The invention discloses a multi-strategy stored-grain insect vision detection method. The method comprises the steps of 1, acquiring a stored-grain insect image; 2, subjecting the stored-grain insect image to singular value decomposition to obtain an image data matrix (img file = 'dest _ path _ image 268190 001. TIF 'wi = '94 'he = '19 ', ); 3, subjecting the stored-grain insect image to 2-layer wavelet decomposition and reconstruction by adopting a sym N/wavelet base; 4, subjecting the stored-grain insect image to binarization treatment by adopting the Otsu method; 5, subjecting the obtained binarized image to expansion operation; 6; based on the Blob algorithm, subjecting the binarized stored-grain insect image to Blob connected domain analysis. In this way, the purpose of detecting stored-grain insects is realized. According to the technical scheme of the invention, the problem in the prior art that the conventional and artificial detection for stored-grain insects is time-consuming and inefficient can be solved. Based on area parameters (img file = 'dest _ path _ image 457863 002. TIF 'wi = '28 'he = '22 ', ) and perimeter parameters(img file = '170735 dest _ path _ image 003. TIF 'wi = '30 'he = '18 '), stored-grain insects can be rapidly, accurately, nondestructively and automatically detected. Meanwhile, the problem in the prior art that existing stored-grain insect vision detection methods cannot overcome the influence of the granary composite environment can also be solved. As a result, the number of stored-grain insects can be accurately detected, while the locations of stored-grain insects can be accurately marked.