CN105825528B - A kind of image symmetrical shaft detection method - Google Patents

A kind of image symmetrical shaft detection method Download PDF

Info

Publication number
CN105825528B
CN105825528B CN201610146362.0A CN201610146362A CN105825528B CN 105825528 B CN105825528 B CN 105825528B CN 201610146362 A CN201610146362 A CN 201610146362A CN 105825528 B CN105825528 B CN 105825528B
Authority
CN
China
Prior art keywords
point
image
parameter
symmetry axis
straight line
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201610146362.0A
Other languages
Chinese (zh)
Other versions
CN105825528A (en
Inventor
修春波
巴富珊
王甜甜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin Polytechnic University
Original Assignee
Tianjin Polytechnic University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tianjin Polytechnic University filed Critical Tianjin Polytechnic University
Priority to CN201610146362.0A priority Critical patent/CN105825528B/en
Publication of CN105825528A publication Critical patent/CN105825528A/en
Application granted granted Critical
Publication of CN105825528B publication Critical patent/CN105825528B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image

Landscapes

  • Image Analysis (AREA)

Abstract

The invention belongs to image procossing and image understanding field, specially a kind of image symmetrical shaft detection method.Characteristic point detection is carried out to image, the cluster of feature point set is realized using clustering method.In each feature point set repeatedly put to extracting using the mode of random sampling, the Kernal Equations of candidate symmetry axis parameter are established to the parameter of perpendicular bisector according to point, the parameter of the peak point correspondence image symmetry axis of total kernel function, is achieved in the automatic detection of image symmetrical axis.The method of the present invention is easy, result is accurate, suitable for image understanding field.

