CN110210491A - Rapid image feature extracting method based on FAST and SIFT - Google Patents
Rapid image feature extracting method based on FAST and SIFT Download PDFInfo
- Publication number
- CN110210491A CN110210491A CN201910333318.4A CN201910333318A CN110210491A CN 110210491 A CN110210491 A CN 110210491A CN 201910333318 A CN201910333318 A CN 201910333318A CN 110210491 A CN110210491 A CN 110210491A
- Authority
- CN
- China
- Prior art keywords
- point
- characteristic point
- sift
- characteristic
- fast
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- 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
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20164—Salient point detection; Corner detection
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of methods that the quick SIFT feature of image may be implemented and extract.Since traditional SIFT method needs to generate image difference of Gaussian pyramid, lead to that computationally intensive, time-consuming.The present invention, which is got around, detects this step according to image difference of Gaussian pyramid progress extreme point, then uses FAST corner detection method, is rapidly and efficiently extracted to the characteristic point of image.It combines SIFT method to carry out the calculating of principal direction to characteristic point later and generates SIFT descriptor, while guaranteeing that rotational invariance, feature point description symbol are reliable and stable, realize the quick of image characteristic point, high efficiency extraction and description.
Description
Technical field
The invention belongs to computer vision fields, more particularly to the rapidly extracting of image characteristic point, image registration, image
The fields such as identification.
Background technique
SIFT (Scale invariant features transform) algorithm is the generally acknowledged strongest feature extracting method of characteristic point comprehensive performance,
Have the invariance such as scaling, translation, scale, stability is strong with noise immunity, is usually used in handling between two images in the presence of translation, rotation
Turn, the registration under the conditions of affine transformation etc..But this method has shortcoming again, that is, due to the structure of Gauss scale pyramid
It builds and the processes operands such as the descriptor computation of characteristic point is big, leading to the algorithm, time-consuming.Existing improvement is mostly for such as
What improves SIFT feature extraction rate, while realizing the promotion of comprehensive performance as far as possible.
Summary of the invention
The purpose of the present invention is to provide a kind of reliable methods that quick SIFT feature may be implemented and extract.Due to SIFT point
Detection is to be detected based on difference of Gaussian spatial extrema, and the building operand of difference of Gaussian image pyramid is very big, consumes very much
When, leading to total algorithm, time-consuming.In order to improve algorithm speed, the present invention around be opened in difference of Gaussian space carry out extremum extracting this
One step carries out characteristic point detection to picture subject to registration using FAST angular-point detection method first, quickly obtains the position of characteristic point
Information.The sub- generation strategy of description for combining SIFT later, generates 128 dimensional feature description vectors around extreme point.By FAST point
Detection method is combined with SIFT method, realizes quick, real-time image characteristic point detection and description.
It realizes the technical solution of the present invention is as follows: the rapid image feature extracting method based on FAST and SIFT, including following
Step:
Step 1: angle point is extracted according to FAST angle point grid step.
Step 2: selecting a certain size neighborhood window, calculates characteristic point surrounding neighbors histogram of gradients, and is characterized a little minute
With principal direction θ.
Step 3: it is calculated according to 16*16 neighborhood territory pixel around key point and generates SIFT feature descriptor.
In step 1, feature point extraction is carried out to image subject to registration using FAST, obtains specifically doing for characteristic point position
Method are as follows:
Centered on pixel p, radius is to have 16 on the bresenham circle of the discretization of three pixels for definition first
Pixel (p1, p2 ..., p16).
A suitable threshold value is defined later, calculates p1, p5, p9, the pixel absolute value of the difference of p13 and center p, if absolutely
Having at least three in value is more than threshold value, then is not otherwise angle point as candidate angular.
If p is candidate angular, calculate this 16 points of p1 to p16 to center p pixel absolute value of the difference, if in absolute value
Having at least continuous n is more than threshold value, then is angle point, is not angle point otherwise, n can be 9 or 12.
Non-maxima suppression finally is carried out to all characteristic points, that is, is directed to some characteristic point, is calculated including itself
3 × 3 or 5 × 5 neighborhoods in characteristic point quantity, if quantity be 1, i.e., in neighborhood only have this characteristic point, then retain
This characteristic point calculates the FAST score value of each characteristic point, that is, it is right with it to calculate each characteristic point if quantity is greater than 1
The sum of upper 16 pixels pixel value absolute value of the difference of bresenham circle answered, retains the maximum characteristic point of FAST score, remaining
Feature point deletion.
