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

A Natural Scene Text Detection Method Based on Linear Structure Extraction

A linear structure, text detection technology, applied in the field of pattern recognition

Inactive Publication Date: 2017-11-21
CENT SOUTH UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Text extraction from natural scene images is a challenging subject due to the diversity of scene texts and the complexity of backgrounds

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 Natural Scene Text Detection Method Based on Linear Structure Extraction
  • A Natural Scene Text Detection Method Based on Linear Structure Extraction
  • A Natural Scene Text Detection Method Based on Linear Structure Extraction

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0081] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0082] Such as figure 1 Shown is a schematic flow chart of the method of the present invention, to figure 2 (a) Take text detection as an example, the specific steps are as follows:

[0083] Step 1, input the original color image, such as figure 2 As shown in (a), get its R, G and B color channels, such as figure 2 (b-d);

[0084] Step 2, respectively calculate the Hessian matrix transformation results of each pixel in different color channels:

[0085]

[0086] Wherein, image I represents the grayscale image of a certain color channel, and H(x, y) represents the Hessian matrix transformation result of pixel point (x, y);

[0087] Step 3, use the Gaussian kernel function with scale δ(δ∈[1,4], step size 1) to perform convolution operation on the transformation result of Hessian matrix, and calculate the eigenvalue λ 1 and lambda 2 (|λ 2 |≤|...

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 text detection method based on linear structure extraction, which considers the text connected area as a combination of linear structures of different shapes; by extracting the linear structure, the text area is roughly positioned; and then through color clustering and color distance-based Region growing, to obtain the complete text connected domain from the initial text region. The geometric and spatial location features of connected regions are further analyzed to obtain candidate text strings. Among them, the extraction of the linear structure adopts the multi-scale Hessian matrix filtering method, which operates on the three channels of R, G and B of the original color image, so that this method can overcome the influence of illumination on the image to a certain extent; through the Gaussian mixture model Color clustering and region growth based on color distance, two-step color analysis method, make this method better detectable for texts of different sizes and stroke widths, and further improve the robustness of this method.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and relates to a natural scene text detection method based on linear structure extraction. Background technique [0002] With the wide application of image acquisition devices such as digital cameras, cameras, and ultra-high-speed scanners, natural scene text images have received extensive attention. These images usually provide a lot of useful information, among which text information is particularly important. This is because text information is easy to recognize and understand, and can be widely used in related systems and devices, such as visual assistance systems for the blind, travel translation systems, information retrieval systems, robot vision navigation, etc. Therefore, text extraction from natural scene images is an important topic in the field of computer vision. In order to recognize text in natural scene images, experts have designed many OCR character recognition sys...

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): G06K9/20G06K9/34G06K9/62
Inventor 邹北骥吴慧陈再良赵于前
Owner CENT SOUTH UNIV