Detection of Cracked Digitized Paintings and Manuscripts Based on Threshold Techniques

摘要

Ancient paintings are cultural heritages that require extensive and intricate preservation undertakings. With the passage of time, paintings can be damaged, and common deteriorations found in the ancient paintings include cracking. Cracks can be caused by many factors, such as aging, drying, and mechanical factors. Fortunately, digital images of ancient paintings can be restored using image processing techniques. This work involved with the development of an automatic crack detection system. First, the binarization technique was applied to the digital images that allowed for the extraction of cracks from the background of these images, based on the thresholding segmentation technique after using the top-hat transformation technique during pre-processing. Thin dark patterns that were misidentified as cracks were then automatically removed. This was then followed by the crack filling via order statistic algorithm. The main contribution of this paper is a simple and robust monochrome method that can be used to restore digital images of historical paintings and manuscripts.