Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Multidimensional image segmentation method

An image segmentation, multi-scale technology, applied in the field of image segmentation of spectral clustering technology, can solve the problems of vector time-consuming, low accuracy of image edge contour extraction, unstable storage and calculation subspace clustering, etc. Effects of time, edge accuracy, complexity and computation time reduction

Active Publication Date: 2017-03-15
KUNMING UNIV OF SCI & TECH +1
View PDF3 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] The normalized cut algorithm based on spectral graph theory can effectively solve the local constraints and consistency problems in image data, but due to the limitations and complexity of similarity matrix construction, storage and calculation, and the instability of subspace clustering, it will make the It cannot produce better segmentation results in complex and similar backgrounds
The effective combination of normalized cut and multi-scale information enables the target to produce good segmentation results even in complex environments, but the data-driven serial operation cannot guarantee that the error merging of pixels in adjacent areas in low-scale images will be propagated to High scale image
The multi-scale normalized cut method can not only effectively solve the problems existing in the above methods, but also introduce a multi-scale space parallel clustering method to greatly make up for the shortcomings of the large order of the similarity matrix and low computational efficiency, but the method still has The accuracy of image edge contour extraction is not high, and it is time-consuming to construct similarity matrices of each scale and solve eigenvectors

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
  • Multidimensional image segmentation method
  • Multidimensional image segmentation method
  • Multidimensional image segmentation method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0061] Embodiment 1: as Figure 1-21 As shown in Fig. 1, a multi-scale image segmentation method, first uses the semi-reconstruction algorithm of anti-symmetric biorthogonal wavelet transform to extract the edge contours of multiple scales of the image to be tested, and directly constructs each scale by combining the intensity and position features of each scale. Scale similarity matrix; secondly, perform matrix downsampling on the similarity matrix of each scale to obtain the normalized similarity matrix and downsampled similarity matrix of each scale; then construct multi-scale normalized similarity matrix, down-sampled multi-scale similarity matrix and cross-scale Constraint matrix; finally, the normalized cut method is used to solve the downsampling eigenvector, and the final result is obtained by upsampling multiplication and discretization.

Embodiment 2

[0062] Embodiment 2: as Figure 1-21 As shown in Fig. 1, a multi-scale image segmentation method, first uses the semi-reconstruction algorithm of anti-symmetric biorthogonal wavelet transform to extract the edge contours of multiple scales of the image to be tested, and directly constructs each scale by combining the intensity and position features of each scale. Scale similarity matrix; secondly, perform matrix downsampling on the similarity matrix of each scale to obtain the normalized similarity matrix and downsampled similarity matrix of each scale; then construct multi-scale normalized similarity matrix, down-sampled multi-scale similarity matrix and cross-scale Constraint matrix; finally, the normalized cut method is used to solve the downsampling eigenvector, and the final result is obtained by upsampling multiplication and discretization.

[0063] The concrete steps of described method are as follows:

[0064] Step1. Input an image to be tested with a size of M×N, con...

Embodiment 3

[0080] Embodiment 3: as Figure 1-21 As shown in Fig. 1, a multi-scale image segmentation method, first uses the semi-reconstruction algorithm of anti-symmetric biorthogonal wavelet transform to extract the edge contours of multiple scales of the image to be tested, and directly constructs each scale by combining the intensity and position features of each scale. Scale similarity matrix; secondly, perform matrix downsampling on the similarity matrix of each scale to obtain the normalized similarity matrix and downsampled similarity matrix of each scale; then construct multi-scale normalized similarity matrix, down-sampled multi-scale similarity matrix and cross-scale Constraint matrix; finally, the normalized cut method is used to solve the downsampling eigenvector, and the final result is obtained by upsampling multiplication and discretization.

[0081] The concrete steps of described method are as follows:

[0082] Step1. Input an image to be tested with a size of M×N, con...

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 relates to a multidimensional image segmentation method, and belongs to an image segmentation field of a spectral clustering technology. The multidimensional image segmentation method comprises steps that an antisymmetrical bi-orthogonal wavelet transformation semi-reconstruction algorithm is used to for multi-dimensional edge contour extraction of a to-be-tested image, and similar matrices of various dimensions are directly established by adopting the combination of the strengths and the position characteristics of the various dimensions; the matrix downsampling of the similar matrices of the various dimensions is carried out to acquire the uniformization similar matrices and the downsampling similar matrices of the various dimensions; the multi-dimensional uniformization similar matrices, the downsampling multi-dimensional similar matrices, trans-dimensional constraint matrices are established; a uniformization segmentation method is used for downsampling characteristic vector solution, and a final result is acquired by adopting multiplication and discretization processing of upsampling. Edges of segmentation results are more accurate, and complexity of establishing the similar matrices and operation time are reduced.

Description

technical field [0001] The invention relates to a multi-scale image segmentation method, which belongs to the field of image segmentation of spectral clustering technology. Background technique [0002] The normalized cut algorithm based on spectral graph theory can effectively solve the local constraints and consistency problems in image data, but due to the limitations and complexity of similarity matrix construction, storage and calculation, and the instability of subspace clustering, it will make the It cannot produce better segmentation results in complex and similar backgrounds. The effective combination of normalized cut and multi-scale information enables the target to produce good segmentation results even in complex environments, but the data-driven serial operation cannot guarantee that the error merging of pixels in adjacent areas in low-scale images will be propagated to High-scale images. The multi-scale normalized cut method can not only effectively solve th...

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): G06T7/12G06T7/194
Inventor 伍星王森柳小勤刘韬张印辉蔡正刘畅伞红军
Owner KUNMING UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com