基于特征加权融合算法的动漫人物图像复原重建方法  

A restoration and reconstruction method for anime character images based on feature weighted fusion algorithm

作　　者：黄君君[1] HUANG Jun-jun(Information Engineering College,Fujian Vocational College of Agriculture,Fujian Fuzhou 350000,China)

机构地区：[1]福建农业职业技术学院信息工程学院,福建福州350000

出　　处：《齐齐哈尔大学学报（自然科学版）》2024年第6期30-36,共7页Journal of Qiqihar University(Natural Science Edition)

基　　金：教育部中国高校产学研基金项目“基于Kinect的三维重建技术在艺术设计类课程中的应用”(2023KY077)。

摘　　要：动漫人物复原图像存在像素缺失会导致所设计人物出现失真显示的情况,从而使主机元件无法准确重建动漫人物模型。针对此问题,提出基于特征加权融合算法的动漫人物图像复原重建方法。利用特征加权融合算法跟踪动漫人物特征,将图像分成多个像素区域,根据像素损失值推导人物图像的变分损失复原函数。利用跟踪所得的人物特征,定义重建节点,再将各个节点对象连接起来,以求解具体的重建表达式,完成动漫人物图像复原重建。实验结果表明,应用构建模型可以保证复原图像中缺失像素与原像素的色彩一致性,色彩一致性高于92%,不会造成设计人物失真显示的情况,能够保障主机元件对动漫人物模型的准确重建。The pixel loss in the restored image of anime characters can lead to distorted display of the designed characters,making it difficult for the host components to accurately reconstruct the anime character model.To solve the above problems,a feature weighted fusion algorithm based on animation character image restoration and reconstruction method is proposed.After the feature weighted fusion algorithm is used to track the animation characters,the image is divided into multiple pixel regions,and the variational loss restoration function is derived according to the pixel loss value.Using the tracked character features,define reconstruction nodes,and then connect each node object to solve the specific reconstruction expression to complete the restoration and reconstruction of anime character images.The experimental results show that the application of the constructed model can ensure the color consistency between the missing pixels and the original pixels in the restored images,and the color consistency is higher than 92%,which will not cause the distorted display of the design characters,and can ensure the accurate reconstruction of the animation character model by the host components.

关 键 词：特征加权融合算法 动漫人物图像 图像复原 图像重建 人物模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]