Low-quality license plate image super-resolution reconstruction method based on stacked multi-dimensional information enhanced convolution

The invention belongs to the field of license plate intelligent information detection, and particularly relates to a low-quality license plate image super-resolution reconstruction method based on stacked multi-dimensional information enhancement convolution, and the method comprises the steps: connecting a noise equalization module, a pixel information enhancement module and a character correction module in series; the low-quality license plate information is subjected to feature extraction and fusion from the angles of adjacent feature association relation, different receptive fields, different dimensions, pixel semantic front and back association analysis and the like, and the high-frequency details of the low-quality license plate image are improved by mining the relevance between local features of license plate image bottom information. Therefore, the influence caused by incomplete license plate image information due to strong noise, fuzzy uncertainty and exposure in super-resolution reconstruction of lo