Unlock instant, AI-driven research and patent intelligence for your innovation.

A Fully Automatic Soft Segmentation Method for Characters on Textured Background

A textured background, fully automatic technology, applied in the field of image processing, can solve problems such as difficult to separate background and characters, and character breakage

Active Publication Date: 2022-02-08
江苏全丰电气有限公司
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The main purpose of the present invention is to provide a fully automatic soft segmentation method for characters under the texture background, which is used to solve the problem that the gray value of the shading and the gray value of the characters in the printed matter are similar in the prior art, which makes it difficult to separate the background and the characters. And the problem that characters may break during the segmentation process of character graphics with large character edges; the specific technical solutions are as follows:

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Fully Automatic Soft Segmentation Method for Characters on Textured Background
  • A Fully Automatic Soft Segmentation Method for Characters on Textured Background
  • A Fully Automatic Soft Segmentation Method for Characters on Textured Background

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] The invention provides a method for fully automatic soft segmentation of characters under a texture background, and the specific process of the method is as follows:

[0041] First, for the input character image I 1 Perform grayscale conversion to obtain a grayscale image I 2 , the present invention needs to ensure that the grayscale image I 2 The overall brightness of the background is brighter than that of the foreground characters, which is beneficial to the subsequent segmentation of the foreground characters; if the grayscale image I 2 There is a situation where the overall brightness of the background is darker than that of the foreground characters, and the present invention requires the grayscale image I 2 Do the grayscale inversion operation, specifically through the formula s=L-1-r to achieve, so that the grayscale image I 2 After the grayscale inversion operation, it conforms to the overall brightness of the character brightness darker than the background;...

Embodiment 2

[0071] Suppose the character image of a document is I 1 , character image I 1 After grayscale, the grayscale image I is obtained 2 Such as figure 2 As shown, the grayscale image I 2 The grayscale histogram of image 3 As shown, the first segmentation threshold t obtained by using the dual-threshold OTSU segmentation algorithm 1 is 53, the first segmentation threshold t 1 and the position of the corrected threshold in the histogram as Figure 4 As shown, the three vertical lines in the figure respectively represent the position of the corrected threshold, the first segmentation threshold and the second segmentation threshold from left to right, and the valley search strategy is used to correct the first segmentation threshold t 1 The implementation is as follows:

[0072] (1) Set the parameter p of the valley search strategy num for 7.

[0073] (2) Statistical grayscale image I 2 The gray histogram of h 34 ~ h 61 The values ​​of are 70, 69,..., 219, as shown in Tab...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a character automatic soft segmentation method under the texture background, firstly, the input character image I 1 After grayscale preprocessing, the grayscale image I is obtained 2 , using the double-threshold OTSU segmentation algorithm to obtain the grayscale image I 2 The two segmentation thresholds of , using the valley search strategy and based on the lower threshold to find the grayscale image I 2 The trough position of the histogram, the histogram is thresholded with the trough position as the threshold to obtain the image I 3 ; Then use the morphology to image I 3 The foreground part is corroded to get the image I 4 , do dilation operation to get image I 5 , the image I 5 Foreground part removal image I 4 The foreground part of is identified as an unknown area, and the identification image I 4 The foreground part of the foreground area, the image I 3 The area except the unknown area and the foreground area within the range is identified as the background area, and the three-part graph is composed of the unknown area, the foreground area and the background area; finally, each character area is obtained by using the matting algorithm and the connected domain analysis method to realize the soft segmentation operation; The invention effectively improves the accuracy of character segmentation.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a fully automatic soft segmentation method for characters under a texture background. Background technique [0002] Character segmentation is a step in character recognition tasks. The task of character recognition is from an image containing characters to the output of recognized characters. This task can be decomposed into multiple steps, including: character area detection, tilt correction, character segmentation, character recognition and other steps. Character segmentation is a core key step in character recognition tasks. In the character recognition tasks of documents, bills, license plates and other objects, the accuracy of character segmentation often determines the performance of related products. Commonly used character segmentation methods mainly include threshold method, connected domain method, projection method and method based on texture features. The...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06V30/148
CPCG06V10/267
Inventor 林忠黄陈蓉卢阿丽周静波
Owner 江苏全丰电气有限公司