Abstract: The presence of unsound kernels in grain reduces its quality and adversely affects processed grain products. The content of unsound kernels also impacts the quality grading of grain according to national purchase standards, leading to economic losses. Traditional detection methods such as manual inspection and chemical reagents are subjective, time-consuming, and do not align with current trends favoring rapid and precise detection. Visible light imaging technology offers fast operation and imaging speed but cannot discern internal grain characteristics. Advanced imaging technologies with high resolution and rapid detection capabilities have thus become pivotal in the field of unsound kernel detection. This paper reviews various imaging techniques utilized for detecting unsound kernels in grain, including visible light imaging, X-ray, thermal imaging, hyperspectral and multispectral imaging, and terahertz imaging. It discusses and compares the strengths and weaknesses of these techniques. The paper introduces the visual appearance and internal feature information of grain separately, highlighting the research progress in combining imaging techniques with machine learning methods for unsound kernel detection. Finally, it outlines current challenges and discusses future directions for improvement, aiming to provide valuable insights for innovative applications in unsound kernel detection in grain.