Picture information extraction method and device, computer device, and storage medium

Abstract

The invention discloses a picture information extraction method and device, a computer device and a storage medium. The method comprises the following steps: obtaining a bill picture to be recognizedand adjusting the skew and the illumination to obtain the pre-processed bill picture; Identifying a plurality of text regions included in the obtained pre-processed bill picture; Acquiring spatial coordinates of each text region in a plurality of text regions, serially connecting corresponding vectors according to a stitching order to obtain a text box sequence; Taking the text box sequence as theinput of the sequence labeling model to obtain the sub-sequence corresponding to the region to be extracted; recognizing the text region corresponding to the sub-sequence of the region to be extracted to obtain the text information corresponding to the region to be extracted. The method does not need to extract and recognize all the text information frames of complex bill, and does not need to calculate the association between the characters in turn, which reduces the computational workload. Moreover, the image recognition technology can be used to train the labeling data of various angles and distorted bill pictures, and has good robustness.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method, device, computer equipment and storage medium for extracting picture information. Background technique [0002] At present, due to the complexity and variety of page text layout relationships of complex bills, the following extraction process is often used: [0003] 1) Text area labeling: label all text areas in the bill image; [0004] 2) Text content extraction and recognition: perform text recognition on all text areas and extract text; [0005] 3) Text content association: Combining the text information and the position of the text in the picture, perform content association on the text information. [0006] The above process has the following disadvantages: it is necessary to extract text information from all text information areas in the bill one by one and associate them one by one, resulting in a large amount of calculation and low execution efficien...

AI Technical Summary

Problems solved by technology

[0007] The embodiment of the present invention provides a picture information extraction method, device, computer equipment, and storage medium, aiming to solve the problem of extracting text information one by one from all text information areas in the bill in the prior art and correlating them one by one, resulting in the processing of a large number of irrelevant texts content, making the entire processing process a large amount of calculation and low execution efficiency

Images