A Clustering and Segmentation Method of Stained Brush Text Images on Ancient Building Wall Inscriptions

Abstract

The invention discloses a method for clustering and segmenting images of dirty brush characters in inscriptions on walls of ancient buildings, belonging to the field of digital restoration of ancient buildings. In this method, firstly, a partial differential model is constructed to denoise the acquired image, and block enhancement is performed according to the illumination characteristics of the inscription image; secondly, the enhanced image is segmented by the method of maximum between-class variance, and morphological processing is performed; then the processing After the image is regionalized, the minimum circumscribed rectangle of the text area is obtained, and the corresponding text area is marked in the enhanced image; finally, the first FCM clustering is performed on the text area to determine the cluster center matrix, and the average gray similarity and The distance penalty function constrains the degree of membership, and after NKFCM clustering and deblurring processing, the final clustering and segmentation image is obtained. This method can not only effectively eliminate the influence of noise on clustering, but also maintain the integrity of segmentation, and extract high-quality inscription text. The invention is mainly used for the clustering and segmentation of the dirty brush characters in the wall inscriptions of ancient buildings.

Description

technical field [0001] The invention belongs to the field of digital restoration of ancient buildings, and specifically relates to a clustering and segmentation method for image clustering and segmentation of dirty brush characters on ancient wall inscriptions, which can not only filter out the noise caused by pollution, but also ensure the complete segmentation of the main body of characters. Background technique [0002] The brush writing on the wall inscriptions of ancient buildings carries a large amount of historical information and is of great value in calligraphy and historical research. However, due to natural weathering and man-made damage, and some inks are decomposed and processed from animal and plant remains, they are prone to mildew, and the inscriptions have been polluted to varying degrees, and some are even difficult to identify. Traditional physical decontamination and restoration are easy to remove while removing pollution. Causes damage to the text body a...

AI Technical Summary

Problems solved by technology

The threshold-based clustering algorithm has a strong dependence on the selection of the threshold, and the selection of the threshold is greatly affected by noise, and the robustness of the algorithm is poor; the model-based clustering algorithm can make full use of the characteristic information of the text, but the training model is limited The amount of sample data required is large, and the prior sample requirements are high, which is not suitable for small sample ancient text clustering; partition-based clustering algorithms, such as the k-means algorithm, can better eliminate differential noise for text segmentation However, the noise filtering effect of blocky pollution with similar gray levels is not good, and the segmented text body is incomplete.

The above method has a better clustering effect on inscription images with less pollution in the text area, but when the pollution seriously affects the main body of the text, the clustered text may have defects or contain more noise

Images