Abstract:

Aiming at the unrealistic and inaccurate problems of 3D reconstruction object texture, a cultural-relic texture mapping algorithm based on Polynomial Texture Maps（PTM） model is proposed. This algorithm takes advantage of the polynomial color dependence and adopts a biquadratic polynomial that characterizes color and brightness dependence coefficient for storing texel data. Thereby, the surface texture of the object can be reconstructed under varying lighting conditions. In addition, aiming at the problems of reconstruct image texture bluring, ghosting and object detail texture missing, which are result from diffuse reflection anomaly?of the original acquired image, the uneven distribution of luminance information, and the inaccurate reconstruction fitting coefficient of the original PTM model, a PTM algorithm for improving the basis function of polynomial and optimizing the fitting coefficient and an optimization method for PTM image acquisition equipment are proposed. Finally, the experiment proves that the authenticity and accuracy of the reconstructed object texture are effectively improved.

Key words: artifacts, texture mapping, biquadratic polynomial, fitting coefficient, Bidirectional Texture Function（BTF）, Polynomial Texture Maps（PTM） model

摘要：

针对三维重建物体纹理不真实与不精确问题，提出了基于PTM（Polynomial Texture Maps）模型的文物纹理映射算法，该算法利用了多项式颜色依赖性的特质，采取表征颜色与亮度依赖关系的双二次多项式系数进行存储纹素数据，从而实现在变化的光照条件下重建物体的表面纹理。此外针对原PTM模型中存在的原始采集图像漫反射异常、亮度信息分布不均匀、重建拟合系数不精确，导致的重建图像纹理模糊、存在重影、物体细节纹理缺失等问题，提出了改进多项式的基函数和优化拟合系数的PTM算法以及PTM图像采集设备的优化方法，最后经实验验证，提出的算法使得重建物体纹理的真实性与精确度均得到了有效提高。

关键词: 文物, 纹理映射, 双二次多项式, 拟合系数, 双向纹理函数（BTF）, PTM模型