Search

The virtual restoration of cultural relics by adopting computer technology is the trend of cultural relics protection and value mining. In order to solve the problem of large error in the existing matching methods of non-thin-walled cultural relic fragments, in this paper, a matching method of cultural relic fragments is proposed, when the thickness feature and the contour feature are combined, and the strategy of coarse matching followed by fine matching is adopted. Firstly, the broken contour of cultural relic fragments is extracted, and the upper and lower contour lines are identified; secondly, the thickness feature is calculated and the thickness histogram is constructed. After that, the initial matching of fragments is completed according to the similarity of the histogram. Thirdly, the fracture contour is segmented, and the fragments of the mismatching relationship in the initial matching are also eliminated by the modified Hausdorff distance to optimize the coarse matching results. Finally, the SVD method is adopted to solve the rigid body transformation parameters, and the ICP algorithm is employed to accurately align the fine matching. The experimental results show that with the method in this paper, excellent matching results have been obtained regarding the fragments of the Terracotta Warriors. Compared with matching methods of other cultural relic fragments, the accuracy and algorithm cost of this method are better.

Background Ancestry estimation of skulls is under a wide range of applications in forensic science, anthropology, and facial reconstruction. This study aims to avoid defects in traditional skull ancestry estimation methods, such as time-consuming and labor-intensive manual calibration of feature points, and subjective results. Results This paper uses the skull depth image as input, based on AlexNet, introduces the Wide module and SE-block to improve the network, designs and proposes ANINet, and realizes the ancestry classification. Such a unified model architecture of ANINet overcomes the subjectivity of manually calibrating feature points, of which the accuracy and efficiency are improved. We use depth projection to obtain the local depth image and the global depth image of the skull, take the skull depth image as the object, use global, local, and local + global methods respectively to experiment on the 95 cases of Han skull and 110 cases of Uyghur skull data sets, and perform cross-validation. The experimental results show that the accuracies of the three methods for skull ancestry estimation reached 98.21%, 98.04% and 99.03%, respectively. Compared with the classic networks AlexNet, Vgg-16, GoogLenet, ResNet-50, DenseNet-121, and SqueezeNet, the network proposed in this paper has the advantages of high accuracy and small parameters; compared with state-of-the-art methods, the method in this paper has a higher learning rate and better ability to estimate. Conclusions In summary, skull depth images have an excellent performance in estimation, and ANINet is an effective approach for skull ancestry estimation.

Skull registration is one of the important steps in craniofacial reconstruction, and its registration accuracy and efficiency have an important impact on the reconstruction results. To solve the problem of low accuracy and efficiency of existing skull registration methods, a hierarchical skull point cloud registration method is proposed in this paper. The whole registration process is divided into a rough registration stage and a fine registration stage. Firstly, feature points are extracted from the pre-processed skull point cloud model, and a local coordinate reference system is established according to the feature points and their neighbor points. The improved spin image is used to construct the local feature descriptor. The feature matching is carried out according to the nearest neighbor algorithm, and the k-means algorithm is used to eliminate the mismatching points to achieve skull rough registration. Then, based on rough registration, we use an improved ICP algorithm to achieve fine registration of the skull. In this process, we use random sampling to reduce the search scale of points and add geometric feature constraints to further eliminate mismatched points. Finally, the whole registration algorithm is applied to the skull point cloud data to verify. The experimental results show that, compared with other methods, the registration effect and efficiency of the proposed method are superior to those of other methods. In order to verify the universality of the method, we also use a common data set for verification. Experiments show that the method is also very effective.

The estimation of ancestry is an essential benchmark for positive identification of heavily decomposed bodies that are recovered in a variety of death and crime scenes. Aiming at the problem of skull ancestry estimation, this paper proposes an improved convolutional neural network method to realize ancestry estimation. We use the six-angle images of the skull as the input of the network. By improving the basic model LeNet5 of the convolutional neural network, we preserve the depth semantics and content information of the image, reduce the number of parameters, and ensure the learning ability of network features. In the experiment, 156 yellow skulls from northern China and 178 white skulls from Xinjiang were used as subjects, 80% of skull samples were used as training sets and 20% as test sets. Experiments on the training set and test set show that the improved CNN network architecture achieves 95.88% accuracy on the training set and 95.52% accuracy on the test set. In addition, we also designed experiments on the contribution of various parts of the skull to ancestor identification. The experimental results show that each region of the skull is useful for ancestor identification, but the effect is different. Compared with other networks, the network structure of this paper has the highest accuracy and better performance.

