Contaminated film character identification method

Abstract

The invention discloses a contaminated film character identification method, comprising the following steps of carrying out binarization processing on a film character image; obtaining sizes of normalcharacter blocks and interval pixels among the character blocks; establishing and annotating feature templates; finding a completely matching point in a to-be-identified character image; carrying outbidirectional character segmentation starting from the complete matching point, thereby obtaining to-be-identified character blocks; carrying out template matching on all to-be-identified character blocks and the feature templates, thereby obtaining an identification result of the to-be-identified character blocks; and carrying out connection on annotations corresponding to each to-be-identifiedcharacter block in sequence, thereby obtaining a whole character string identified from the to-be-identified character image. According to the method, few feature templates are constructed in two modes of the to-be-identified image characters, the to-be-identified characters are segmented into a plurality of character blocks, and all character blocks are matched with the feature templates, so thespecial character image is identified, the workload is low, the efficiency is high and the accuracy is achieved.

Description

technical field [0001] The invention relates to the field of image recognition, in particular to a method for recognizing characters on contaminated film. Background technique [0002] For the text on the film, especially some text on the medical film, it needs to be recognized accurately and quickly, so as to associate with the relevant data information in the hospital database, realize the self-printing function of the film, and multiple departments can share data, share Machines, thereby reducing queuing phenomenon, shortening queuing time and reducing costs. [0003] Commonly used optical character recognition software can recognize regular and clear characters. Printed characters printed on film are normally white characters printed on a black background, such as figure 1 As shown, it is easier to recognize, but there are also black shadows formed by printing white characters on a white background, such as figure 2 "2017" as shown; or part of a character is printed o...

AI Technical Summary

Problems solved by technology

[0003] Commonly used optical character recognition software can recognize regular and clear characters. Printed characters printed on film are normally white characters printed on a black background, such as figure 1 As shown, it is easier to recognize, but there are also black shadows formed by printing white characters on a white background, such as figure 2 "2017" as shown; or part of a character is printed on a black background and part is printed on a white background, such as figure 2 For the character "1" shown, although the latter two cases are relatively easy to recognize by the human eye, it is difficult for the machine to recognize

Images