Abstract: Flour mixing is the key technological step of noodle making. The dough crumbs developed in flour mixing directly determine the final quality of noodle. However, the lack of methods for objectively detecting the quality of dough crumbs restricts the development of noodle industrialization. The formation of dough crumbs is a wet granulation process in which cereal flour particles are mixed with moisture to form the stable aggregates, which is influenced by flour components, amount of water addition, methods of dough mixing and the addition of other supplementary materials. The good dough crumbs is characterized by the relatively uniform particle size, the uniform moisture distribution within and between the dough crumbs, and the gluten network that wrap the starch effectively. The deep learning analysis of dough crumbs images divides the formation of dough crumbs into four stages: Wetting adhesion, aggregation forming, fracture dispersion and stable equilibrium. The stage of stable equilibrium plays an important role in the accurate identification of the end point of dough mixing. At present, the traditional approaches for particle size evaluation of dough crumbs are based on visual inspection, time control and sensory evaluation, which cannot achieve accurate determination by instruments or online quality inspection. Basing on image analysis technology, the shadow area of dough crumbs can indirectly reflect the state of dough crumbs and guide the industrial production of noodles. This article presents a review of the fundamental characteristics of dough crumbs in the wheat noodle production process. Specifically, it analyzes the dynamic changes in moisture content, particle size, gluten network structure, and the related influencing factors during the formation of dough crumbs. The article further explores the correlation between dough crumbs and the noodle quality. Additionally, the review covers the recent developments in image detection methods and other detection techniques utilized for dough crumbs detection.