基于MIP图像拼接系统的大型古生物化石数字化应用研究

LARGE FOSSIL DIGITALIZATION BASED ON THE MOSAIC OF IMAGE PROGRAM(MIP)

在大型古生物化石数字化过程中,为了充分展示化石的细节信息,往往需要拍摄大量的图像。为了实现大型古生物化石数字化数据的完整性,需要对这些大量的图像进行精密的图像拼接处理。基于这种应用需求的前提下,本文在自主研发的Mosaic of Image Program(MIP)图像拼接系统的基础上,对高精度的相机检校、畸变检校及改正和拼接缝的保真处理等方面进行研究,形成系统的古生物化石彩色合成影像数字化流程。在宜州化石馆的实际处理中,完成了杨氏锦州龙、蜥脚类恐龙、孔子鸟等大型古生物化石的数字化,几何失真小于0.36mm(畸变矫正精度优于1像元,拼接精度优于2像元,像片分辨率0.12mm)。同时采用基于SIFT的自动辐射归一化处理算法对拼接影像进行辐射均衡处理,矫正拼接影像辐射亮度的不均衡。

In fossil-digitalization process,a multitude of digital images were captured for exhibiting adequate details of fossils.In order to guarantee the integrity of digital data,high-accurate image stitching needs to be implemented with those images.Thus,for meeting the requirement of large fossil digitalization,we proposed a systematic framework based on the foundation of selfdeveloped Mosaic of Image Program（MIP）,including collected camera calibration,distortion correction and image stitching process etc.To validate the correctness and feasibility of our study,the proposed framework was utilized in the digitalization project of Yizhou Fossil Museum.Digitalization data of Jinzhousaurus yangi,sauropod dinosaurs,Dromaeosaurus,Confuciusornis and other large fossils were obtained,and the integrated geometric distortion is less than0.36mm（the accuracy of distortion correction and stitching is better than 1and 2pixel,respectively,with the image resolution of 0.12 mm）. With respect to radiometric differences between adjacent images, we employed automatic relative radiometric normalization method with SIFT algorithm to effectively eliminate those radiometric differences.The results of the applied project demonstrate that our framework can address our focusing problems effectively.