Microplastics are widely distributed in ecosystems and are increasingly found in food. This poses a potential threat to human health. However, current detections of microplastic in food primarily focused on the simple matrices, such as water, milk, and beverages, with relatively few methods available for complex matrices. Due to the strong matrix interference, non-destructive detection of microplastics in food has always been challenging. Thus, in this study, infrared spectral imaging approach was employed in tandem with chemometrics to perform nondestructive and in-situ characterization of microplastics in twelve diverse Chinese diets including meat and seafood stuffs. Results demonstrate that the proposed method can efficiently characterize common microplastics, such as polypropylene (PP), polyethylene terephthalate (PET), and polyethylene (PE), etc., in various complex matrices. The IR spectral imaging was subsequently applied to the detection of microplastics in seafood samples collected from 24 provinces across China. Results revealed the widespread presence of microplastics in seafood diets with significant regional variations. Overall, this study offers an innovative and applicable means for detecting microplastics in complex foods and provides a reference for the rapid detection of microplastics in various materials.