In the beverage, food and drug industry, more and more machine vision systems are being used for the defect detection of Polyethylene Terephthalate (PET) bottle caps. In this paper, in order to address the result of cylindrical distortions that influence the subsequent defect detection in the imaging process, a very fast image stitching algorithm is proposed to generate a panorama planar image of the surface of PET bottle caps. Firstly, the three-dimensional model of the bottle cap is established. Secondly, the relative poses among the four cameras and the bottle cap in the three-dimensional space are calculated to obtain the mapping relationship between three-dimensional points on the side surface of the bottle cap and image pixels taken by the camera. Finally, the side images of the bottle cap are unfolded and stitched to generate a planar image. The experimental results demonstrate that the proposed algorithm unfolds the side images of the bottle cap correctly and very fast. The average unfolding and stitching time for 1.6-megapixel color caps image can reach almost 123.6 ms.

Machine vision-based methods are usually applied to detect impurities in transparent liquids like water or alcohols. Oils have higher viscosity and lower light transmittance, so traditional vision detection methods do not work well. A weak oil impurity detection algorithm was proposed using event stream data. In this method, the binary image of the oil was firstly captured using an event camera. To reduce the noise interference from the event camera, image filtering and morphological operations were applied. Then, image algebra operations were used to reduce the oil container’s bottom pattern. Finally, impurity detection was performed through the YOLOv5 network. Three common edible oils serve as the experimental samples. Small flying insects, raw material fragments, metal fragments, hair strands, and tin beads of various sizes are selected as the weak impurities. Experiments were performed on a dataset containing 3000 sample images. To the best of our knowledge, existing algorithms can detect impurities with the minimum size of 0.4mm, and most of the experimental samples are transparent liquids. The proposed method can be applied to detect impurities in water and oils, and the detectable impurity size limit is 0.2mm.

Vehicle-to-Vehicle (V2V) communication plays a critical role in ensuring network reliability in highly dynamic environments. Most existing channel models are based on the wide-sense stationary uncorrelated scattering propagation assumption, which becomes increasingly inadequate under high-mobility conditions and in dense urban communication scenarios. To overcome these limitations, this paper introduces a novel beam-domain channel model (BDCM) that effectively exploits angular sparsity by transforming the channel representation from the spatial domain to the angular domain. This transformation enables the channel model to accurately characterize inherent sparsity characteristics of V2V multiple-input multiple-output (MIMO) channels. In the proposed channel model, we derive and analyze key non-stationary channel propagation characteristics, with a particular focus on the level-crossing rate (LCR), a metric that evaluates how often the signal envelope crosses a specified threshold. This theoretical formulation establishes a solid theoretical basis for capturing the inherent non-stationarity and rapid fading dynamics in V2V environments. Simulation results demonstrate that the proposed BDCM can efficiently and accurately capture the complex characteristics of V2V channels.

Edible oil may be mixed with tiny solid impurities like raw material fragments, hair, metal fragments and etc. during the production and manufacturing process. For food safety reasons, these tiny impurities need to be detected in the quality control process. As compared with manual detection ways, computer vision-based impurity detection methods can greatly reduce the workload and improve the detection efficiency and accuracy. However, there are currently no datasets for the detection of solid impurities in edible oils. For this reason, a dual-modal edible oil impurity (DMEOI) dataset is developed and made publicly available. The dataset is composed of 14520 event and full-pic images and annotated in two ways. Five common edible oils and four typical impurities are included in the dataset. The DMEOI dataset can be used for liquid impurity detection in either single-modal or dual-modal way. To prove the usability of the dataset, four object detection algorithms are applied and compared.

Public food image datasets have primarily focused on recognition or segmentation, with limited resources for comprehensive nutrition analysis, particularly for Chinese cuisine which exhibits high nutritional variability due to diverse cooking methods and regional influences. To address this gap, we introduce a multimodal food image dataset designed to support full-cycle nutritional analysis, including segmentation, category recognition, and nutrient content estimation, tailored specifically to diabetic diets. This dataset is aligned with clinical data from Chinese diabetes cohorts to facilitate accurate dietary management and glucose prediction. It provides a valuable resource for advancing computational methods in meal-level nutrition assessment for diabetic populations.