Optical character recognition method and apparatus, electronic device and storage medium

An electronic device, comprising:

• at least one processor;
• and
• a memory communicatively connected with the at least one processor;
• wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a method for optical character recognition, wherein the method comprises:
• determining, for a to-be-recognized image, a text bounding box of surrounding a text area in the to-be-recognized image, and extracting a text area image from the to-be-recognized image according to the text bounding box, determining an adjustment mode of the text area image, and adjusting a dimension of the text area image according to the determined adjustment mode;
• determining a bounding box surrounding text lines in the text area image, and extracting a text-line image from the text area image according to the bounding box;
• and
• performing text sequence recognition on the text-line image, and obtaining a recognition result,
• the method further comprises:
• inputting the to-be-recognized image to a pre-trained text scale prediction model, and obtaining a text mask graph with a single channel and a text scale graph;
• wherein values of pixels in the text mask graph respectively represent probabilities of the corresponding pixels belonging to a text area, and values of pixels in the text scale graph respectively represent ratios of dimensions of the shortest edges of text lines to which the corresponding pixels belong to a preset dimension;
• wherein the determining a text bounding box surrounding a text area in the to-be-recognized image comprises:
• determining the text bounding box surrounding the text area in the to-be-recognized image according to the text mask graph;
• and
• determining, for any text area image, values of pixels of the text area image in the text scale graph respectively, the pixels of the text area image having the same value in the text scale graph,
• wherein the adjustment mode of the text area image comprises:
• under a precondition of retaining an aspect ratio of the text area image, adjusting width and height dimensions of the text area image to make the dimensions of the shortest edges of the text lines after adjustment equal to the preset dimension.

A non-transitory computer readable storage medium with computer instructions stored thereon, wherein the computer instructions are used for causing a computer to perform a method for optical character recognition, wherein the method comprises:

• determining, for a to-be-recognized image, a text bounding box surrounding a text area in the to-be-recognized image, and extracting a text area image from the to-be-recognized image according to the text bounding box, determining an adjustment mode of the text area image, and adjusting a dimension of the text area image according to the determined adjustment mode;
• determining a bounding box surrounding text lines in the text area image, and extracting a text-line image from the text area image according to the bounding box;
• and
• performing text sequence recognition on the text-line image, and obtaining a recognition result,
• the method further comprises:
• inputting the to-be-recognized image to a pre-trained text scale prediction model, and obtaining a text mask graph with a single channel and a text scale graph;
• inputting the to-be-recognized image to a pre-trained text scale prediction model, and obtaining a text mask graph with a single channel and a text scale graph;
• wherein values of pixels in the text mask graph respectively represent probabilities of the corresponding pixels belonging to a text area, and values of pixels in the text scale graph respectively represent ratios of dimensions of the shortest edges of text lines to which the corresponding pixels belong to a preset dimension;
• wherein the determining a text bounding box surrounding a text area in the to-be-recognized image comprises:
• determining the text bounding box surrounding the text area in the to-be-recognized image according to the text mask graph;
• and
• determining, for any text area image, values of pixels of the text area image in the text scale graph respectively, the pixels of the text area image having the same value in the text scale graph,
• wherein the adjustment mode of the text area image comprises:
• under a precondition of retaining an aspect ratio of the text area image, adjusting width and height dimensions of the text area image to make the dimensions of the shortest edges of the text lines after adjustment equal to the preset dimension.