对特征块与参考帧进行配准.实验结果表明,匹配精度有明显提高.
展开▼