CN-115512354-B - Method and device for identifying guideboard characters

CN115512354BCN 115512354 BCN115512354 BCN 115512354BCN-115512354-B

Abstract

The application relates to the technical field of image recognition, in particular to a recognition method and a recognition device of guideboard characters, wherein the method comprises the steps of cutting out a first sub-image containing guideboard characters to be recognized from a target image; the method comprises the steps of processing a first sub-image into a second sub-image only containing black and white, judging whether a black-and-white distribution area in the second sub-image meets a preset area distribution condition, if so, identifying guideboard characters in the second sub-image to generate an identification result, and if not, expanding a cutting area and returning to the step of cutting the first sub-image containing the guideboard characters to be identified in a target image, wherein the problem of incomplete extraction of the guideboard characters can be effectively solved, and therefore the accuracy of guideboard character extraction is improved.

Inventors

LI ZHENGXU
ZHU LEI
JIA SHUANGCHENG

Assignees

智道网联科技(北京)有限公司

Dates

Publication Date: 20260508
Application Date: 20221027

Claims (8)

1. A method of identifying a guideboard character, comprising: cutting out a first sub-image containing the guideboard characters to be identified from the target image; processing the first sub-image into a second sub-image comprising only black and white; Judging whether a black-and-white distribution area in the second sub-image meets a preset area distribution condition, if so, identifying guideboard characters in the second sub-image to generate an identification result, and if not, expanding a cutting area and returning to the step of cutting a first sub-image containing the guideboard characters to be identified in the target image; the processing of the first sub-image into a second sub-image comprising only black and white comprises the steps of processing the foreground of the first sub-image into black as a foreground color area, processing the background of the first sub-image into white as a background color area, or processing the foreground of the first sub-image into white as a foreground color area, processing the background of the first sub-image into black as a background color area; Correspondingly, the judging whether the black-and-white distribution area in the second sub-image meets the preset area distribution condition comprises the following steps: judging whether the area of the foreground color area in the second sub-image reaches a preset threshold value or not, and judging whether the position of the foreground color area in the second sub-image is at a specified position or not; And under the condition that the area of the foreground color area in the second sub-image does not reach a preset threshold value, before expanding the clipping area, the method further comprises the following steps: A clipping ratio is determined based on a ratio of the foreground color region in the second sub-image, and a clipping region is determined based on the clipping ratio.
2. The method as recited in claim 1, further comprising: the first sub-image is subjected to noise filtering treatment and then is subjected to amplification treatment; accordingly, the processing the first sub-image into a second sub-image including only black and white includes: The first sub-image after the enlargement processing is processed into a second sub-image containing only black and white.
3. The method of claim 1, wherein after generating the recognition result, further comprising: and if the identification result does not accord with the guideboard identification rule, correcting the identification result.
4. The method of claim 1, wherein cropping a first sub-image containing a guideboard character to be identified in the target image comprises: Determining a designated graph in the target image, wherein the designated graph is a graph containing guideboard characters to be identified; A first sub-image associated with the specified graphic is cropped from the target image.
5. A guideboard character recognition apparatus, comprising: the first clipping unit is used for clipping a first sub-image containing the guideboard characters to be identified from the target image; A first processing unit for processing the first sub-image into a second sub-image containing only black and white; A first judging unit, configured to judge whether a black-and-white distribution area in the second sub-image meets a preset area distribution condition; the first recognition unit is used for recognizing the guideboard characters in the second sub-image and generating a recognition result under the condition that the black-white distribution area in the second sub-image meets the preset area distribution condition; A first enlarging unit, configured to enlarge a clipping region when a black-and-white distribution region in the second sub-image does not meet a preset region distribution condition, and control the first clipping unit to re-clip the enlarged clipping region; The first processing unit includes: The first processing module is used for processing the foreground of the first sub-image into black as a foreground color area, and processing the background of the first sub-image into white as a background color area; or a second processing module, configured to process the foreground in the first sub-image into white as a foreground color area, and process the background of the first sub-image into black as a background color area; Correspondingly, the first judging unit comprises: The first judging module is used for judging whether the area of the foreground color area in the second sub-image reaches a preset threshold value or not, and the second judging module is used for judging whether the position of the foreground color area in the second sub-image is at a specified position or not; under the condition that the first judging module judges that the area of the foreground color area in the second sub-image does not reach the preset threshold value, the method further comprises the following steps: and a first determining unit, configured to determine a clipping ratio based on a duty ratio of the foreground color region in the second sub-image, and determine a clipping region based on the clipping ratio.
6. The apparatus as recited in claim 5, further comprising: the second processing unit is used for carrying out noise filtering processing on the first sub-image and then amplifying processing; correspondingly, the first processing unit is specifically configured to process the first sub-image after the amplification processing into a second sub-image that only includes black and white.
7. The apparatus as recited in claim 5, further comprising: and the first correction unit is used for correcting the identification result if the identification result does not accord with the guideboard identification rule.
8. The apparatus of claim 5, wherein the first clipping unit comprises: The image determining module is used for determining a designated graph in the target image, wherein the designated graph is a graph containing a guideboard character to be identified; And the first clipping module is used for clipping a first sub-image associated with the designated graph from the target image.

