Improved region growing algorithm for the calibration of flaking deterioration in ancient temple murals

Qi Zhang,Hongyan Cui,Jianfang Cao,Yanfei Li

doi:10.1186/s40494-018-0235-9

Qi Zhang, Hongyan Cui + Show 2 more

Open Access

https://doi.org/10.1186/s40494-018-0235-9

Copy DOI

Abstract

Ancient murals are vulnerable to varying degrees of damage due to long-term exposure to external environmental factors such as light, temperature and humidity. Enabling people to appreciate the original features of murals has become a concern for field experts. The development of computer technology makes it possible to use intelligent information processing to simulate and restore ancient murals. This paper proposes an improved region growing algorithm based on threshold segmentation to automatically calibrate the flaking-related deterioration of murals in response to erosion by taking temple murals from the Song Dynasty in Kaihua Temple as the study object. First, we analyze the color characteristics of the flaking area, mark the suspected flaking-damaged points by threshold segmentation, and use these points as seeds for the area growth and expansion of the flaking area. We then calculate the color mask. Next, in the YCbCr and HSV color spaces, the brightness, chroma, and saturation characteristics of the flaking area are analyzed. The masks for the brightness, chroma, and saturation of the flaking area are obtained by threshold segmentation, and all the feature masks are merged. Finally, the mask of the flaking area obtained by data fusion is added to the original image to calibrate the flaking deterioration. Compared with current calibration algorithms based on multiscale mural deterioration, the experimental results show that the average error and error standard deviation of the proposed calibration algorithm are 1.91 and 1.82, respectively, without noise and 1.97 and 1.85, respectively, with noise. The errors are reduced, and the calibration performance is improved and stable. This work provides a good foundation for the virtual and practical restoration of ancient murals.

Highlights

Ancient mural art represents an important cultural heritage and includes a substantial amount of cultural, historical, and artistic information that vividly documents the social features of various ethnic groups and periods and has important historical, scientific, and artistic values
To protect the abovementioned ancient temple murals, this paper proposes the region growing algorithm fused with threshold segmentation (TS-RG), which automatically calibrates the flaking area in temple murals based on the mural of Kaihua Temple from the Song Dynasty
The experiment was conducted on a mural image database consisting of 100 images with a resolution of 2600 × 2600 (30 pieces of murals with flaking deterioration in the paint layer and 70 pieces of murals with flaking deterioration in the earth layer)

Summary

Introduction

Ancient mural art represents an important cultural heritage and includes a substantial amount of cultural, historical, and artistic information that vividly documents the social features of various ethnic groups and periods and has important historical, scientific, and artistic values. Li et al [8] investigated the effectiveness of reinforcing mural paintings at the Mogao Grottoes in Dunhuang, China, by conducting on-site inspections to determine if new murals that had been previously repaired presented new forms of deterioration These studies focused on ancient murals from the perspectives of aesthetics, biology, and chemistry, proposed the need for mural protection and laid the foundation for the protection of ancient murals using intelligent information processing technologies. For the automatic calibration of murals, Wang et al [15] extracted crack information using a top-hat transformation from the images of Tang dynasty mural paintings to realize automatic recognition and calibration This method relied heavily on the selection of structural elements and experienced an overcalibration problem. Scholars studying the intelligent processing of ancient murals are still in the beginning and exploration stages of this field

Methods

Results

Conclusion