CN-116152794-B - Ship name recognition method, electronic equipment and storage medium

CN116152794BCN 116152794 BCN116152794 BCN 116152794BCN-116152794-B

Abstract

The application discloses a ship name recognition method, electronic equipment and a storage medium, wherein the ship name recognition method comprises the steps of obtaining a plurality of images to be recognized, carrying out track association on ship detection boxes and text detection boxes in the images to be recognized to determine target ships and target texts which are associated with each other, carrying out track fitting on the basis of the tracks of the text detection boxes associated with the target texts to obtain an included angle between a track fitting straight line and a horizontal plane, correcting the text detection boxes according to the included angle, and recognizing the target texts on the basis of the corrected text detection boxes to obtain a ship name recognition result of the target ships. Namely, the text detection box corresponding to the target text box is corrected through the included angle between the track fitting straight line of the target text and the horizontal plane, so that the recognition accuracy of the target text associated with the target ship is improved.

Inventors

LI YUE
CHEN YASONG
YANG KUNXING
WU CHANGJIN

Assignees

浙江大华技术股份有限公司

Dates

Publication Date: 20260508
Application Date: 20221228

Claims (9)

1. A ship name recognition method, characterized in that the ship name recognition method comprises: Acquiring a plurality of images to be identified; Track association is carried out on ship detection boxes and text detection boxes in the images to be identified so as to determine a target ship and a target text which are associated with each other, wherein a ship track corresponding to the ship detection boxes and a text track corresponding to the text detection boxes are respectively obtained, and the matched ship track and the matched text track are used for determining the target ship and the target text; Performing track fitting on the tracks of the text detection boxes related to the target text to obtain an included angle between a track fitting straight line and a horizontal plane, wherein the tracks of the text detection boxes related to the target text in the images to be identified are obtained; correcting the text detection box according to the included angle, and identifying the target text based on the corrected text detection box to obtain a ship name identification result of the target ship.
2. The ship name recognition method according to claim 1, wherein, Correcting the text detection box according to the included angle, including: performing quality scoring on the target texts in the plurality of images to be identified to obtain a target image corresponding to the target text with the highest quality score, wherein the target image comprises the target text and a text detection box associated with the target text; and rotating the image center of the target image by the included angle so as to enable the target text in the target image to be parallel to a horizontal plane, and obtaining the corrected text detection box.
3. The ship name recognition method according to claim 2, wherein, The text detection box based on correction identifies the target text, obtains a ship name identification result of the target ship, and comprises the following steps: Taking the minimum circumscribed rectangle of the corrected text detection box as a text recognition box; Generating a plurality of text candidate boxes based on the text recognition boxes; And carrying out text recognition on the text candidate boxes, grading the text recognition results, and taking the text recognition result with the highest grading as the ship name recognition result of the target ship.
4. The ship name recognition method according to claim 1, wherein, Track association is carried out on ship detection boxes and text detection boxes in the images to be identified so as to determine a target ship and target text which are associated with each other, and the track association method comprises the following steps: acquiring ship detection frames and text detection frames in a plurality of images to be identified; Respectively carrying out target tracking on the ship detection frame and the text detection frame to obtain corresponding ship tracks and text tracks; And when the ship track is matched with the text track, determining that the ship corresponding to the ship track is a target ship, and determining that the text corresponding to the text track is a target text.
5. The ship name recognition method according to claim 4, wherein, After the target text is acquired, the method further comprises: Performing quality scoring on the target texts in the plurality of images to be identified to obtain text quality scores of the target texts in the images; And caching the target binding information of the target ship and the associated text quality score of the target text into a target cache corresponding to the target ship, wherein the target binding information is the binding information of the target ship and the target text.
6. The ship name recognition method according to claim 5, wherein, After caching the target binding information, further comprising: Traversing the target text in a plurality of images to be identified; If the target text does not have the target ship associated with it and the target text is located in another ship detection box, binding the target text with the target ship in the ship detection box, updating the target cache, and/or If the target text does not have the target ship associated with it and the target text is outside of a ship detection box of the target ship, discarding the target text.
7. The ship name recognition method according to claim 5, wherein, After caching the target binding information, further comprising: inquiring the moving track of the target ship, if the moving track of the target ship is not searched, newly establishing the moving track of the target ship, and inserting the moving track, the target binding information corresponding to the target ship and the target text quality score into a target cache.
8. An electronic device comprising a memory and a processor coupled to the memory, the memory storing at least one computer program that, when loaded and executed by the processor, is configured to implement the ship name recognition method of any one of claims 1-7.
9. A computer readable storage medium, characterized in that the computer readable storage medium has at least one program, which when loaded and executed by a processor, is adapted to carry out the ship name recognition method according to any one of claims 1-7.