Description

Method and device for identifying guideboard characters Technical Field The invention relates to the technical field of image recognition, in particular to a recognition method and device of guideboard characters. Background Currently, there are many fields, such as the autopilot field, the street view map field, the dynamic perception field, etc., that have a need for recognizing images, especially identifying guideboard characters in images. In the prior art, a neural learning network is generally adopted to directly identify images, but the method in the prior art often cannot identify guideboard characters completely due to the influences of shooting distance, shooting angle, shielding and the like, so that the accuracy of identification is reduced. Disclosure of Invention In view of the above, the present invention provides a method and apparatus for recognizing guideboard characters to improve accuracy of guideboard recognition. In order to achieve the above purpose, the present invention provides the following technical solutions: The first aspect of the application provides a method for identifying guideboard characters, which comprises the following steps: cutting out a first sub-image containing the guideboard characters to be identified from the target image; processing the first sub-image into a second sub-image comprising only black and white; Judging whether the black-and-white distribution area in the second sub-image meets the preset area distribution condition, if so, identifying the guideboard characters in the second sub-image to generate an identification result, and if not, expanding the cutting area and returning to the step of cutting the first sub-image containing the guideboard characters to be identified in the target image. Optionally, the method further comprises: the first sub-image is subjected to noise filtering treatment and then is subjected to amplification treatment; accordingly, the processing the first sub-image into a second sub-image including only black and white includes: The first sub-image after the enlargement processing is processed into a second sub-image containing only black and white. Optionally, the processing the first sub-image into a second sub-image only comprising black and white comprises processing the foreground of the first sub-image into black as a foreground color area, processing the background of the first sub-image into white as a background color area, or processing the foreground of the first sub-image into white as a foreground color area, processing the background of the first sub-image into black as a background color area; Correspondingly, the judging whether the black-and-white distribution area in the second sub-image meets the preset area distribution condition comprises the following steps: And judging whether the area of the foreground color area in the second sub-image reaches a preset threshold value or not, and/or judging whether the position of the foreground color area in the second sub-image is at a specified position or not. Optionally, before expanding the clipping region, if the area of the foreground color region in the second sub-image does not reach the preset threshold, the method further includes: A clipping ratio is determined based on a ratio of the foreground color region in the second sub-image, and a clipping region is determined based on the clipping ratio. Optionally, after generating the identification result, the method further includes: and if the identification result does not accord with the guideboard identification rule, correcting the identification result. Optionally, the cropping the first sub-image containing the guideboard character to be identified from the target image includes: Determining a designated graph in the target image, wherein the designated graph is a graph containing guideboard characters to be identified; A first sub-image associated with the specified graphic is cropped from the target image. The second aspect of the present application provides a recognition apparatus for a guideboard character, comprising: the first clipping unit is used for clipping a first sub-image containing the guideboard characters to be identified from the target image; A first processing unit for processing the first sub-image into a second sub-image containing only black and white; A first judging unit, configured to judge whether a black-and-white distribution area in the second sub-image meets a preset area distribution condition; the first recognition unit is used for recognizing the guideboard characters in the second sub-image and generating a recognition result under the condition that the black-white distribution area in the second sub-image meets the preset area distribution condition; The first enlarging unit is used for enlarging the clipping area and controlling the first clipping unit to re-clip the enlarged clipping area under the condition that the black-white distribution area in the second s