Virtual restoration method for cultural relics

Abstract

The invention relates to a cultural relic virtual restoration method, and specifically, the method comprises the steps: firstly, depicting the global correlation of a cultural relic image through low-rank Tucker decomposition and Stiefel manifold; then, Gaussian noise is removed by utilizing a Frobenius norm, and meanwhile, sparse noise is removed by adopting an L0 norm aiming at the deficiency of an L1 norm; furthermore, a weighted L0 norm is introduced into a gradient image domain to explore a local segmentation smooth structure of the cultural relic image. In addition, the patent designs an ADMM algorithm based on a hard threshold to solve the provided model. In each step of algorithm iteration, a rank increase strategy is adopted to dynamically adjust a tensor rank, so that the burden of searching an optimal tensor rank is avoided, and the calculation cost is reduced. Therefore, the damaged area of the cultural relic is complemented and repaired, and the original appearance of the cultural relic is recovered while the cultural relic is protected.

Description

Technical Field

[0001] The invention belongs to the field of artificial intelligence and pattern recognition, and specifically relates to a method for virtual restoration of cultural relics. Background Art

[0002] Traditional methods of cultural relic restoration mainly rely on manual operations, which inevitably have a direct impact on the cultural relic itself during the restoration process, and may cause secondary damage. In addition, traditional methods are usually time-consuming, costly, and cannot completely restore the original appearance of the cultural relic. In this case, it becomes particularly urgent to find new, more efficient and non-destructive restoration methods.

[0003] The virtual restoration method of cultural relics uses modern technology to digitally protect and restore cultural heritage, and can restore damaged or missing cultural relics without touching the real objects. This not only helps to preserve cultural heritage, but also provides important support for historical research. To promote the protection of cultural heritage, virtual restoration avoids direct contact with the real objects, reduces the secondary damage that may be caused by physical intervention, and improves the safety of cultural relics protection. Timely intervention can be made on the damage of cultural relics, thereby slowing down the further degradation of cultural relics. Using computer-aided technology, the three-dimensional model of cultural relics can be quickly and accurately obtained, and data analysis and processing can be performed. Virtual restoration can accurately restore the missing parts of cultural relics through high-precision simulation technology and maintain the original style. With the rapid development of computer technology and digital technology, digital and virtual technology have provided new means for the protection and restoration of cultural relics. The development of high-precision three-dimensional scanning technology, image processing technology, virtual reality (VR) and augmented reality (AR) technology has made the digitization and virtual restoration of cultural relics possible. Through these technologies, cultural relics can be accurately digitally recorded and virtually reconstructed to achieve lossless restoration and permanent preservation.

[0004] At present, the virtual restoration methods of cultural relics are deeply studied in terms of feature calculation, contour extraction, fracture surface matching, high-level semantic annotation, and three-dimensional model retrieval. Through advanced computer technology combined with heavy stone cultural relics, it is possible to open areas that cannot be reached by human power and reveal the heavy secrets that cannot be penetrated by sight, so that history can be reproduced digitally and culture can be passed on in an information-based manner. It not only demonstrates the great potential of digital technology in the field of cultural relics protection, but also lays a solid foundation for the subsequent restoration, protection, and even research and application of cultural relics. However, although China has made significant progress in the research and application of virtual restoration methods for cultural relics, there are still some challenges. For example, how to obtain more accurate cultural relics data, how to reduce the cost of technology application, and how to increase the popularity of technology still need to be solved.

[0005] In the present invention, a method for virtual restoration of cultural relics is proposed. Summary of the invention

[0006] The present invention relates to a method for virtual restoration of cultural relics. Specifically, first, low-rank Tucker decomposition and Stiefel manifold are used to characterize the global correlation of cultural relic images. Then, the Frobenius norm is used to remove Gaussian noise. At the same time, in view of the shortcomings of the L1 norm, the L0 norm is used to remove sparse noise. Further, the weighted L0 norm is introduced in the gradient image domain to explore the local segmented smooth structure of the cultural relic image. In addition, this patent designs an ADMM algorithm based on hard threshold to solve the proposed model. In each step of the algorithm iteration, the rank increase strategy is used to dynamically adjust the tensor rank, thereby avoiding the burden of finding the optimal tensor rank and reducing the computational cost. Finally, the effectiveness of the proposed method is verified by quantitative evaluation and visual comparison of a series of experiments on simulated data sets and actual data sets. In particular, for the removal of mixed noise, the method proposed in this patent is up to 2.07dB higher than the best comparison result.

