Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Image Semantic Annotation Method Based on Energy Model

A technology of semantic annotation and energy model, which is applied in semantic analysis, image analysis, image data processing, etc., can solve problems such as semantic confusion, decrease in annotation accuracy, and failure to consider spatial constraints, so as to improve accuracy and avoid semantic confusion.

Inactive Publication Date: 2017-07-21
ZHEJIANG UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, only the constraint relationship between semantics (that is, labels) is considered, and the spatial constraint relationship between semantics is not considered, so it is easy to cause semantic confusion, resulting in a decrease in labeling accuracy

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
  • Image Semantic Annotation Method Based on Energy Model
  • Image Semantic Annotation Method Based on Energy Model
  • Image Semantic Annotation Method Based on Energy Model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0078] The present invention will be described in detail below in conjunction with specific embodiments.

[0079] An image semantic annotation method based on an energy model, comprising:

[0080] (1) Divide the original image into several regional images, and extract the visual feature vectors of each regional image.

[0081] In this embodiment, the visual feature fuzzy c-means (FCM) clustering algorithm is used to segment the original image and extract the visual feature vectors of the images in each region. The visual feature vector can be a feature vector based on the Moving Picture Experts Group-7 (MPEG-7) feature, or a feature vector based on the scale-invariant feature transform (Scale-invariant feature transform, SIFT) A feature vector, in this embodiment, is a feature vector based on SIFT features.

[0082] (2) According to the visual feature vector of each region, use the trained SVM classifier to determine the candidate semantic labels of each region image, and th...

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 an image semantic labeling method based on an energy model. In the image semantic labeling method, the image-semantic label in different regions of the original image is constructed by using the corresponding potential of the regional image-semantic label and the interaction potential between the semantic labels. The full energy function under the corresponding relationship is used to fuse the image context information and appearance information by using the full energy function. Compared with the contextual object classification models in the prior art that only use co-occurrence information between objects, or only use fixed spatial relationships and simple frequency counting methods, the full energy function proposed in the image semantic tagging method of the present invention passes through the fuzzy space The relationship makes full use of the spatial structure information between objects, effectively avoids the problem of semantic confusion, and improves the accuracy of semantic annotation.

Description

technical field [0001] The invention relates to the technical fields of image retrieval and automatic image labeling, in particular to an image semantic labeling method based on an energy model. Background technique [0002] In order to classify image scenes semantically, objects in the segmented regions need to be annotated after image segmentation. Object annotation in image segmentation regions directly affects the accuracy of scene semantic understanding and classification. Many researchers have carried out object labeling work in images, basically using the underlying visual features of image regions to classify objects. In recent years, researchers have carried out object recognition with fusion of context information, but its accuracy has not met the actual needs and needs to be improved. [0003] In order to solve the problem of combining contextual information into an object classification framework, machine learning techniques are generally borrowed to fuse visua...

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): G06F17/30G06K9/66G06K9/62G06T7/11
CPCG06F40/30G06F18/29G06F18/24
Inventor 姚敏赖盛章李昌英吴朝晖
Owner ZHEJIANG UNIV
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
Eureka Blog
Learn More
PatSnap group products