The reconstruction of multi-angled aggregates (e.g., manufactured aggregates) is increasingly concerned. Based on the compactly supported spherical “Difference of Gaussians” (DOG) wavelets, this paper proposed a novel shape reconstruction method (called the DOG method). In addition, this paper proposed a concrete mesostructure modelling method using the aggregates reconstructed by the DOG method. The results showed that the DOG method showed higher accuracy in reconstructing the multi-angled manufactured aggregates compared to the traditional spherical harmonic (SH) series method, which is because that it could avoid the ringing effect when reconstructing shapes using a high-order series. Additionally, the DOG method outperformed the SH series method in generating the new shape with localized different from original shape, attributed to the wavelet's tight support property. Hence, the DOG method has great potential in randomly generating the shapes of recycled aggregates, which is often adhered by the residual mortar only at some localized locations.

中文翻译：

---

基于球形DOG小波框架的骨料形状重构与混凝土细观结构建模


多角度骨料（例如人造骨料）的重建越来越受到关注。基于紧支持球面“高斯差分”（DOG）小波，本文提出了一种新颖的形状重建方法（称为DOG方法）。此外，本文提出了一种利用DOG方法重建骨料的混凝土细观结构建模方法。结果表明，与传统的球谐函数（SH）级数方法相比，DOG方法在重建多角度人造骨料方面表现出更高的精度，这是因为它可以避免使用高阶级数重建形状时的振铃效应。此外，由于小波的紧支撑特性，DOG 方法在生成局部不同于原始形状的新形状方面优于 SH 系列方法。因此，DOG 方法在随机生成再生骨料形状方面具有巨大潜力，而再生骨料通常仅在某些局部位置被残余砂浆粘附。