Description

A kind of image symmetrical shaft detection method
Technical field
The invention belongs to image procossings and image understanding field, are related to a kind of image symmetrical shaft detection method, more particularly to A kind of image symmetrical shaft detection method based on cluster analysis.
Background technology
Symmetry is one of principal shape attribute of image object.Symmetry Detection image understanding, target identification and There is important application value in the fields such as three-dimension object reconstruct.The feature for determining to assist realizing image of symmetry characteristic carries It takes, the functions such as object detection and identification.
At present, the detection of image symmetrical characteristic is widely used in image procossing, to identification and positioning, the scene understanding of object Etc. significant.In image retrieval, using object symmetry, efficient compression and coding can be carried out to digital picture, So that the storage of image is more quick with transmitting；In biomedical sector, the symmetry of image provides weight for medical diagnosis The parameter guidance wanted.With deepening continuously for symmetrical Journal of Sex Research, Symmetry Detection is led in intelligent transportation, three-dimensional reconstruction, robot The fields such as boat, remote sensing image processing possess broader practice prospect.
Common symmetry detection methods have the mirror symmetry detection method based on pattern matching method, pair based on phase information Title property detection method and the detection method based on curve differential calculus property etc..But existing method usually exists to noise-sensitive, place There are the limitations such as large error, computationally intensive for reason process.
Therefore, designing a kind of symmetry axis detection method being simple and efficient has good application value.
Invention content
The technical problems to be solved by the invention are that being existed according to the both sides of image symmetrical axis largely has same characteristic features Point set designs a kind of image symmetrical shaft detection method based on cluster analysis.
The technical solution adopted in the present invention is：A kind of image symmetrical shaft detection method carries out characteristic point detection to image, Detection zone is limited to the edge of target, and the cluster of feature point set is realized using clustering method.Utilize random sampling Mode carries out repeatedly putting to extracting in each feature point set, and candidate symmetry axis ginseng is established to the parameter of perpendicular bisector according to point Several Kernal Equations, the parameter of the peak point correspondence image symmetry axis of total kernel function, are achieved in the automatic of image symmetrical axis Detection.
It is an object of the invention to design a kind of image symmetrical shaft detection method based on cluster analysis, be capable of detecting when to Determine the symmetry axis of image, there is good practicability.
Description of the drawings
Fig. 1 is original image.
Fig. 2 is edge detection results figure.
Fig. 3 is Corner Detection result figure.
Fig. 4 is characteristic point Neighborhood Graph.
Fig. 5 is evaluation index figure.
Fig. 6 is cluster result figure when classifying number C=8.
Fig. 7 is symmetry axis and frame intersection point relational graph.
Fig. 8 is symmetry axis testing result figure.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
The present invention is using the angle point in Harris angular-point detection method detection images.To reduce operand, first using Sobel Operator carries out edge extracting to image, then carries out angle point differentiation for marginal point.Fig. 1 is given original image, and Fig. 2 is uses Sobel operators carry out edge extracting acquired results.
Angle point is extracted using Harris angular-point detection methods to each point on image border, Fig. 3 is Corner Detection result.
Using the angle point detected as characteristic point.If the feature point coordinates detected is (i, j), gray value is h (i, j), Fig. 4 provides 3 × 3 Neighborhood Graphs of characteristic point (i, j).
Neighborhood gray scale difference value H (i, j) at characteristic point is calculated using formula (1)：
The textural characteristics L of characteristic point (i, j) is calculated using LBP (Local binary patterns) texture blending method (i, j).By at characteristic point grey scale pixel value, the feature vector of neighborhood gray scale difference value and texture eigenvalue composition characteristic point Xij：
Xij=[h (i, j) H (i, j) L (i, j)]T (2)
Wherein, " T " is transposition symbol.
The cluster analysis of all characteristic points, and the Cluster Assessment index J being defined as follows are realized using C means clustering methods：
Wherein, SWFor scatter matrix in total class, SBThe scatter matrix between class chooses different classification number C values, it is equal that C is respectively adopted It is worth the cluster analysis that clustering method realizes characteristic point, Cluster Assessment index J is taken into cluster result during maximum as final spy Levy cluster result.
Fig. 5 Cluster Assessment index relationships obtained by different C values.As it can be seen that when number of classifying is 8, Cluster Assessment index J reaches Maximum value, Fig. 6 provide cluster result during classification number C=8.As seen from Figure 6, most of angle point can correctly classify, but also deposit In the angle point of mistake classification, that is, there is a situation where that partial symmetry angle point is not clustered into one kind, but this kind of angle point negligible amounts.
It is all kinds of it is middle carry out multiple random sampling respectively, a pair of of characteristic point is chosen in sampling every time, will sample and be obtained every time The perpendicular bisectors of 2 characteristic points be determined as a candidate symmetry axis.The candidate symmetry axis of one of image is straight line, In order to determine a candidate symmetry axis, it is thus necessary to determine that the angle of inclination of a point and candidate symmetry axis on candidate symmetry axis.By It will necessarily meet at 2 points with four frames of image in candidate symmetry axis, as shown in fig. 7, Fig. 7 midpoints (a(i) 1, b(i) 1) and point (a(i) 2, b(i) 2) represent candidate symmetry axis i and image frame intersection point.It can be acquired in two intersection points using formula (4) and formula (5) The transverse and longitudinal coordinate of heart point：
If the inclination angle of candidate symmetry axis i is θ(w) 0, then defining i-th of kernel function is：
Wherein, σ1、σ2、σ3For kernel functional parameter.A, b and θ is the parameter of any bar straight line, i.e., (a, b) is one on straight line Point, θ are the inclination angle of straight line.Total kernel function that all candidate's symmetry axis are formed is F：
Make straight line parameter (a during total kernel function F acquirements maximum value0, b0, θ0) determined by straight line be image master couple Claim axis.Fig. 8 gives final symmetry axis testing result.
It is an advantage of the current invention that symmetry axis detection method is applicable not only to the symmetrical figure of horizontal direction, incline for having The symmetric figure of rake angle is equally applicable, and this method is easy, result is accurate.The present invention is suitable for image understanding field.

Claims (1)

  1. A kind of 1. image symmetrical shaft detection method, which is characterized in that characteristic point detection is carried out to image, using clustering method It realizes the cluster of feature point set, in each feature point set repeatedly put to extracting using the mode of random sampling, according to point The Kernal Equations of candidate symmetry axis parameter are established to the parameter of perpendicular bisector, the peak point correspondence image of total kernel function is symmetrical The parameter of axis；It, will be special using the angle point detected as characteristic point using the angle point in Harris angular-point detection method detection images The feature vector of grey scale pixel value, neighborhood gray scale difference value and texture eigenvalue composition characteristic point at sign point, is gathered using C mean values Class method realizes the cluster analysis of characteristic point, is maximum cluster result as final cluster result using Cluster Assessment index J, Wherein,
    SWFor scatter matrix in total class, SBThe scatter matrix between class；According to the parameter of the perpendicular bisector of random sampling point pair, establish The Kernal Equations of candidate symmetry axis parameter be：
    Wherein, σ1、σ2、σ3For kernel functional parameter, a(i) 0, b(i) 0, θ(i) 0The transverse and longitudinal coordinate of any on respectively candidate symmetry axis i And the inclination angle of candidate symmetry axis i, a, b and θ are the parameter of any bar straight line, i.e., (a, b) is a bit on straight line, and θ is straight line Inclination angle, total kernel function that all candidate's symmetry axis are formed is F：
    Make straight line parameter (a during total kernel function F acquirements maximum value0, b0, θ0) determined by straight line be image main symmetry axis.
CN201610146362.0A 2016-03-14 2016-03-14 A kind of image symmetrical shaft detection method Expired - Fee Related CN105825528B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610146362.0A CN105825528B (en) 2016-03-14 2016-03-14 A kind of image symmetrical shaft detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610146362.0A CN105825528B (en) 2016-03-14 2016-03-14 A kind of image symmetrical shaft detection method

