软件导刊2025，Vol.24Issue(5)：207-213,7.DOI:10.11907/rjdk.241303
面向直接体绘制图像的像素级色彩迁移
Pixel-level Color Transfer for Direct Volume Rendering Image
摘要
Abstract
Direct volume rendering images rely on transfer functions to map optical properties to pixels for imaging.To change the visual effect of direct volume rendering images,the transfer function must be continuously adjusted,but this process is extremely cumbersome.Therefore,a pixel level color transfer method is proposed,which not only completes the color style transfer of direct volume rendering images,but also has good image structure protection.Specifically,it is necessary to perform pixel matching in the image features extracted by convolutional neural networks,edit high-level information from the image to low-level information,establish pixel mapping relationships between two imag-es,and propose a new color modification formula to achieve a balance between transfer effects and structural protection.The experiment shows that the proposed method is superior to mainstream methods of the same type in terms of color reproduction and image structure protection.In the transfer experiment of Bonsai volume data,a gray histogram overlap of 0.84 was achieved,which is much higher than the 0.21 and 0.28 of the comparative methods;In the Lobster body data with a smaller gap,it also reached 0.70,still ahead of the comparison methods of 0.47 and 0.59,which is more efficient compared to manually adjusting the transfer function.关键词
直接体绘制/色彩迁移/图像特征/灰度直方图/像素匹配Key words
direct volume rendering/color transfer/image feature/gray histogram/pixel matching分类
计算机与自动化引用本文复制引用
李维杰,潘家辉..面向直接体绘制图像的像素级色彩迁移[J].软件导刊,2025,24(5):207-213,7.基金项目
广东省普通高校重点领域专项项目(2023ZDZX2021) （2023ZDZX2021）