In step 2, a certain size neighborhood window is selected, calculates characteristic point surrounding neighbors histogram of gradients, and is spy
The specific practice of sign point distribution principal direction θ are as follows:
Coordinate is the point of (x, y), and mould and direction are expressed as follows:
M (x, y)={ [L (x+1, y)-L (x-1, y)]2+[L(x,y+1)-L(x,y-1)]2}1/2
Wherein L (x, y) represents image subject to registration.
It is sampled in the neighborhood window centered on characteristic point later, with the ladder of gradient orientation histogram statistics neighborhood territory pixel
Direction is spent, the histogram of gradients within the scope of 0 °~360 ° is averagely divided into 36 columns, every 10 degree of columns.The peak of histogram of gradients
Value corresponding angle represents the principal direction of characteristic point, and when the ratio between minor peaks and main peak value are more than 80%, this direction is considered
The auxiliary direction of this feature point.One characteristic point may be designated with multiple directions, to enhance robustness.
In step 3, is calculated according to 16*16 neighborhood around key point and generates SIFT feature descriptor, specific practice are as follows:
First centered on characteristic point, reference axis is rotated to be to the principal direction of characteristic point, i.e., in characteristic point neighbors around
By reference axis rotate θ angle (θ is characterized principal direction), after rotation in neighborhood pixel new coordinate are as follows:
Later centered on characteristic point, the pixel of 16 × 16 neighborhoods around it is evenly dividing as 4 × 4 fritters, every
The histogram of gradients that eight directions are drawn on a fritter, forms seed point.Therefore the descriptor of each characteristic point is by 16 seeds
Point is formed, and each seed point has the information in 8 directions, to obtain the feature point description symbol of 4 × 4 × 8=128 dimension.
Detailed description of the invention:
Fig. 1 is flow chart of the present invention.
Fig. 2 is FAST corner detection schematic diagram.
Claims (2)
1. the rapid image feature extracting method based on FAST and SIFT, which comprises the following steps:
Step 1: the characteristic point of image is extracted according to FAST angle point grid step.
Step 2: selecting a certain size neighborhood window, calculates characteristic point surrounding neighbors histogram of gradients, and is characterized a distribution master
Direction θ.
Step 3: it is calculated according to 16*16 neighborhood around key point and generates SIFT feature descriptor.
2. the rapid image feature extracting method according to claim 1 based on FAST and SIFT, which is characterized in that make
Characteristic point detection is carried out to image subject to registration with FAST point detecting method, instead of in traditional SIFT method in difference of Gaussian image
Extreme point is detected in pyramid, is greatly reduced characteristic point and is detected the calculation amount of this step, it can be achieved that real time characteristic points detect.Later
The calculating of principal direction is carried out to characteristic point in conjunction with SIFT method and generates SIFT feature descriptor, thus guaranteeing feature point description
While according with reliable and stable, quick, efficient image feature point extraction and description are realized.Whole process technical requirements are as follows:
Centered on pixel p, radius is to have 16 pixels on the bresenham circle of the discretization of three pixels for definition first
Point (p1, p2 ..., p16).
A suitable threshold value is defined later, calculates p1, p5, p9, the pixel absolute value of the difference of p13 and center p, if in absolute value
Having at least three is more than threshold value, then is not otherwise angle point as candidate angular.
If p is candidate angular, calculate this 16 points of p1 to p16 to center p pixel absolute value of the difference, if having in absolute value to
Few continuous n are more than threshold value, then are angle points, are not angle point otherwise, n can be 9 or 12.
Then to all characteristic points carry out non-maxima suppression, that is, be directed to some characteristic point, calculate including itself 3 ×
The quantity of characteristic point in 3 or 5 × 5 neighborhoods only has this characteristic point that is, in the neighborhood, then retains this if quantity is 1
Characteristic point calculates the FAST score value of each characteristic point, that is, it is corresponding to calculate each characteristic point if quantity is greater than 1
The sum of upper 16 pixels pixel value absolute value of the difference of bresenham circle, retains the maximum characteristic point of FAST score, remaining feature
Point deletion.
Then a certain size neighborhood window is selected, characteristic point surrounding neighbors histogram of gradients is calculated, and is characterized a distribution master
Direction θ, specific practice are as follows:
Coordinate is the point of (x, y), and mould and direction are expressed as follows:
M (x, y)={ [L (x+1, y)-L (x-1, y)]2+[L(x,y+1)-L(x,y-1)]2}1/2
Wherein L (x, y) represents image subject to registration.
It is sampled in the neighborhood window centered on characteristic point, the gradient direction of neighborhood territory pixel is counted with gradient orientation histogram,
Histogram of gradients within the scope of 0 °~360 ° is averagely divided into 36 columns, every 10 degree of columns.The peak value of histogram of gradients is corresponding
Angle represents the principal direction of characteristic point, and when the ratio between minor peaks and main peak value are more than 80%, this direction is considered this feature
The auxiliary direction of point.One characteristic point may be designated with multiple directions, to enhance robustness.
