CN109635815A - One kind being based on morphologic target's feature-extraction method - Google Patents
One kind being based on morphologic target's feature-extraction method Download PDFInfo
- Publication number
- CN109635815A CN109635815A CN201811360606.0A CN201811360606A CN109635815A CN 109635815 A CN109635815 A CN 109635815A CN 201811360606 A CN201811360606 A CN 201811360606A CN 109635815 A CN109635815 A CN 109635815A
- Authority
- CN
- China
- Prior art keywords
- image
- corrosion
- target
- distance
- complexity
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Length-Measuring Devices Using Wave Or Particle Radiation (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses one kind based on morphologic object edge feature extracting method, and image is generated black and white binary image;By the individual image of simple target Area generation and normalize to fixed dimension;Calculate the characteristic parameter of the original image;Determine structural element type, size and corrosion times N, preimage is corroded, calculate the ratio of major and minor axis after corrosion every time, complexity, etch away parts area with inward flange ratio, corrode mean value, the variance of back edge distance center point distance, the deflection angle of the offset of central point and front and back central point twice.Simple target uniform sizes are only being [360 by the feature that the present invention extracts, deflection when 360] by target image is influenced, morphological erosion and ratio of long axis to short axis, complexity, morphology complexity, off-centring distance, deflection angle and corrosion back edge to center distance mean value and variance be all rotational invariants, it is not influenced by target direction angle, is conducive to the identification of target.
Description
Technical field
The invention belongs to image processing techniques, and in particular to one kind is based on morphologic target's feature-extraction method.
Background technique
Feature extraction is the key technology of image recognition, and the feature of image has color characteristic, textural characteristics, geometrical characteristic etc.
Deng.The geometrical characteristic of target is common feature in image recognition.Usual target geometrical characteristic parameter mainly has: perimeter, area,
The longest axis, azimuth, boundary matrix and form factor etc..Accurately the shape of target is described be image recognition problem
One of.How the edge details of careful description target image, while having that obtain the general shape information of target be to need to continue
The problem of exploration.
Summary of the invention
Goal of the invention: it is an object of the invention to solve the deficiencies in the prior art, provides a kind of based on morphology
Target's feature-extraction method, the present invention carries out multiple morphological erosion processing to target image, utilizes the image of corrosion front and back
The long axis and short axle ratio, complexity of target after calculating original image and corrosion;Corrode the offset and deflection angle of rear center;
Morphology complexity (the corrosion of the variance of the distance of image target edge distance center and mean value and etch away parts after corrosion
Fall area divided by the inward flange length for eroding area) feature as target image.By repeatedly to target image
Corrosion is decomposed, while obtaining edge details description and the overall shape approximate description of target.
Technical solution: one kind of the invention is based on morphologic object edge feature extracting method, comprising the following steps:
(1) it generates simple target image: original image being subjected to binary conversion treatment, and makes object pixel 1, background pixel is
0, then to treated, binary image carries out empty filling, and the maximum target image of area is individually saved as figure to be analyzed
Picture, and uniform sizes (such as [360,360]);
(2) original target image A is calculated0Ratio of long axis to short axis, complexity, image object central point C0, in Edge Distance
The mean value and variance of the distance of heart point.
(3) structural element shape, size and the number of corrosion of erosion operation are set: choosing disk-like structure element, setting
The times N of corrosion, structural element are sized such that
(4) n times etching operation is carried out to above-mentioned target image using structural element, the image after n times corrosion is then respectively
A1,A2,…,AN;
(5) clarification of objective is calculated using the image of corrosion front and back: calculating A1... ..., ANLong axis and short axle ratio with
And complexity, and utilizeThe complexity of etch away parts is calculated, wherein sjAnd LjFor the area of image after jth time corrosion
And perimeter;
(6) image A after corroding is calculated1... ..., ANCorresponding center C1... ..., CNRelative to image A before corroding0Center
C0Offset distance;Utilize center line C1- C0And C2- C0... ... CN- C0Calculate the deflection angle of center line;
(7) the marginal point coordinate for calculating image after corroding calculates marginal point to center CjDistance, seek all marginal points
To the variance and mean value of central point distance;
In step (3), morphological erosion is carried out to target image, the image section eroded includes that the edge of target is thin
Information is saved, remaining part is the approximate shapes of target, the trend partial information comprising target after corrosion.Each morphological erosion
Picture breakdown is all details and approximate part by operation, and multiple etching operation realizes the multi-level decomposition to target image.
