MULTIMODAL FOOD TEXTURE PREDICTION USING TEMPLATE MATCHING TECHNIQUES

Abstract

Accurate texture forecasts are essential for a number of industries, including food quality assurance, automated cooking systems, and consumer satisfaction surveys. This study presents a novel, all-encompassing method for predicting food texture by integrating visual and tactile data using advanced template matching algorithms. Crunchy, soft, chewy, and crispy are among the textures that the system correctly detects and labels by combining information from RGB pictures, depth maps, and force-feedback signals. By integrating real-time sensory input with a meticulously maintained library of known texture patterns, we can create incredibly accurate texture estimations using a process known as template matching. Extensive experiments have shown that the suggested approach performs better than traditional one-dimensional texture analysis methods under various lighting, occlusion, and surface contamination conditions. By allowing robots to evaluate food similarly to people, this study advances human-computer interaction in culinary robotics and smart food processing systems.