Publications (2)

Publication Number Publication Date
CN105825528A CN105825528A (en) 2016-08-03
CN105825528B true CN105825528B (en) 2018-07-10

Family

ID=56987847

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610146362.0A Expired - Fee Related CN105825528B (en) 2016-03-14 2016-03-14 A kind of image symmetrical shaft detection method

Country Status (1)

Country Link
CN (1) CN105825528B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106780528B (en) * 2016-12-01 2019-09-17 广西赛联信息科技股份有限公司 Image symmetrical shaft detection method based on edge matching

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4182665B2 (en) * 2002-02-08 2008-11-19 日産自動車株式会社 Approaching object detection device
CN103559494A (en) * 2013-10-30 2014-02-05 中国矿业大学(北京) Method for detecting symmetry axis of plane figure object
CN105184830A (en) * 2015-08-28 2015-12-23 华中科技大学 Symmetry image symmetric axis detection positioning method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4182665B2 (en) * 2002-02-08 2008-11-19 日産自動車株式会社 Approaching object detection device
CN103559494A (en) * 2013-10-30 2014-02-05 中国矿业大学(北京) Method for detecting symmetry axis of plane figure object
CN105184830A (en) * 2015-08-28 2015-12-23 华中科技大学 Symmetry image symmetric axis detection positioning method

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Fast Global Reflectional Symmetry Detection for Robotic Grasping and Visual Tracking;Wai Ho Li et al.;《Australasian Conference on Robotics》;20141210;全文 *
Robust image Segmentation Using FCM with spatial constraints based on new kernel-induced distance measure;Songcan Chen et al.;《IEEE Transactions on Systems, Man, and Cybernetics, Part B》;20040831;第34卷(第4期);全文 *
具有对称性的凸边界图形对象识别与图像配准研究;洪亮;《中国博士学位论文全文数据库 信息科技辑》;20141115(第11期);全文 *
基于规则和对称分析的眼睛定位方法;李金生 等;《微计算机信息》;20091231;第25卷(第5-3期);全文 *
基于角点特征的粘连车辆分割方法;陈舒 等;《计算机工程》;20130430;第39卷(第4期);第3节 *
大规模图像集自动摘要技术研究;赵烨;《中国博士学位论文全文数据库 信息科技辑》;20150715(第7期);全文 *

Also Published As

Publication number Publication date
CN105825528A (en) 2016-08-03

Similar Documents

Publication Publication Date Title
CN103325112B (en) Moving target method for quick in dynamic scene
Jin et al. 3D fast automatic segmentation of kidney based on modified AAM and random forest
Sharma et al. A review on image segmentation with its clustering techniques
CN105574527B (en) A kind of quick object detecting method based on local feature learning
EP2079054B1 (en) Detection of blobs in images
CN107239777B (en) Tableware detection and identification method based on multi-view graph model
Shahin et al. A novel white blood cells segmentation algorithm based on adaptive neutrosophic similarity score
CN104123554B (en) SIFT image characteristic extracting methods based on MMTD
CN109975196B (en) Reticulocyte detection method and system
Wu et al. Image segmentation
CN110490913A (en) Feature based on angle point and the marshalling of single line section describes operator and carries out image matching method
CN109035196A (en) Saliency-Based Image Local Blur Detection Method
CN108280469A (en) A kind of supermarket's commodity image recognition methods based on rarefaction representation
CN108182705A (en) A kind of three-dimensional coordinate localization method based on machine vision
Perumal et al. Multilevel morphological fuzzy edge detection for color images (MMFED)
CN110910497B (en) Method and system for realizing augmented reality map
Kovacs et al. Orientation based building outline extraction in aerial images
CN105825528B (en) A kind of image symmetrical shaft detection method
CN115965613A (en) Cross-layer connection construction scene crowd counting method based on cavity convolution
Wei et al. An algorithm for segmentation of lung ROI by mean-shift clustering combined with multi-scale HESSIAN matrix dot filtering
Gim et al. A novel framework for white blood cell segmentation based on stepwise rules and morphological features
Sahu et al. Digital image texture classification and detection using radon transform
Li et al. A novel edge detector for color images based on MCDM with evidential reasoning
CN113205540A (en) Multi-scale automatic anisotropic morphological direction derivative edge detection method
Wang et al. Database of human segmented images and its application in boundary detection

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20180710

Termination date: 20210314

CF01 Termination of patent right due to non-payment of annual fee