简体中文
From Jun. 24 to Jul. 4, 2014, Dunhuang Research Institute and Tianjin University continued their cooperation to conduct relevant researches and testing concerning the "Risk Monitoring and Key Technology Research Assessment for Dunhuang Mogao Grottoes" as the sub-topic of "Key Technology Research of Risk Control and Demonstration Project of World Cultural Heritage" and the national natural science foundation topic namely "Disease Evolution Monitoring Technology Research Ancient Murals Based on Image Analysis".
According to the research mission, the main content of this research and testing includes the following four parts, namely the disease mural relocation technology and equipments, the application of spherical panoramic photography, the application of depth camera in disease analysis and mural disease identification system.
Through research based on the analysis of external images of ancient mural diseases of different types and different periods, we explored quantitative analysis and standards of disease evolution degree and monitoring technology of disease evolution. of ancient murals based on the precise relocation of cameras, image comparison and gap analysis at a relatively microscopic level. In order to ensure a reliable and accurate monitoring and measurement, the research is based on the supporting technology which is characterized by precision acquisition images via extraction, segmentation, comparison of image features and self-feedback relocation . To obtain inspection and appraisal conclusion of disease evolution status of mural disease areas, a overall mural and mural groups, the technology research of comprehensive analysis of disease evolution of mural disease area at a large-scale scope from macro perspective will also be included in this project.
The main content of on-spot work consists of determining the first batch of mural disease monitoring sites, acquiring panoramic images of monitored caves, initial data acquisition of monitoring sites, relocated platform validation for partial monitoring sites, 3D deep information acquisition of monitoring sites via depth cameras and further improving interactive mural disease identification systems and other contents.
Among these works, the working method of determining the disease monitoring site is, by historical data analysis, focusing on the severity changes of four typical diseases including herpes, crisp base, armor-off, cracks, cave distribution (south, north, upper and lower), size, smooth level of wall surface, age, workload and actual demand of the project. After an on-spot investigation over 30 Mogao Grottoes, 11 were selected as the monitoring caves including Cave 25, 70, 85, 87, 92, 132, 276, 351, 356, 437 and 465.
LadyBug panoramic camera is mainly adopted in spherical panoramic photography to conduct panoramic photography test of Cave 85, 351 and 356. LadyBug camera can be used to conduct simultaneous shooting through its multiple cameras and acquire images of all areas except except the bottom of the camera by automatic image stitching. Task group hopes that, through panoramic camera application, the reference coordinate system can be established for all monitoring sites in the cave and integrate monitoring information (location and disease evolution degree) into the panoramic image.
Panoramic Image of Cave 365
Data acquisition of the first monitoring site mainly aims at 11 caves confirmed this time with the guarantee availability and accuracy. The data to be collected includes the number of cave, disease type, image and description of reference point, camera position, overall image of reference point, light source parameters, camera parameters, front illumination image, side illumination image and other contents.
Relocation verification is achieved through monitoring sites within Cave 85, 351 and 356. The platform is precisely acquired with high-definition industrial camera system as the core. Camera three-point prompt relocation approach and four-point prompt relocation approach are adopted for implementation.
Working Site of Relocation Shooting
The main content of using depth camera to collect 3D depth information of monitoring sites is using capture depth camera to conduct deep shooting of more than ten monitoring sites in Cave 351, 356, 132, 85. Through deep shooting, the degree of mural disease can be reflected from previous qualitative analysis such as metering light shooting into quantitative analysis. Thus, mural changes can be measured 0.01mm away from the datum plane.
Shooting Scene with Depth Camera
In studying the mural disease identification system, the research group conducted. precise and artificial disease identification of collected image data for the establishment of typical disease database. This time, mainly targeting incompatible 32-bit operating system last year, a new system versatile for 32-bit and 64-bit operating system has been developed. Meanwhile, the paging output during the print and related staff’s demand have also been collected. The system is intended to provide an efficient, convenient and easy auxiliary disease identification system aiming at practical work requirement including disease identification, status survey, design of repair plan of ancient murals represented by Dunhuang Mogao Grottoes murals. Besides, with machine learning technique as the basis, a semi-automated interactive mural disease identification and diseases status investigation can be realized. The system has been adopted for on-spot testing in Cave 254.
Mural Disease Identification System