The complexity of image after corroding is calculated in step (5), and calculates the complexity of etch away parts, is conducive to mesh
Target fully describes.
The relatively original initial target image A of central point of image after corroding every time is calculated in step (6)0Central point it is inclined
The angle of the distance of shifting and the front and back deflection of corrosion centers point object original central point twice, after describing multi-level corrosion decomposition
The changes in distribution rule of image detail.
Mean value and side of the edge of image relative to the distance of current central point after calculating is corroded every time in step (7)
Difference, the shape after on the one hand describing corrosion, the variation after on the other hand describing corrosion every time.
The utility model has the advantages that the feature that the present invention extracts is only when being [360,360] by simple target uniform sizes by target figure
The influence of the deflection of picture, morphological erosion and ratio of long axis to short axis, complexity, morphology complexity, off-centring distance, deflection
Angle and corrosion back edge to center distance mean value and variance be all rotational invariants, not by the shadow at target direction angle
It rings, is conducive to the identification of target.And entire extraction process is successively to corrode target, successively seeks feature, is realized to target
Layer-by-layer careful description.
Detailed description of the invention
Fig. 1 is flow chart of the invention;
Fig. 2 is the picture of trolley to be analyzed in embodiment;
Fig. 3 is the result figure of 5 morphological erosions in embodiment;
Fig. 4 is major and minor axis ratio, complexity, morphology complexity and the corruption twice in embodiment after 5 morphological erosions
Central point deflection angle is lost as morphology decomposes the variation of number;
Fig. 5 is the distance of central point offset distance in embodiment after 5 morphological erosions, object edge distance center point
Variance and mean value.
Specific embodiment
Technical solution of the present invention is described in detail below, but protection scope of the present invention is not limited to the implementation
Example.
As shown in Figure 1, one kind of the invention is based on morphologic object edge feature extracting method, comprising the following steps:
(1) generate simple target image: image is subjected to binary conversion treatment, and makes object pixel 1, background pixel 0, to image into
The filling of row cavity, individually saves as image to be analyzed, and uniform sizes for area maximum target image;(2) the original mesh is calculated
Logo image A0Ratio of long axis to short axis, complexity, image object central point C0, Edge Distance central point distance mean value and variance.
(3) structural element shape, size and the number of corrosion of erosion operation are set: choosing disk-like structure element, sets time of corrosion
Number N, structural element are sized such that(4) etching operation, n times corrosion are carried out to target image using structural element
Image afterwards is then respectively A1,A2,…,AN;(5) clarification of objective is calculated using the image of corrosion front and back: calculating A1... ..., AN
Long axis and short axle ratio and complexity;Calculate image A after corroding1... ..., ANCenter C1... ..., CNRelative to corruption
Image A before losing0Center C0Offset distance;Utilize center line C1- C0And C2- C0... ... CN- C0Calculate the deflection of center line
Angle;(6) the marginal point coordinate for calculating image after corroding calculates marginal point to center CjDistance, seek all marginal points and arrive
The variance and mean value of central point distance;(7) it utilizesThe complexity of etch away parts is calculated, wherein sjAnd LjFor jth time
The area and perimeter of image after corrosion, the value range of j are 1~N.
Embodiment:
The present embodiment is realized in matlab R2016b software.