Craniofacial reconstruction is to estimate a person’s face model from the skull. It can be applied in many fields such as forensic medicine, archaeology, and face animation. Craniofacial reconstruction is based on the relationship between the skull and the face to reconstruct the facial appearance from the skull. However, the craniofacial structure is very complex and the relationship is not the same in different craniofacial regions. To better represent the shape changes of the skull and face and make better use of the correlation between different local regions, a new craniofacial reconstruction method based on region fusion strategy is proposed in this paper. This method has the flexibility of finding the nonlinear relationship between skull and face variables and is easy to solve. Firstly, the skull and face are divided into five corresponding local regions; secondly, the five regions of skull and face are mapped to low-dimensional latent space using Gaussian process latent variable model (GP-LVM), and the nonlinear features between skull and face are extracted; then, least square support vector regression (LSSVR) model is trained in latent space to establish the mapping relationship between skull region and face region; finally, perform regional fusion to achieve overall reconstruction. For the unknown skull, first divide the region, then project it into the latent space of the skull region, then use the trained LSSVR model to reconstruct the face of the corresponding region, and finally perform regional fusion to realize the face reconstruction of the unknown skull. The experimental results show that the method is effective. Compared with other regression methods, our method is optimal. In addition, we add attributes such as age and body mass index (BMI) to the mappings to achieve face reconstruction with different attributes.

For any positive integer n, let $$\sigma (\mathrm{n})$$ and p(n) denote the sum of divisors and the least prime divisor of n respectively. Let a, b be positive integers. In this paper we prove the following two results: (i) If 4 | a and $$\gcd (a, b)=1$$ , then a and b do not satisfy $$\sigma (a)= \sigma (b)=a+b$$ . (ii) If $$a>10^{8}$$ and $$p(a)>2\log _{2}a+1$$ , where $$\log _{2}{a}$$ is the logarithm of a with base 2, then a and b do not satisfy $$\sigma (a)=\sigma (b)=a+b+\lambda $$ , where $$\lambda \in \{0,\pm 1\}$$ .

The disadvantages of BOW (Bag of words model) for image classification include the large amount of data in generating a codebook by clustering, redundant code words that may affect the classification results and so on. The process of BOW for the classification can be improved through the Laplace weights to improved fuzzy C means algorithm, and obtaining codebook with more ability to distinguish between images the optimization methods for Codebook based on interdependence between the image code word co-occurrence relationships. Experimental results show that the improved clustering algorithm and code optimization method proposed in this paper can help to establish a codebook with more ability in expression and discrimination, which can significantly improve the accuracy of image classification.

In this paper the level-set method is introduced to mesh simplification of the three-dimensional model, especially for cultural heritage. First, constructs a signed distance function, the zero set of which define is the surface. Then, introduces a energy functional and minimizes it to derive a second order geometric partial differential equation (GPDE) in the level-set formulation. Thus, the mesh simplification process is transferred into volumetric diffusion process of an implicit model. This method is used for simplification of digitalization of large scene and relics in cultural heritage. Finally, the level-set mesh simplification algorithm are compared and analyzed with mesh simplification algorithm based on point-collapse on the visual quality and geometric error. The experiment shows that it can be applied any topology of mesh. Even model is large-scale simplified, it is still able to maintain a considerable number of important geometric features and better overall visual effect while maintaining a low error level.

We present a pipeline for eaves tiles geometry reconstruction from single binary image, which collected in their monotype album. This pipeline consists of shape from shading based spatial points reconstruction, feature-preserved point cloud smoothing for high quality models, and 3D printing techniques to obtain the duplication of eaves tiles. Compared with other reverse engineering methods, we reduce the high demand of original point cloud data in modeling, efficiently reconstruct the large number of eaves tiles, and by duplicating we reappear the lost culture relic face after thousands of years. We illustrate the performance of our pipeline on Qin Dynasty eaves tiles.

We propose a novel approach for automatic relic positioning in archaeological site using shape matching technique. The method rapidly locates the initial position of relic in scene model by sketch based image registration. First we extract apparent ridges on models to draw the sketches, subsequently the relic can be substantially located by image registration process. We use 3D partial shape matching to identify the positioning result further. In contrast to the differential invariants used in many literatures, this paper addresses the integral invariants which exhibited robust and stable in dealing with high noise relic models. The suggested approach has been well applied in archaeology report generation during the process of 3rd excavation of Qin Terracotta Warriors and offers an outstanding efficiency.

A Level-of-Detail model is proposed to improve the efficiency of hair rendering. The model considers the simplification on both strands density and controlling points of strand curves, which common used uniquely by traditional solutions. On the other hand, we provide artists a novel constraint tool that inspired by existing hair styling method. Artists can determine LOD parameters value in 3D space by their own ideas but not only by viewing distance, whether there is a hair model or not. Another benefit of spatial constraint using is that those separating hair strands can clamp together dynamically if they are growing under working from the same hairstyle constraint. Thus strands can merge into hair wisp model through K-d tree searching for nearest control points. Result shows a notable improvement on photo realistic hair rendering with our model.

Craniofacial similarity measuring is a difficult scheme in the field of craniofacial reconstruction. In this paper, we define a hemisphere as a referenced model, which is used to select the craniofacial homologous vertices and to construct a new transformation coordinate system. Then the hemisphere is transformed into different craniofacial models. The transformations are a basis for craniofacial similarity measuring. We adopt a modified version of relative angle-context distribution (RACD) to measure one-vertex transformation irrespective of global affine transformation between the hemisphere and one craniofacial model, which we call BRACD (bending relative angle-context distribution). What's more, the BRACDs of all vertices will reflect the whole bending information. Following χ2 statistics, the dissimilarity between different craniofacial models can be computed from above values. Therefore a similarity measuring formula can estimate craniofacial similarity. The experimental results show that the hit-rate of our approach is higher than RACD.

This paper provides a compression method of two-dimensional contour model based on compressed sensing. First, this method gets the 2-D geometric signal through discrete representing the two-dimensional contour model. Then, construct a basis using Laplace operator of the Two-dimensional contour model, thus we get the sparse representation of the 2-D geometric signal based on this basis. Last, we complete compressing the Two-dimensional contour model, through random sampling geometry signals based on Compressed Sensing. In the recovery process, we reconstruct the 2-D geometric signal through optimizing 1-norm of the sparse signal. This method completed the compression of Two-dimensional contour model in the sampling process. Experimental results show that the compression ratio of this method is high, restore effect is good and is suitable for large-scale data compression.

Because the greatly most algorithms of real color image enhancement did not relate to the spectral sensitivity of most human vision, the color restoration functions of some real color image enhancement algorithms are greatly at random and not proved , and the real color images enhanced which are based on illumination-reflectance model have the loss of details and the ‘halos’, we proposed a new algorithm to overcome these disadvantages. Firstly, we transform the real color image from RGB space to HSV space which is approximately orthonormal system between color and value. Secondly, the illumination and the reflectance of value are separated by homomorphic decomposition based on illumination-reflectance model. We have discovered that the high dynamic range of image including high bright lights is mainly caused by the reflectance. Thirdly, the details of reflectance are preserved by wavelet transform. Fourthly, the dynamic range of reflectance is compressed by Butterworth filtering. Lastly, the energy of the saturation of real color image in HSV space is attenuated according to the spectral sensitivity of most human vision. Experiments show that the color of image enhanced adapts well to the most people vision and is not obviously distorted, the preservation of the details is perfectly and the ‘halos’ is effectively restrained. Real color images enhanced by the algorithm are much better than the multi-scales Retinex’s with color restoration (MSRCR).

Because the color restoration functions of real color image enhancement are greatly at random and not proved, and the enhanced real color images which are based on illumination-reflectance model have the loss of details and the 'halos', we proposed a new algorithm to overcome these disadvantages. Firstly, we transform the real color image from RGB space to HSV space which is approximately orthonormal between color and value, and keep the color of image. Secondly, the illumination and the reflectance of value are separated by homomorphic decomposition based on illumination-reflectance model. We have discovered that the high dynamic range of image including high bright lights is mainly caused by the reflectance. Thirdly, the details of reflectance are preserved by wavelet transform. Lastly, the dynamic range of reflectance is compressed by Butterworth filtering. Experiments show that the color of image enhanced is not obviously distorted, the preservation of the details is good and the 'halos' is effectively restrained. Real color image enhanced by the algorithm is much better than the multi-scale Retinex's with color restoration (MSRCR).

Ontology for insect recognition is a powerful tool to express semantics of insects and to share the knowledge of insects. In this paper, the framework of ontology based insect recognition is proposed. The framework includes training layer, ontology layer and recognition layer as well as methods for communicating between different layers. The training layer includes insect image samples database, feature extraction, insect feature knowledge database, insect type knowledge database; the ontology layer includes insect image visual feature ontology, insect image media feature ontology, classification module, insect morphology and taxonomy ontology.

Because the most algorithms of real color image enhancement did not relate to the spectral sensitivity of most people vision, the color restoration functions of some real color image enhancement algorithms are greatly at random and not proved , and the enhanced real color images which are based on illumination-reflectance model have the loss of details and the ‘halos’, we proposed a new algorithm to overcome these disadvantages. Firstly, we transform the real color image from RGB space to HSV space which is approximately orthonormal between the color and the value. Secondly, the illumination and the reflectance of the value are separated based on a new illumination-reflectance model described by stationary wavelet transform. We have obtained the conclusion that the illumination of image is mainly preserved in the low frequency part of stationary wavelet decomposition and the reflectance is mainly preserved in the high frequency parts. Thirdly, the dynamic range of illumination is compressed by Gaussian filtering. Lastly, the energy of the saturation of real color image in HSV space is attenuated according to the spectral sensitivity of most people vision. Experiments show that the color of image enhanced adapts well to the most people vision and is not obviously distorted, the preservation of the details is excellent and the ‘halos’ is effectively restrained. Real color image enhanced by the algorithm is much better than by the multi-scales Retinex's with color restoration (MSRCR).

Real color images are images with noise. The application of gray-scale image enhancement methods in the three-channel of images (R,G,B), on the one hand, increased noise, on the other hand, tended to have a color deviation, which impacted the effect of enhancement. In view of human visual system is more sensitive to lightness than hue, saturation and noise affects lightness more than a,b color channel in CIELab space. It proposed a method of image denoised and enhanced in CIELab space. In this paper, we transform a real color image from RGB space to CIELab space at first, then maintain a,b channel unchanged, decompose lightness channel into high frequency and low frequency information using wavelet transform. And then it directly compresses dynamic range of the j-scale low frequency information based on a new Illumination—Reflectance Model ,uses new threshold function to denoise in all scale high frequency information. At last, it transforms into a real color images from CIELab space to RGB space. Experiments show that image has a good dynamic compression range and slight color deviation, which improve the visual effects and the true effects of color.

Digital image enhancement provides good dynamic range compression in each red, green and blue (RGB)channel, but it fails in color rendition. In this letter, we propose to transform a real-color image from RGB space to HSV space at first.Then, saturation channel is enhanced by Butterworth high-pass homomorphic filter and value channel is enhanced by multi-scale Gaussian high-pass homomorphic filters. In addition , we propose anew method to fix on the cut-off frequency of high-pass filter in homomorphic enhancement algorithm and explain the ratio coefficient of high-pass filter.Experiments show the compression of real-color image dynamic range is obviously effective, and it successes in color rendition.

The goal of adding content-based 3d model retrieval to semantic web is to develop a rich set of 3d model semantic representation to enable both humans and machines to generate and understand model descriptions and processing which can be used to enable fast efficient retrieval from model collections. The purpose of Ontology for Content-based 3d Model retrieval is intended to describe model information regardless of storage, feature extraction, creation. In our research, the ontology includes the information on media features, low level visual descriptors and their relationship and non media features of 3d model. It is represented in RDF Schema and implemented in protege.

In order to share and reuse the insect knowledge the infrastructure of the semantic grid is needed and ontology provides the basis of management of semantics for the semantic grid. Therefore the ontology for insect morphology and taxonomy (IMTO) is developed. IMTO includes insect order sub-ontology and family sub-ontology. Protege 3.1 is used for ontology development of concepts and relationships that represent insect morphology and taxonomy

We present a new method to estimate the pose of human head from a monocular image in this paper. The method only uses a face model that formed by five points (four eye-corners and subnasale) in human face, does not need any auxiliary equipment. The estimation involves the research result of the anthropometry in the Mongolian race's feature we had obtained, and employs the theory of the projective geometry. We divide the pose computation into separate estimation of the orientation about X, Y and Z-axis, respectively. For the automatic detection of these points, active appearance models (AAM) are utilized in our experiment. It had been shown that the results of experiment -were very close with actual rotational angles of human face. Especially when facial rotation in a naturally small angle, less than 30/spl deg/, we can obtain the approximate accurate value of rotational angles by using this method.