Finally centered on characteristic point, reference axis is rotated to be to the principal direction of characteristic point, i.e., will be sat in characteristic point neighbors around
Parameter rotate θ angle (θ is characterized principal direction), after rotation in neighborhood pixel new coordinate are as follows:
Then centered on characteristic point, the pixel of 16 × 16 neighborhoods around it is evenly dividing as 4 × 4 fritters, each small
The histogram of gradients that eight directions are drawn on block, forms seed point.Therefore the descriptor of each characteristic point is by 16 seed dots
At each seed point has the information in 8 directions, to obtain the feature point description symbol of 4 × 4 × 8=128 dimension.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910333318.4A CN110210491A (en) | 2019-04-24 | 2019-04-24 | Rapid image feature extracting method based on FAST and SIFT |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910333318.4A CN110210491A (en) | 2019-04-24 | 2019-04-24 | Rapid image feature extracting method based on FAST and SIFT |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110210491A true CN110210491A (en) | 2019-09-06 |
Family
ID=67786230
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910333318.4A Pending CN110210491A (en) | 2019-04-24 | 2019-04-24 | Rapid image feature extracting method based on FAST and SIFT |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110210491A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106485740A (en) * | 2016-10-12 | 2017-03-08 | 武汉大学 | A kind of combination point of safes and the multidate SAR image registration method of characteristic point |
CN106558072A (en) * | 2016-11-22 | 2017-04-05 | 重庆信科设计有限公司 | A kind of method based on SIFT feature registration on remote sensing images is improved |
CN108921175A (en) * | 2018-06-06 | 2018-11-30 | 西南石油大学 | One kind being based on the improved SIFT method for registering images of FAST |
-
2019
- 2019-04-24 CN CN201910333318.4A patent/CN110210491A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106485740A (en) * | 2016-10-12 | 2017-03-08 | 武汉大学 | A kind of combination point of safes and the multidate SAR image registration method of characteristic point |
CN106558072A (en) * | 2016-11-22 | 2017-04-05 | 重庆信科设计有限公司 | A kind of method based on SIFT feature registration on remote sensing images is improved |
CN108921175A (en) * | 2018-06-06 | 2018-11-30 | 西南石油大学 | One kind being based on the improved SIFT method for registering images of FAST |
Non-Patent Citations (3)
Title |
---|
常旭剑: "基于FAST检测及SIFT描述的特征检测算法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
蒋少华: "《多源图像处理技术》", 31 July 2012 * |
陈慧岩: "《智能车辆理论与应用》", 31 July 2018 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106960451B (en) | Method for increasing number of feature points of image weak texture area | |
CN108256394B (en) | Target tracking method based on contour gradient | |
EP1835460B1 (en) | Image processing system, learning device and method, and program | |
CN110334762B (en) | Feature matching method based on quad tree combined with ORB and SIFT | |
CN106709500B (en) | Image feature matching method | |
CN104933434A (en) | Image matching method combining length between perpendiculars (LBP) feature extraction method and surf feature extraction method | |
Uchiyama et al. | Toward augmenting everything: Detecting and tracking geometrical features on planar objects | |
CN106408597A (en) | Neighborhood entropy and consistency detection-based SAR (synthetic aperture radar) image registration method | |
CN110110608B (en) | Forklift speed monitoring method and system based on vision under panoramic monitoring | |
Bai et al. | Scene text localization using gradient local correlation | |
CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
Ram et al. | Analysis of Harris corner detection for color images | |
Liang et al. | Robust sea-sky-line detection for complex sea background | |
CN108319961B (en) | Image ROI rapid detection method based on local feature points | |
CN114331879A (en) | Visible light and infrared image registration method for equalized second-order gradient histogram descriptor | |
CN103533332B (en) | A kind of 2D video turns the image processing method of 3D video | |
CN103336964B (en) | SIFT image matching method based on module value difference mirror image invariant property | |
CN110197184A (en) | A kind of rapid image SIFT extracting method based on Fourier transformation | |
CN112418210B (en) | Intelligent classification method for tower inspection information | |
CN112232249A (en) | Remote sensing image change detection method and device based on depth features | |
CN104408430B (en) | License plate positioning method and device | |
Zou et al. | Geodesic-based pavement shadow removal revisited | |
CN109544608B (en) | Unmanned aerial vehicle image acquisition characteristic registration method | |
CN110210491A (en) | Rapid image feature extracting method based on FAST and SIFT | |
CN110210511A (en) | A kind of improvement PCA-SIFT method for registering images based on cosine measure |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |