CN108010045A - Visual pattern characteristic point error hiding method of purification based on ORB - Google Patents
Visual pattern characteristic point error hiding method of purification based on ORB Download PDFInfo
- Publication number
- CN108010045A CN108010045A CN201711297537.9A CN201711297537A CN108010045A CN 108010045 A CN108010045 A CN 108010045A CN 201711297537 A CN201711297537 A CN 201711297537A CN 108010045 A CN108010045 A CN 108010045A
- Authority
- CN
- China
- Prior art keywords
- characteristic point
- point
- matching
- feature
- orb
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
-
- 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
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
Abstract
The present invention relates to a kind of visual pattern characteristic point error hiding method of purification based on ORB, it comprises the following steps:Read the left and right two shot under different viewing angles to be detected and open image, structure scale image pyramid is carried out to image, and carry out grid processing;The detection of characteristic point, the characteristic point extracted are carried out in each small grid of every layer of pyramid diagram picture, and determines feature point coordinates;The characteristic point close to image border is removed, and calculates the barycenter direction of residue character point;Calculate ORB feature point descriptions;Characteristic point on two images is slightly matched;Screen characteristic point slightly matching pair;To error hiding to rejecting again;RANSAC algorithm iterations are carried out to remaining matching double points, and export the matching image after purification.The invention enables the characteristic point detected to be evenly distributed, and avoids characteristic point caused by multiple characteristic points are easily crowded together and clusters effect, and matched speed is also improved while improving matching accuracy.
Description
Technical field
The invention belongs to computer vision, image processing field, more particularly to a kind of visual pattern characteristic point based on ORB
Error hiding method of purification.
Background technology
Feature detection, description and matching technique be realize image co-registration, image flame detection, image mosaic and target identification with
One of committed step of tracking, and image procossing, the one of machine vision navigation field big research hotspot.Image recognition, video
The realization of many technologies such as tracking, image mosaic, three-dimensional reconstruction, will rely on detection, description and the matching of image characteristic point.It is high
The images match of accuracy is to determine the condition and key of robot motion, in the vision SLAM of distinguished point based, feature
It is low with often there are many error hiding information, causing to calculate the pose accuracy obtained in result, pose estimation easily occurs
The situation of failure, therefore, rejecting these mispairing points has very big necessity.
ORB algorithms are a kind of outstanding characteristic point detection, description and matching algorithms, and calculating speed is than SIFT, SURF algorithm
1 ~ 2 order of magnitude is improved, is broadly divided into three parts:Have directive FAST characteristic points detection, there is invariable rotary characteristic
BRIEF feature point descriptions and characteristic point thick matching.The method of feature detection portion is used on image pyramid
The characteristics of FAST Corner Detection Algorithms, the algorithm, arithmetic speed was very fast as its name.Feature point description part is then
It is using binary-coded BRIEF description.ORB algorithm arithmetic speeds are fast, and illumination robustness is good, and affinity is also good, but
Without scale invariability, cause ORB that there is limitation in application field.
During images match due to the extraction process and feature point description of characteristic point can not all accomplish it is absolute accurate,
So there are many erroneous matchings, the matching of mistake to be broadly divided into two kinds of situations for obtained initial matching centering:First, carry out
Matched characteristic point is wrong, second, the characteristic point on image can not be matched.RANSAC algorithms are a kind of classical disappear
Except the method for feature error hiding, there is the advantages that matching precision is high, reliability is high, strong robustness, but significant drawback is to take out at random
Matching pair is taken, does not account for the quality of matching pair, by thick matched all characteristic points to being all iterated, when error hiding pair
Resulted in when more and waste many times.
The content of the invention
In view of this, the object of the present invention is to provide a kind of visual pattern characteristic point error hiding method of purification based on ORB,
The ORB characteristic points being evenly distributed are detected using improved FAST feature point detection algorithms, use bi-directional matching crossing filtering side
Then method carries out secondary error hiding using given threshold, the method for the Hamming distance that sorts and picks first to matching to carrying out primary screening
Remove, finally recycle RANSAC algorithm iterations.Matched speed is also improved while being allowed to improve matching accuracy.
The present invention is realized using following scheme:A kind of visual pattern characteristic point error hiding method of purification based on ORB, it is wrapped
Include following steps:
Step S1:Read the left and right two shot under different viewing angles to be detected and open image, structure scalogram is carried out to image
Grid processing is carried out as pyramid, and to every layer of pyramid diagram picture;
Step S2:In each small grid of every layer of pyramid diagram picture feature is carried out using improved FAST feature point detection algorithms
The detection of point, the characteristic point extracted using the data structure storage of Octree, and determine feature point coordinates;
Step S3:The characteristic point close to image border is removed, and calculates the barycenter direction of residue character point;
Step S4:Calculate ORB feature point descriptions;
Step S5:The characteristic point on two images is slightly matched using based on the method for Hamming distance;
Step S6:The slightly matching pair of filter method screening characteristic point is intersected using bi-directional matching;
Step S7:To the remaining matching double points of step S6 by given threshold, the method for the Hamming distance that sorts, to error hiding to again
Once reject;
Step S8:RANSAC algorithm iterations are carried out to the remaining matching double points of step S7, and export the matching image after purification.