Image will be subjected to binary conversion treatment first, and make object pixel 1, background pixel 0 carries out cavity to image
Filling finds area maximum target image using bwlabel function combination regionprops function and individually saves as figure to be analyzed
Picture, and uniform sizes [360,360];Disk-like structure element is chosen, sets the times N of corrosion, structural element is sized such that
360/(2*(N+2));Etching operation is carried out to target image using structural element, which is denoted as A0.It reuses
Bwlabel function combination regionprops function seeks area, major and minor axis, center, the marginal point of original object;Utilize area
Complexity is sought with edge length, seeks ratio of long axis to short axis using length shaft length, is sat using the coordinate and central point of marginal point
Mark the mean value and variance of the distance for calculating marginal point to center.First time corrosion is carried out to image, image is denoted as A after corrosion1, together
Sample seeks area, major and minor axis, center, the marginal point of image after corrosion using bwlabel function combination regionprops, equally
The mean value and variance of computation complexity ratio of long axis to short axis and marginal point distance center, it is another to calculateIt is defined as etch away parts
Morphology complexity, wherein S0For A0Area, S1And L1For A1Area and perimeter.Calculate A1Center C1And A0Center C0
Distance, the most corrode rear center offset.C is calculated simultaneously1And C0The deflection of two o'clock line.Corrosometer is counted in stating again
Characteristic quantity separately calculates C1And C0Deflection and C2And C0Deflection deflection angle of the difference as target's center.According to setting
The number of fixed corrosion, above-mentioned whole features after gradually calculating corrosion.
The present invention repeatedly corrodes target image using structural element, to image after corrosion and the image eroded point
Complexity value is not sought, and seeks the long axis of corrosion front and back and the ratio of short axle as clarification of objective, contains target image
Detail edges information and approximate trend term information calculate the offset and deflection angle of corrosion rear center's point every time, description corrosion
Fall the Mass Distribution of part, calculate corrosion back edge to center distance mean value and variance, describe the big of image after corroding
Small and complexity.Feature after calculating corrosion every time realizes more complete shape description, conducive to the identification of target.Successively divide
The variation of parameter is also the index of reflection target shape after solution.
In order to verify its feasibility, the present embodiment is analyzed by taking Fig. 2 automobile as an example.The result of 5 morphological erosions
See Fig. 3.Major and minor axis ratio, complexity, morphology complexity after 5 morphological erosions and twice corrosion centers point deflection angle
As Fig. 4 is shown in the variation that morphology decomposes number, the central point offset distance, object edge distance center point after morphological erosion
Distance variance and mean value see Fig. 5.The invention mainly relates to morphology operations and connected region to solve two class operations, takes into account spy
The requirement of sign description accuracy rate and real-time, the needs being able to satisfy under certain condition.
Claims (1)
1. one kind is based on morphologic object edge feature extracting method, it is characterised in that: the following steps are included:
(1) it generates simple target image: original image being subjected to binary conversion treatment, and makes object pixel 1, background pixel 0, so
To treated, binary image carries out empty filling afterwards, and the maximum target image of area is individually saved as image to be analyzed,
And uniform sizes;
(2) original target image A is calculated0Ratio of long axis to short axis, complexity, image object central point C0, Edge Distance central point
Distance mean value and variance;
(3) structural element shape, size and the number of corrosion of erosion operation are set: choosing disk-like structure element, setting corrosion
Times N, structural element is sized such that
(4) n times etching operation is carried out to above-mentioned target image using structural element, the image after n times corrosion is then respectively A1,
A2,…,AN;
(5) clarification of objective is calculated using the image after corrosion every time: calculating A1... ..., ANLong axis and short axle ratio and
Target complexity;Calculate image A after corroding1... ..., ANCorresponding center C1... ..., CNRelative to image A before corroding0Center
C0Offset distance;Utilize center line C1-C0And C2-C0... ... CN-C0Calculate the deflection angle of center line;Scheme after calculating corrosion
The marginal point coordinate of picture calculates marginal point to center CjDistance, seek all marginal points to central point distance variance and
