复杂场景下动画图像轮廓特征匹配拼接仿真  

Simulation of Animated Image Contour Feature Matching Correction in Complex Scene

作　　者：燕耀[1] 雷桐[1] 赵曦 YAN Yao;LEI Tong;ZHAO Xi(Xi'an Polytechnic University,Xi'an Shanxi 710048,China)

机构地区：[1]西安工程大学,陕西西安710048

出　　处：《计算机仿真》2020年第8期202-205,218,共5页Computer Simulation

基　　金：教育部人文社会科学研究青年基金项目(18YJC760110);陕西省教育厅研究计划项目(18JK0331)。

摘　　要：采用传统的方法对动画图像轮廓特征进行匹配拼接时,存在特征匹配准确率低、拼接效率不高和拼接后图像失真严重等问题。针对上述问题,提出一种基于复杂场景下动画图像轮廓特征的匹配拼接方法。方法将复杂场景下动画图像进行适当缩放并采用高斯低通滤波器进行降噪,利用双阈值方法检测平滑图像的轮廓,提取轮廓特征并对其构建匹配关系。选出最佳匹配对进行图像轮廓拼接。仿真结果证明,其较传统拼接方法特征匹配准确率更高、运行方法所用时间更短,具有较强的抗失真能力,为提高动画制造工作效率和提升产量有很大的帮助。In the traditional method,the visual effects after image correction cannot reach the ideal result and the correction completion time is too long.In this article,a method to correct asymmetric image distortion based on bilinear interpolation was proposed.By analyzing the distortion of asymmetric image,we used the phase difference theory to build the asymmetric image distortion correction model,so that the model could be effectively combined with the optical imaging law.Then,we calculated the initial distortion coefficient of asymmetric image.After that,we got the best asymmetric coefficient by optimizing the objective function.Finally,we reconstructed the asymmetric distortion image by bilinear interpolation method.Thus,we achieved the uniformity correction of asymmetric image distortion.Simulation results prove that the proposed method has better visual effect,less root mean square error and shorter correction time.

关 键 词：动画图像 轮廓特征 图像轮廓匹配 图像拼接 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]