|
|
|
 |
|
|
|
|
|
|
莫高窟壁畫顏料多光譜數字化識別系統的研發與應用=Research and Application of the Multispectral Digital System for Identifying the Pigments of Mogao Murals |
|
|
|
| Author |
柴勃隆 (著)=Chai, Bo-long (au.)
;
肖冬瑞 (著)=Xiao, Dong-rui (au.)
;
蘇伯民 (著)=Su, Bo-min (au.)
;
馮偉 (著)=Feng, Wei (au.)
;
于宗仁 (著)=Yu, Zong-ren (au.)
|
|---|
| Source |
敦煌研究=Dunhuang Research
|
|---|
| Volume | n.3 (總號=n.169) |
|---|
| Date | 2018 |
|---|
| Pages | 123 - 130 |
|---|
| Publisher | 敦煌研究編輯部 |
|---|
| Publisher Url |
http://www.dha.ac.cn/
|
|---|
| Location | 蘭州, 中國 [Lanzhou, China] |
|---|
| Content type | 期刊論文=Journal Article |
|---|
| Language | 中文=Chinese |
|---|
| Note | 作者單位：國家古代壁畫保護重點科研基地、敦煌研究院保護研究所、天津大學計算機學院、敦煌研究院國家古代壁畫與土遺址保護工程技術研究中心 |
|---|
| Keyword | 多光譜=Multispectral; 壁畫顏料檢測=mural pigment detection; 色彩空間=color space; 色彩相似度=color similarity |
|---|
| Abstract | 多光譜攝影無損調查技術已成熟應用於對漫漶壁畫內容的再現提取,調查壁畫中顏料類別的使用及分佈。在以往的研究中對已知的24種莫高窟壁畫繪畫材料做了多光譜圖像獲取,並經過科學篩選、驗證,建立了"莫高窟壁畫顏料多光譜圖像色彩標準數據庫",專業人員可通過相同光譜條件下對現實壁畫獲取光譜圖像,並與顏料多光譜圖像標準數據庫對比分析,可初步獲悉莫高窟壁畫中顏料的類別及應用分佈狀態。但是人眼對色彩的認知存在個體差異,導致人工比對存在主觀不一致、識別難、精度低等問題,且不利於非專業人員分析研究。本研究採用多光譜攝影系統,以"顏料多光譜圖像色彩標準數據庫"[1]為基礎,開發顏料多光譜圖像數字化識別系統。目的利用軟件算法對壁畫多光譜圖像中的目標顏料通過RGB色彩空間模型向HSV色彩空間模型數值的轉換,自動查詢、匹配顏料多光譜圖像數據庫中不同顏料HSV色彩數值的最近相似度[2 ],以達到自動識別顏料類別的目標,解決人工比對中存在的不足,提高鑒別的精度和效率,從而提升分析的科學性。
Multispectral photography investigation technology has previously been used to reproduce the contents of obscure murals and to analyze the use and distribution of the pigments in these murals. As for the Mogao cave murals, a Database of the Mogao Cave Mural Pigments Multispectral Photography Image Standards has been established by a rigorous system of selection and verification from the captured multispectral images of the 24 known types of painting materials established by previous research results. Based on comparative terms, one can obtain the class and distribution status of pigments of Maogao grotto murals through analysis conducted by trained professionals comparing and analyzing the images of the murals captured under spectral conditions aligned with the pigment spectrum from the database. Individual differences in the perception of colors, however, is likely to lead to subjective inconsistencies, recognition difficulty, low precision etc., and is not a process untrained individuals can complete to satisfaction. This study adopts the multispectral photography system to develop a multispectral digital image identification system for mural pigments based on the color data obtained from multispectral images. The process begins by selecting RGB values of the target pigment from obtained multispectral images of wall paintings and converting them to HSV color space. By automatic querying, the most numerically similar HSV color values are then matched with pigment values from the multispectral image database, thus providing a scientific identification of pigment category to a high degree of accuracy and efficiency that avoids the problems related to subjective manual comparison. |
|---|
| Table of contents | 1 多光譜圖像在壁畫保護研究中的應用背景 124
2 多光譜圖像系統簡介 124
3 軟件系統功能設計與實現 125
3.1 色彩空間模型的選取與轉換 125
3.2 軟件算法及代碼實現 126
3.3 軟件運行流程 126
3.4 算法實現: 126
4 軟件功能描述與效果測試 126
4.1 模擬測試 126
4.2 現場壁畫多光譜圖像識別測試 128
5 結論 130 |
|---|
| ISSN | 10004106 (P) |
|---|
| Hits | 222 |
|---|
| Created date | 2020.10.26 |
|---|
| Modified date | 2020.10.28 |
|---|
 |
Best viewed with Chrome, Firefox, Safari(Mac) but not supported IE
|
|
|