Value;It utilizesThe complexity of etch away parts is calculated, wherein sjAnd LjFor the area and perimeter of image after jth time corrosion.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811360606.0A CN109635815B (en) | 2018-11-15 | 2018-11-15 | Target feature extraction method based on morphology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811360606.0A CN109635815B (en) | 2018-11-15 | 2018-11-15 | Target feature extraction method based on morphology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109635815A true CN109635815A (en) | 2019-04-16 |
CN109635815B CN109635815B (en) | 2023-04-28 |
Family
ID=66068161
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811360606.0A Active CN109635815B (en) | 2018-11-15 | 2018-11-15 | Target feature extraction method based on morphology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109635815B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079613A (en) * | 2019-12-09 | 2020-04-28 | 北京明略软件***有限公司 | Gesture recognition method and apparatus, electronic device, and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105374045A (en) * | 2015-12-07 | 2016-03-02 | 湖南科技大学 | Morphology-based image specific shape dimension objet rapid segmentation method |
CN106327522A (en) * | 2016-08-25 | 2017-01-11 | 上海航天控制技术研究所 | Infrared small target detection method based on multi-direction morphological filtering complex cloud background |
CN108734716A (en) * | 2018-04-21 | 2018-11-02 | 卞家福 | A kind of fire complex environment image detecting method based on improvement Prewitt operators |
-
2018
- 2018-11-15 CN CN201811360606.0A patent/CN109635815B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105374045A (en) * | 2015-12-07 | 2016-03-02 | 湖南科技大学 | Morphology-based image specific shape dimension objet rapid segmentation method |
CN106327522A (en) * | 2016-08-25 | 2017-01-11 | 上海航天控制技术研究所 | Infrared small target detection method based on multi-direction morphological filtering complex cloud background |
CN108734716A (en) * | 2018-04-21 | 2018-11-02 | 卞家福 | A kind of fire complex environment image detecting method based on improvement Prewitt operators |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079613A (en) * | 2019-12-09 | 2020-04-28 | 北京明略软件***有限公司 | Gesture recognition method and apparatus, electronic device, and storage medium |
CN111079613B (en) * | 2019-12-09 | 2023-11-03 | 北京明略软件***有限公司 | Gesture recognition method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109635815B (en) | 2023-04-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101398894B (en) | Automobile license plate automatic recognition method and implementing device thereof | |
CN107564017B (en) | Method for detecting and segmenting urban high-resolution remote sensing image shadow | |
CN109285162A (en) | A kind of image, semantic dividing method based on regional area conditional random field models | |
CN107230214B (en) | SAR image water area automatic detection method based on recursive OTSU algorithm | |
CN104217436B (en) | SAR image segmentation method based on multiple features combining sparse graph | |
CN104598922B (en) | Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm | |
CN107358619A (en) | Multi-Level Threshold Image Segmentation method based on chicken group's optimization | |
CN109635815A (en) | One kind being based on morphologic target's feature-extraction method | |
CN111161300B (en) | Niblack image segmentation method based on improved Otsu method | |
CN115409831A (en) | Star point centroid extraction method and system based on optimal background estimation | |
CN110633749B (en) | Three-dimensional point cloud identification method based on improved viewpoint feature histogram | |
CN113341388B (en) | Rapid implementation method for radar target two-dimensional CA-CFAR detection based on step-by-step processing | |
CN104574352A (en) | Crowd density grade classification method based on foreground image | |
CN109446882A (en) | Logo feature extraction and recognition methods based on the characteristic quantification that gradient direction divides | |
CN103218820A (en) | Camera calibration error compensation method based on multi-dimensional characteristics | |
CN102800101A (en) | Satellite-borne infrared remote sensing image airport ROI rapid detection method | |
CN113344047A (en) | Platen state identification method based on improved K-means algorithm | |
CN102005058B (en) | Rapid implementation method aiming at OPTA (One-Pass Thinning Algorithm) of image | |
CN112819832A (en) | Urban scene semantic segmentation fine-grained boundary extraction method based on laser point cloud | |
CN109829511B (en) | Texture classification-based method for detecting cloud layer area in downward-looking infrared image | |
CN104123553A (en) | License plate positioning method and system based on cascading morphological transformation | |
CN103793913A (en) | Spectral clustering image segmenting method combined with mean shift | |
CN113920391B (en) | Target counting method based on scale generation self-adaptive truth diagram | |
CN110827309A (en) | Polaroid appearance defect segmentation method based on super-pixels | |
CN110309793A (en) | A kind of SAR target identification method based on video bits layering interpretation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |