Spectral image intelligent analysis method for natural food additive

The invention relates to the technical field of multi-image analysis, in particular to an intelligent spectral image analysis method for natural food additives. The method comprises the following steps of: filtering an image of each layer of an image pyramid of a spectral grayscale image to obtain a corresponding initial spectral filtering image; obtaining the noise degree of the initial spectral filtering image, and obtaining the combination weight of pixel points in the initial spectral filtering image in combination with the hierarchy of the image and the surrounding gradient distribution of the pixel points; constructing a final spectral filtering image according to the gray value of the corresponding pixel point in the initial spectral filtering image of the image pyramid of the spectral gray image and the combined weight; and analyzing the mixing degree of the natural food additives and the food raw materials based on the position distribution of the additive pixel points in the final spectral filtering