Abstract:

Plant animation synthesis is a research hotspot nowadays. Extraction and reconstruction of plant motion information are basic and key works in computer to realize efficient and realistic plant animation synthesis. Taking maize as research object, video sequence of real plant motion  was captured and converted into image sequences and then several key frames were selected. By studying digital image processing algorithms, image foreground object was extract,  and a clear and complete binary skeleton image of plant was get through thinning and deburring the target. Based on the plant skeleton images, an approach to extract plant animation feature points data were proposed. Finally the extracted plant animation feature points were mapped to 3 dimensional plant models and data-driven plant animation synthesis was  realized. These experimental results showed the validity and effectiveness of this method.

Key words: plant animation, video, image sequence, image thinning, feature extraction

摘要：

植物动画合成是当今的一项研究热点，运动信息的提取与重构是在计算机上实现高效、逼真植物动画合成的基础和关键。以玉米为研究对象，采集真实的植物运动视频，将视频转换为图像序列并选取若干关键帧。研究数字图像处理算法，提取图像前景目标，并对目标细化和毛刺去除得到清晰、完整的二值植物骨架图像。基于植物骨架图像提出了植物动画特征点数据提取方法，并将特征点数据映射到三维植物模型，实现了数据驱动的植物动画。

关键词: 植物动画；视频、图像序列；图像细化；特征提取