Study on the complete extraction method of ancient ceramic texture features

Abstract

To address the problems such as background noise and uneven illumination during the extraction of ancient ceramic patterns. A nondestructive extraction method for missing patterns in ancient ceramics is proposed. Combined with the characteristics of the pattern, the median filter matrix equation is designed to denoise the image. By defining the median filter equation function for the high-frequency approximation coefficient of the wavelet transform, the details of the ancient ceramic pattern are clearer. The double histogram equalization algorithm is built to maintain the average brightness of the image, which is helpful to the extraction and analysis. The verification shows that the method can effectively eliminate the interference of noise on the extraction of the ancient ceramic pattern, and the accuracy of the segmentation reaches 94%. It effectively promotes the restoration and protection of ancient ceramic cultural relics.