In an embodiment of the present invention, the method for structure scale image pyramid is in the step S1:According to improved
The size of Filtering Template in FAST feature point detection algorithms, the Gauss down-sampled to the picture to be detected progress Gauss of input
Down-sampled mesoscale value is, wherein N is the size of Filtering Template;Grid processing is carried out to every layer of pyramid diagram picture
Method be:It is divided into 8 layers of pyramid diagram picture, first layer is divided into 14*20 grids;The second layer is divided into 12*16 grids;Third layer is divided into
10*13 grids;4th layer is divided into 8*11 grids;Layer 5 is divided into 6*9 grids;Layer 6 is divided into 5*7 grids;Layer 7 is divided into 4*
6 grids;8th layer is divided into 3*4 grids.
Further, the detection of characteristic point is carried out including following using improved FAST feature point detection algorithms in step S1
Step:A pixel p is chosen from picture and its pixel value is set to, a suitable threshold value t is set, is chosen with the picture
Vegetarian refreshments is the center of circle, is 16 pixels on the circle of radius with 3;If there are n continuous pixels, their pixel on circle
Value all compares+ t is big, or all compares- t is small, then it is assumed that it is an angle point.
Preferably, n takes 12 or 9.
In an embodiment of the present invention, the barycenter side of characteristic point is drawn in the step S3 by calculating the square of characteristic point
To specific method is:(p+q) rank square of any one feature vertex neighborhood is in definition image:,
Then the center-of-mass coordinate C of its neighborhood is:()(,), pass through formula
Draw the barycenter direction θ of this feature point;The barycenter direction of all residue character points is drawn using the formula.
In an embodiment of the present invention, the specific method of calculating ORB feature point description is in the step S4:Step
S41:M related coefficient is calculated close to 0.5 random point pair using exhaust algorithm;Step S42:Will be random in step S41
Group point rotates the barycenter direction of the characteristic point according to step S3, on the corresponding image pyramid layer of characteristic point,
Using formulaBinary descriptor is generated, in formulaFor postrotational random point
It is right,It is image block p in pixelThe gray value at place;M binary bits strings are drawn i.e. according to the formula
It is characterized description.
In an embodiment of the present invention, the thick matching process of characteristic point is in the step S5:Calculate special on two images
Sign point description vectors between Hamming distance, Hamming distance it is smaller represent characteristic point between similarity it is higher, when characteristic point is retouched
The similarity of son is stated when being more than certain threshold value, then it is assumed that Feature Points Matching is successfully.
In an embodiment of the present invention, the specific method of the thick matching pair of screening characteristic point is in the step S6:S61:It is false
If the collection for the characteristic point that left figure detects is combined into: A = {|∈ A, i=1,2 ... k }
The collection for the characteristic point that right figure detects is combined into: B = {| ∈ B, i=1,2 ... h }
To some characteristic point in AIts two characteristic point with the closest and secondary neighbour of Euclidean in B is found out, if arest neighbors
The ratio between Euclidean distance and time neighbour's Euclidean distance are less than given threshold value, representIt is a pair of of matching with this arest neighbors characteristic point
Point pair;S62:Bi-directional matching crossing filtering method is combined on the basis of S61 to matching to screening, to each spy in A
Sign pointCorresponding characteristic point is found in B, to each characteristic point in BCorresponding characteristic point is found in A;S63:
Reverse matching mechanisms are introduced, if the characteristic point in AMatch point in B is, and the characteristic point in BMatching in A
Put and be;Then think that this to being correct, if matching is unsuccessful, directly weeds out matching.
In an embodiment of the present invention, given threshold, the secondary error hiding of sequence Hamming distance progress pick in the step S7
Except method is:A smallest hamming distance threshold value is set, the Hamming distance of the remaining matching double points of step S6 is ranked up
To minimum range, it will match and match point of the Hamming distance more than smallest hamming distance certain multiple rejected.
In an embodiment of the present invention, RANSAC algorithm iteration methods are in the step S8:Step S81:To step S7
The set of remaining N number of point pair to be matched carries out Unitary coordinate, i.e.,, wherein i=1,2 ..., N;Step S82:
4 points pair are randomly selected from N number of to be matched centering, solve 8 parameters of homography matrix H;Step S83:Calculate remaining
(N-4) a characteristic point passes through the Euclidean distance d, d=‖ between the characteristic point and other characteristic points to be matched that matrix H converts
HA-A' ‖;If d < t, t are threshold value, then the characteristic point to be matched is interior point, otherwise is exterior point;Step S84:Repeat S82~S83
Step K times, interior quantity under meter each time, wherein K is iterations;Step S85:The most set conduct of points in selection
Point set in final, 8 parameters in corresponding H conversion are as parameter Estimation optimal value.
