Method and software for automatically judging multi-direction and multi-scale new crown antigen detection result

The invention discloses an automatic distinguishing method for a multi-direction and multi-scale new crown antigen detection result, which comprises the following steps of: constructing a priori structural element, performing priori flexible morphological operator filtering on an original image of a shot antigen kit detection result, calculating a priori flexible morphological gradient, extracting the edge of an antigen kit, and extracting a to-be-distinguished area of test paper. Interference such as a reflective area is eliminated, potential areas are sequentially positioned, and letter discrimination area pre-detection is carried out; carrying out discriminant letter modeling in an angle dimension to form a training set, and combining samples into a tensor form to form a sample feature matrix; the RGB color image is converted to an HSV color space, alphabet image blocks are expanded to extract features, and feature matching operators are constructed to classify alphabets C and T; performing color histogram