Semantic segmentation method for power tower/pole images

A tower pole image and semantic technology, applied in the field of power tower pole image semantic segmentation, can solve the problems of power tower poles with many edges, lack of flexibility, and does not consider the surrounding labeling conditions, etc., to achieve a single and consistent improved loss function Sexual Enhancement Effects

Inactive Publication Date: 2013-07-24
NORTH CHINA ELECTRIC POWER UNIV (BAODING)
View PDF5 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The segmentation of power tower images has different characteristics from the segmentation of ordinary images: for example, there are more edges of power towers; another example, from the sense of perception, the image of power towers is easily confused with the background
The form of the energy function in the existing image segmentation algorithm is not flexible, and it depends more on the similarity between the image to be segmented and the training data set. In a certain group, the initial labeling of the same superpixel is only based on the feature Differentiate and assign different loss functions, regardless of the labeling of the surrounding neighborhood; or the loss function only considers the influence of the number of superpixels in the group

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
  • Semantic segmentation method for power tower/pole images
  • Semantic segmentation method for power tower/pole images
  • Semantic segmentation method for power tower/pole images

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0021] figure 1 is the flow chart of the method for semantic segmentation of power tower images, such as figure 1 As shown, a kind of power tower image semantic segmentation method that the present invention proposes comprises:

[0022] Step 1: The power tower image is preprocessed, segmented into superpixels, the best matching dataset is selected and features are extracted.

[0023] Preprocessing refers to the work before the segmentation process: including removing noise, transforming the original image into superpixels, and forming an oversegmentation image (Oversegmentation Image). Superpixel (Superpixel) value is a collection of more than a dozen or dozens of pixels with certain common characteristics, and the superpixel can be further segmented to obtain the segmentation result.

[0024] The superpixel formation process can extract the RGB values ​​of all pixels in the image, compare them with the RGB values ​​of the surrounding pixels, and set a threshold as needed. I...

Embodiment 2

[0052] Binary labeling is used for reasoning, that is, the label takes values ​​from {0, 1}, representing two different types of objects. In the actual segmentation problem, the labels are diverse, as long as more object types are learned from the training data set. Can. exist figure 2 and image 3 Taking S as the central superpixel, the "sensitivity" and high-order loss function design criteria in the design of the loss function in the present invention are specifically described through the change of its neighborhood. The following will calculate S if the loss function value marked "1" is taken.

[0053] figure 2 In the four neighborhoods of S, it changes gradually. In (a), the labels in the neighborhood of S are all "0". Therefore, according to the Markov property, the probability of S taking the label "1" is the smallest, that is, the loss function should be the largest. Calculated, the loss function value is 1.

[0054] In (b), (c), (d), as the number of superpixe...

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 semantic segmentation method for power tower/pole images and belongs to the technical field of computer vision and pattern recognition. The method comprises the steps of: pre-processing the power tower/pole images, segmenting the power tower/pole images into super pixels, selecting best matched data sets and extracting features; giving an initialized semantic annotation to each super pixel according to similar relationships between the power tower/pole images and the data sets; respectively substituting the initialized semantic annotations into an annotation space and a feature space, and calculating an annotation-space high-order loss function value and a feature-space high-order loss function value; and optimizing a loss function, and obtaining a global minimum value of the loss function, thereby reaching the aim of semantic image segmentation. According to the method, during the semantic segmentation of the power tower/pole images, the consistency of the same object and edges of different objects are correspondingly enhanced, so that the problems in existing image segmentation and analysis algorithms that the loss function is single, the sensitivity is lower, and the misclassification rate is high are solved.

Description

technical field [0001] The invention belongs to the technical field of computer vision and pattern recognition, and in particular relates to a semantic segmentation method of an electric tower pole image. Background technique [0002] The inspection of electric equipment by electric robots has been widely used. The power robot realizes the inspection of the power pole and tower by taking the image of the power pole and tower and analyzing the captured image. During the inspection of electric robots, the segmentation and analysis of images of electric power towers has become the most challenging problem in data processing and analysis. The segmentation of power tower images has different characteristics from ordinary images: for example, there are more edges of power towers; another example, from the sense of perception, the image of power towers is easily confused with the background. The form of the energy function in the existing image segmentation algorithm is not flexi...

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 Applications(China)
IPC IPC(8): G06T7/00
Inventor 吴华杨国田李郅诚柳长安刘春阳
Owner NORTH CHINA ELECTRIC POWER UNIV (BAODING)
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap