摘要
针对建筑物三维模型缺乏强烈的真实感,同时目前对纹理的许多处理工作是依赖于人工这一问题,该文结合建筑物点云数据,对基于手持数码相机拍摄像片的建筑物立面纹理优化方法进行了研究。采用单位四元数法解算大倾角像片外方位元素,解算结果满足纹理精度要求;结合纹理面积、倾斜角度以及树木遮挡3项指标,实现从大量像片中自动选取最优像片;依据色调信息识别出树木遮挡区域并对该区域进行了很好地补偿;在近景影像中引入蚁群算法优化拼接线,实现了多条带多张纹理的自动拼接。最后以手持数码相机拍摄的像片进行实验,实验证明:本文方法可以自动有效地提高纹理质量,得到的立面纹理目视效果真实美观。
According to the fact that three-dimensional models of buildings generally lack a strong sense of reality.At the same time,the many processing on the texture is dependent on artificial,this paper studies the texture optimization method of building facades based on photographs by handheld digital camera and buildings' point cloud data.Using unit quaternion method to solve exterior orientation elements of large angle images,calculation results can meet the precision requirements of the texture;Taking the texture area,tilt angle and tree occlusion into consideration to automatically select the best image from a large number of images;recognizing trees occlusion areas by hue information and the areas get good compensations;the paper utilizes ant colony algorithm in close range images to optimize seam line and automatically mosaic multi textures of multi bands.Experiments are carried out on photographs by handheld digital camera,the result shows that the method in this paper can automatically and effectively improve the textures quality,and the facade texturesare real and beautiful.
出处
《测绘科学》
CSCD
北大核心
2017年第1期164-170,176,共8页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41471360)
关键词
建筑物三维模型
数码相机
建筑物点云
建筑立面
纹理优化
three-dimensional models of buildings
digital camera
buildings' point cloud
building facades
texture optimization