基于场记板中数字识别的自动顺片技术

影片素材拍摄完成后,需要根据拍摄进度将每个场景的待用镜头按照剧情顺序重新排列,以便为后续的精细剪接做准备。由于每个镜头的拍摄是从场记板开始;且场记板记录了当前镜头的顺序信息,因此提出一种基于场记板上数字识别的自动顺片技术;该技术首先利用基于场记板信息的两级直方图突变镜头检测算法对视频序列进行镜头分割,在每个镜头中找到含场记板的帧;并从中分割出场记板区域。然后采用霍夫变换几何检测方法,从场记板区域中定位出数字区域。最后基于改进的BP神经网络对手写数字进行识别,从而完成自动顺片。实际应用结果表明,提出的新技术完全自动化、计算精度高、大大提高了工作效率,能够满足实际应用的需求。

An Automatic Assembling Technology Based on The Digit Recognition in The Clap-stick

After shooting the video material, in order to make it a structured and coherent film, we need to edit it. Because each shot often starts with a clap-stick, and the order information of the current shot is written on the clap- stick digitally, so this paper presents an automatic assembling technology based on the digit recognition. Firstly, the proposed method applies the two-level color histogram method based on the clap-stick information to segment shots from video sequence, searches the image frames that contain a clap-stick in each shot, and splits the clap-stick area. Then, the regions of the clap-stick are located by the Hough transform. The proposed method completes the digit recognition based on improved BP neural network, and finally sorts all the shots. Practical application results show that the proposed method in this paper is fully automated and high accuracy. It improves work efficiency, and can be widely applied to the field of film editing.