Description

Ship name recognition method, electronic equipment and storage medium Technical Field The present application relates to the field of computer vision, and in particular, to a ship name recognition method, an electronic device, and a storage medium. Background With the increasing maturity and popularity of computer technology, in the process of identifying ship names, text regions are often detected in a text detection stage, and characters in the text regions are identified in a text identification stage. In the research and practice process of the prior art, the ship name is interfered during the running process of the ship, so that the ship name recognition effect is reduced, and the recognition accuracy is influenced. Disclosure of Invention The application mainly solves the technical problem of providing a ship name recognition method, electronic equipment and a storage medium, which can determine a related target ship and a related target text according to a related ship detection frame and a related text detection frame, acquire an included angle between a track fitting straight line of the target text and a horizontal plane, further correct the text detection frame, improve the influence on ship name recognition and improve recognition accuracy. The ship name recognition method comprises the steps of obtaining a plurality of images to be recognized, carrying out track association on ship detection frames and text detection frames in the images to be recognized to determine target ships and target texts which are associated with each other, carrying out track fitting on the basis of the tracks of the text detection frames associated with the target texts to obtain an included angle between a track fitting straight line and a horizontal plane, correcting the text detection frames according to the included angle, and recognizing the target texts on the basis of the corrected text detection frames to obtain a ship name recognition result of the target ships. In some embodiments, the track fitting is performed based on the text detection box track associated with the target text, and an included angle between a track fitting straight line and a horizontal plane is obtained; the method comprises the steps of obtaining tracks of text detection boxes associated with target texts in a plurality of images to be identified, carrying out text detection box center point track fitting on the tracks of the text detection boxes to obtain track fitting straight lines corresponding to the target texts, and determining the included angle between the track fitting straight lines and the horizontal plane as the included angle between the target texts and the horizontal plane. In some embodiments, the correcting the text detection box according to the included angle includes performing quality scoring on the target texts in the plurality of images to be identified, and obtaining a target image corresponding to the target text with the highest quality score, wherein the target image includes the target text and the text detection box associated with the target text, and performing image center rotation on the target image according to the included angle, so that the target text in the target image is parallel to a horizontal plane, and the corrected text detection box is obtained. In some embodiments, the text detection box based on correction identifies the target text, and obtains a ship name identification result of the target ship, wherein the method comprises the steps of taking the minimum circumscribed rectangle of the text detection box after correction as a text identification box, generating a plurality of text candidate boxes based on the text identification box, carrying out text identification on the text candidate boxes, and grading the text identification result to obtain the highest grading text identification result as the ship name identification result of the target ship. In some embodiments, the track association is performed on the ship detection boxes and the text detection boxes in the images to be identified to determine the related target ship and the related target text, and the track association comprises the steps of acquiring the ship detection boxes and the text detection boxes in the images to be identified, respectively performing target tracking on the ship detection boxes and the text detection boxes to acquire corresponding ship tracks and text tracks, and determining that the ship corresponding to the ship tracks is the target ship and the text corresponding to the text tracks is the target text when the ship tracks and the text tracks are matched. In some embodiments, after the target text is acquired, the method further comprises the steps of carrying out quality scoring on the target text in the images to be identified to acquire text quality scores of the target text in each image, and caching target binding information of the target ship and the associated text quality scores of the