
ABSTRACT The impacts of extrusion on the physical and chemical characteristics of rice, white bean, and powdered red pitaya for the production of third‐generation snacks were evaluated. A mixture of white rice and white bean with incorporation of encapsulated red pitaya (90/10) was conditioned to a moisture content of 27% and processed at extrusion temperatures (ET) ranging from 90°C–129°C and screw speeds (SPs) ranging from 75–125 rpm. The water content, water solubility index (WSI), absorption index (WAI), expansion index (EI), total betalain content (TBC), total phenolic content (TPC), antioxidant activity (AA), color parameter, hardness, microstructure, and sensorial analysis were evaluated. The results indicate that ET and SP significantly affected the studied variables. The hardness, EI, WAI, and WSI were greater at higher ETs and SPs, whereas lower ETs and higher SPs preserved more bioactive compounds, such as TBC, TPC, and AA. After microwave expansion, EI, bulk density, hardness, and TBC decreased because microwave heating produced pores in the expanded snack, reducing mass per unit volume and increasing the product's crispiness, whereas WSI, WAI, TPC, and AA increased. The increase in bioactive compounds after expansion may be attributed to cell wall disruption, which promotes the release of bound compounds and increases their extractability. After processing, microstructural modifications were detected, causing damage to the native starch structure. The treatment at 90°C and 125 rpm presented the highest bioactive values, with a content of 24.35 mg betalains/g, generating a retention of 44% and high AA content of 2.05 μmolTE/g. For its part, the treatment at 125 rpm and 129°C showed the highest sensory acceptability (43.75%), and greater degree of acceptability, hardness, crunchiness, and color. These findings suggest that obtaining snacks with elevated phytochemical content and adequate functional and structural properties is possible.
ABSTRACT Destemming is a critical pretreatment operation in chili pepper processing; however, robust mechanical separation of curved‐stem cultivars remains an unresolved engineering challenge because of irregular stem geometry and stochastic fruit orientation. This work presents a discrete element method (DEM)‐driven optimization framework for a rotary drum destemming system engineered for curved‐stem peppers. The geometric and mechanical properties of the S17 cultivar were determined experimentally, and a multi‐sphere DEM model coupled with the Hertz–Mindlin bonding contact model was constructed to reproduce fruit kinematics and peduncle fracture. Macroscopic validation against angle‐of‐repose measurements returned a mean relative error of 5.04%, below the 8% acceptance threshold. The bonded‐contact parameters were derived from the measured shear and tensile properties of S17 peduncles. Quantitative motion descriptors extracted from the simulations identified wall climbing as the stage that governs stem capture. Quadratic regression models describing the effects of drum rotational speed, drum radius, and rod count on the stem capture rate and the stem breakage rate were established ( R 2 = 0.9230 and 0.8734, respectively), and response surface methodology identified an optimum of 0.049 r s −1 , 1.299 m, and 33 rods. Under these conditions, a full‐scale prototype achieved a net stem cutting efficiency of 97.2% with a fruit tip damage rate of 2.38%, exceeding the stem‐separation efficiencies reported for drum‐type pepper harvester separators. The single‐pass simulated index and the measured multi‐pass efficiency were reconciled by a capture‐probability model. The results provide a DEM‐based parameter‐optimization procedure for separation equipment handling irregularly shaped food materials.
Atlantic salmon is a globally sought-after, high-quality protein source, making the assurance of its quality and safety paramount for both consumer health and industrial sustainability. In automated fillet processing systems, the use of single visible-light imaging is often hampered by reflection artifacts on the fillet surface, which frequently lead to false positives during foreign body detection. To address this problem, we propose a reliable automatic detection method for endogenous foreign bodies in fillets by combining RGB and Multispectral Imaging (MSI) data. We introduce a novel, task-driven Cross-modal Dynamic Collaborative Self-Learning Network (CDCSNet) based on a feature-level fusion strategy. This network enables the modalities to adaptively learn the optimal fusion patterns tailored to the specific characteristics of the foreign bodies. Furthermore, a Dynamically Channel-Gated Decoupling Module (DCGDM) is incorporated into the backbone to effectively mitigate redundant modal information and promote crucial feature complementarity between the two branches. Experimental results demonstrate that the proposed CDCSNet has better performance than other multimodal fusion approaches in both accuracy and robustness. This work provides a promising, multimodality-based solution for intelligent food foreign bodies detection in the food processing industry.
ABSTRACT Hardy kiwifruit ( Actinidia arguta ) is a high‐value berry that is highly susceptible to postharvest latent internal damage. Due to the “epidermal masking effect” during normal ripening, internal defects are often concealed, resulting in a high missed detection rate. To address this, the study proposes a decision‐level fusion detection method mimicking the human visual–tactile mechanism. Based on a four‐point discrete tactile model, the firmness standard deviation (FSD) is introduced to quantify flesh uniformity. By exploiting biomechanical heterogeneity, FSD decouples overall maturity from localized defects, resolving the bottleneck of quantifying latent damage. Concurrently, a surface defect index is extracted via an image enhancement algorithm for integrated internal and external quality grading. A hierarchical gated fusion model was developed, wherein the tactile branch pre‐screens high‐risk samples and the visual branch finely grades potentially acceptable fruits, eliminating full‐scale computation and reducing redundancy. Results demonstrate that the fusion model achieves a grading accuracy of 88.3%, which is a 16.8% improvement over the unimodal visual approach. Crucially, the missed detection rate for latent damage in externally intact fruits is drastically reduced from 59.5% to 10.8%. This work provides a theoretical foundation and engineering value for the mechano‐optical synergistic perception and intelligent postharvest processing of perishable fruits.
ABSTRACT Three‐dimensional food printing (3DFP) has emerged as a promising method for creating personalized nutrient‐rich foods with specific functionality and structure. In this study, a high‐protein, edible seed‐based ink (SBI) was developed using equal proportions of sunflower, watermelon, and black sesame seeds, alongside a naturally sweetened date‐based ink (DBI) composed of date pulp (DP) and date syrup (DS). The rheological properties of both inks were systematically analyzed, focusing on shear‐thinning and structural recovery behavior. The inks had loss factor (tan δ ) values of 0.40–0.44, signifying appropriate viscoelastic properties for extrusion, with structural recovery of 64.99% for the SBI and 58.80% for the DBI, validating their capacity to preserve shape integrity. For both inks, the optimization of printing parameters was performed independently. The SBI exhibited optimal performance with a 1.2 mm nozzle diameter, 67% infill density, rectilinear infill pattern, and 600 mm/min printing speed, 30 rpm motor speed, and 3 bar pressure, with the selection further corroborated by structural stability post freeze‐drying. The DBI was optimized using a 1.2 mm nozzle, 33% infill, grid infill pattern, 600 mm/min printing speed, 30 rpm motor speed, and 2 bar pressure. After post‐processing, the a w of the SBI construct decreased from 0.91 ± 0.01 to 0.63 ± 0.02, indicating shelf stability, with a protein content of 32.37% ± 1.14%, signifying a protein‐rich food product. Under optimum conditions, dual‐ink 3D printed structures were printed in alternating structure (seed‐date) and sandwich (seed‐date‐seed) configurations. The alternating structure exhibited superior sensory acceptance. Overall, this study demonstrates the effective integration of rheological optimization with dual‐ink printing to create functional high‐protein 3D printed foods.