Object segmentation method based on hidden kernel space sparse shape expression

A technology of target segmentation and sparse representation, which is applied in image analysis, image data processing, instruments, etc., and can solve the problem of poor segmentation results of neighbor sparse representation.

Inactive Publication Date: 2016-12-21
ZHEJIANG UNIV
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] (1) The technical problem to be solved is to provide an image segmentation method involving the sparse representation of the hidden kernel space, which overcomes the problem of poor segmentation results caused by the sparse representation of the existing shape neighbors

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
  • Object segmentation method based on hidden kernel space sparse shape expression
  • Object segmentation method based on hidden kernel space sparse shape expression
  • Object segmentation method based on hidden kernel space sparse shape expression

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] As shown in the figure, this embodiment provides a sparse shape representation object segmentation method based on hidden kernel space, including the following steps:

[0044] S1: Perform KPCA processing on the training shape set;

[0045] S2: Based on the KPCA processing results, the hidden kernel shape space is established;

[0046] S3: Establish a high-level sparse representation model based on the hidden kernel shape space;

[0047] S4: Establish the underlying driving energy function based on the probability shape, and at the same time establish the dual connection item between the underlying energy and the high-level energy;

[0048] S5: Initialize the sparse coefficients and the underlying probability shape model;

[0049] S6: Calculate the dual connection item by using the sparse coefficient;

[0050] The dual connection item provided in this embodiment can be calculated by using the initialization sparse coefficient in the initialization phase, and can be ca...

Embodiment 2

[0073] In this embodiment, a two-layer segmentation model framework is established while constructing a hidden kernel shape space and a sparse representation of the hidden kernel space; the specific steps of the two-layer segmentation model framework are as follows:

[0074] A) acquiring image data and performing KPCA processing on the training shape set;

[0075] B) Based on the KPCA processing results, the hidden kernel shape space is established;

[0076] C) Build a high-level sparse representation based on the hidden kernel shape space;

[0077] D) Establish the underlying driving energy function based on the probability shape, and simultaneously establish the dual connection item between the underlying energy and the high-level energy;

[0078] E) Initialize the sparse coefficients and initialize the underlying probability shape;

[0079] F) Use the sparse coefficient to calculate the dual connection term as the constraint of the underlying function;

[0080] G) Optimi...

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 object segmentation method based on hidden kernel space sparse shape expression. The method comprises that kernel principal components are extracted from an original prior shape training set; according to a KPCA extraction result, a hidden kernel space is established, and a hidden kernel space sparse shape expression model is constructed; a sparse-coefficient-based dual constraint item and a probability-shape-based bottom-layer variation driving energy function are constructed; a target function is solved via an alternative iteration method; and a hidden kernel space sparse shape expression result is used to monitor evolution of the probability-shape-based variation energy function, and an evolution curve derived from the energy function is utilized to realize image segmentation. According to the invention, the problems that a present shape-neighbor-based sparse expression segmentation method is low in object segmentation capability and accuracy and that the sparse expression segmentation effect is not good in an original shape domain are solved.

Description

technical field [0001] The invention relates to the field of image segmentation and shape representation, in particular to an object segmentation method based on sparse representation of hidden kernel space shape. Background technique [0002] Object shape representation and segmentation is one of the core tasks in image processing and computer vision. At present, the existing methods can be generally divided into edge-based shape segmentation methods and region-based shape segmentation methods. Most of these methods define an image-based variational energy function and minimize the energy function to drive the evolution of the shape curve. After the energy function converges, a curve based on the energy function is finally used to mark the target area. This method has a good segmentation effect for the case where the target shape is relatively complete. However, when the target has defects, occlusions, or adhesion with background noise similar to the target, simply solving...

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): G06T7/00G06K9/46
CPCG06T2207/20081G06T2207/20076G06V10/40G06V10/513
Inventor 姚劲草于慧敏
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
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