Any image is perceived by human eyes subconsciously as a coherent structure with two large contrasting substructures: figure and ground (Koffka 1936, Rubin 1921). The figure, also called the foreground, is derived out of gray-scale pixels that are darker or lighter than the pixel mean value. Subsequently, it can be decomposed into many substructures (sets of interconnected pixels) with farmore smalls than larges. Likewise, decomposed substructures can be divided into figure and ground parts, then decomposed into corresponding substructures. Therefore, substructures of an image and substructures of decomposable substructures can be derived recursively via foreground extraction

(Jiang and de Rijke 2023).

In this tutorial, substructures and the substructures of decomposable substructures are derived recursively by pixel mean value until all substructures are no longer decomposable, considering the recurring notion of far more small substructures than large ones.