Quantifying Bread Physical Features: A Comparative Analysis Between Standard Method, Structure from Motion, and RGB-D Cameras

The assessment of bread volume and physical characteristics is fundamental to understanding its quality, texture, and overall consumer satisfaction. In this thesis, we present a comprehensive comparative analysis of three methodologies for quantifying bread physical features: the standard seed method, the Structure from Motion (SfM) technique, and RGB-D cameras. Firstly, we delve into the traditional seed method, which involves measuring bread volume by displacement of seeds in a graduated cylinder. This method has long been employed as a standard practice but is limited by its manual nature and potential for measurement errors. Secondly, we explore the Structure from Motion (SfM) approach, a photogrammetric technique that reconstructs three-dimensional (3D) models from a sequence of overlapping 2D images. SfM offers a non-invasive and relatively efficient alternative to the seed method, with the potential for high accuracy and repeatability. Finally, we investigate the use of RGB-D cameras, which capture both color (RGB) and depth (D) information simultaneously. This advanced imaging technology provides a holistic perspective on bread morphology, enabling precise volumetric analysis and texture characterization. Through a series of experiments and comparative studies, we evaluate the performance, accuracy, and practicality of each method in quantifying bread physical features. We assess factors such as measurement precision, time efficiency, equipment requirements, and cost-effectiveness to provide insights into their applicability in different contexts. By elucidating the strengths and limitations of each methodology, this research contributes to the advancement of bread quality assessment techniques. Our findings inform bakery industry practices, research methodologies, and technological advancements, ultimately enhancing the understanding and optimization of bread production processes.