[0007] Data collection and preprocessing: Use laser scanning, structured light scanning and other equipment to perform three-dimensional scanning of cultural relics to obtain geometric data. Use high-resolution cameras to capture images of the surface of cultural relics to obtain texture information. De-noise and pre-process the acquired data to ensure data quality. Low-rank matrix decomposition: Represent three-dimensional point cloud data and image data in matrix form. Apply algorithms such as singular value decomposition (SVD) or principal component analysis (PCA) to decompose the matrix into low-rank matrices and sparse matrices. Repair and reconstruction: Fill in the damaged areas of cultural relics through reconstruction of low-rank matrices and sparse matrices. Combined with image processing technology, texture mapping is performed on the repaired geometric data to complete the color and texture information of the missing parts.

[0008] The specific steps of the above-mentioned cultural relic virtual restoration method are as follows:

[0009] Although the L1 norm is the best convex approximation of the L0 norm in the unit sphere, its exact restoration condition is not easy to determine and has been proven to be a biased estimate, which in many cases cannot meet the sparsity requirements of the true solution. In order to overcome the limitations of L1 regularization, this patent proposes a low-rank Tucker decomposition virtual restoration framework for artifacts.

[0010]

[0011] U i ∈St(I i , r i ), i = 1, 2, 3. (1)

[0012] where τ, λ and μ are regularization parameters.

[0013] The advantages of this patent are as follows:

[0014] In the framework of virtual restoration of cultural relics with low-rank Tucker decomposition, the method proposed in this patent uses the L0 norm to remove sparse noise and reduce the error estimation caused by the L1 norm. The L0 norm is introduced to explore the sparsity of the gradient image domain, and the number of non-zero gradients is directly controlled to better characterize the local smoothness of the space-spectrum. The proposed algorithm uses a rank increase strategy to dynamically adjust the tensor rank, which not only avoids the burden of selecting the tensor rank, but also improves the computational efficiency of the algorithm. DETAILED DESCRIPTION

[0015] Figure 1 This is a quantitative comparison chart of the synthetic dataset INDIAN PINES

[0016] The present invention will be further described below in conjunction with embodiments:

[0017] Example 1 Figure 1 The following are the quantitative comparison results of the synthetic dataset INDIAN PINES. The best results of each quality indicator are marked in bold. The numerical results show that the L0RLRTD method proposed in this patent has higher denoising accuracy. In addition, combined with the visual comparison results, it can be found that the deep learning-based FFDNet method cannot successfully repair the artifact images contaminated by mixed noise in Case 3 and Case 4. This is mainly because FFDnet is a single-band deep denoiser, which ignores the correlation between different bands of HSI, resulting in poor restoration effect.

Claims

1. A method for virtual restoration of cultural relics, characterized in that: The steps include: (I) Data collection and preprocessing: Use laser scanning, structured light scanning and other equipment to perform three-dimensional scanning of cultural relics to obtain geometric data. Use high-resolution cameras to capture images of the surface of cultural relics to obtain texture information. De-noise and preprocess the acquired data to ensure data quality. (II) A virtual restoration method for cultural relics is designed. The alternating direction multiplier method based on the hard threshold operator is used to solve the proposed model. At the same time, the rank increase strategy is used to dynamically adjust the tensor rank in each iteration of the algorithm, thereby avoiding the rank selection burden and improving the computational efficiency. The damaged area of ​​the cultural relic is filled by low-rank Tucker decomposition and alternating direction multiplier method reconstruction. Combined with image processing technology, texture mapping is performed on the restored geometric data to complete the color and texture information of the missing parts. The specific steps are as follows: Although the L1 norm is the best convex approximation of the L0 norm in the unit sphere, its exact restoration condition is not easy to determine and has been proven to be a biased estimate, which in many cases cannot meet the sparsity requirements of the true solution. In order to overcome the limitations of L1 regularization, this patent proposes a virtual restoration framework for cultural relics. U i ∈St(I i ,r i ),i=1,2,3. where τ, λ and μ are regularization parameters.

Images