The advantage of the invention is that solving traditional RANSAC algorithms randomly selects matching pair, the matter of matching pair is not accounted for
Amount quality, by thick matched all characteristic points to being all iterated, when error hiding to it is more when cause to waste the plenty of time and ask
Topic.RANSAC is improved with reference to FAST, to characteristic point to carrying out prescreening and secondary rejecting, improves improved algorithm
It ensure that the timeliness of algorithm while matching accuracy.
Brief description of the drawings
Fig. 1 is the overall flow figure of the embodiment of the present invention.
Fig. 2 is the image schematic diagram of a specific embodiment of the invention.
Embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
The present invention provides a kind of visual pattern characteristic point error hiding method of purification based on ORB, and main flow schematic diagram is such as
Shown in Fig. 1.It comprises the following steps:
Step S1:Read the left and right two shot under different viewing angles to be detected and open image, structure scalogram is carried out to image
Grid processing is carried out as pyramid, and to every layer of pyramid diagram picture;
Step S2:In each small grid of every layer of pyramid diagram picture feature is carried out using improved FAST feature point detection algorithms
The detection of point, the characteristic point extracted using the data structure storage of Octree, and determine feature point coordinates;
Step S3:The characteristic point close to image border is removed, and calculates the barycenter direction of residue character point;
Step S4:Calculate ORB feature point descriptions;
Step S5:The characteristic point on two images is slightly matched using based on the method for Hamming distance;
Step S6:The slightly matching pair of filter method screening characteristic point is intersected using bi-directional matching;
Step S7:To the remaining matching double points of step S6 by given threshold, the method for the Hamming distance that sorts, to error hiding to again
Once reject;
Step S8:RANSAC algorithm iterations are carried out to the remaining matching double points of step S7, and export the matching image after purification.
In the present embodiment, the method for structure scale image pyramid is in the step S1:It is special according to improved FAST
The size of Filtering Template in sign point detection, the Gauss down-sampled middle ruler down-sampled to the picture to be detected progress Gauss of input
Angle value is, wherein N is the size of Filtering Template;The method of grid processing is carried out to every layer of pyramid diagram picture is:
A total of 8 layers of pyramid diagram picture, the ranks number of every layer of grid processing are as shown in the table:
Table 1
In an embodiment of the present invention, the definition of improved FAST feature point detection algorithms is:If the gray value of certain point is than it
The gray value of enough points is small or big in surrounding neighbors, then the point may be angle point.Schematic diagram as shown in Figure 2
A pixel p is chosen from picture and its pixel value is set to, a suitable threshold value t is set, is chosen with the picture
Vegetarian refreshments is the center of circle, is 16 pixels on the circle of radius with 3.If there is n on circle(N generally takes 12 or 9)It is a continuous
Pixel, their pixel value all compare+ t is big, or all compares- t is small, then it is assumed that it is an angle point.
Octree structure:Each node of Octree represents the volume element of a square, and each node has eight sons
Node, the volume element represented by this eight child nodes are added together the volume for being equal to father node, and General Central point is as section
The bifurcated center of point.
The barycenter direction of characteristic point is drawn in the step S3 by calculating the square of characteristic point, specific method is:Definition figure
(p+q) rank square of any one feature vertex neighborhood is as in:, then the center-of-mass coordinate C of its neighborhood
For:()(,), can calculation formulaDraw the matter of this feature point
Heart direction θ;The barycenter direction of all residue character points is drawn using the formula.
In the present embodiment, the specific method of calculating ORB feature point description is in the step S4:
Step S41:M related coefficient is calculated close to 0.5 random point pair using exhaust algorithm;
Step S42:Random groups point in step S41 rotates the barycenter direction of the characteristic point according to step S3,
On the corresponding image pyramid layer of characteristic point, using formulaGenerate binary system description
Son, in formulaFor postrotational random point pair,It is image block p in pixelThe gray value at place;According to institute
State formula and show that m binary bits strings are feature point description.
In the present embodiment, the thick matching process of characteristic point is in the step S5:Characteristic point on two images is calculated to retouch
State the Hamming distance between vector, the similarity that Hamming distance is smaller to be represented between characteristic point is higher.When feature point description
When similarity is more than certain threshold value, it may be considered that Feature Points Matching success.
In the present embodiment, the specific method of the thick matching pair of screening characteristic point is in the step S6:
S61:Assuming that the collection for the characteristic point that left figure detects is combined into: A = {|∈ A, i=1,2 ... k }
The collection for the characteristic point that right figure detects is combined into:B = {| ∈ B, i=1,2 ... h }
To some characteristic point in AIts two characteristic point with the closest and secondary neighbour of Euclidean in B is found out, if arest neighbors
The ratio between Euclidean distance and time neighbour's Euclidean distance are less than given threshold value, representIt is a pair of of matching with this arest neighbors characteristic point
Point pair, such as formulaIt is shown, in formula、Represent respectivelyBetween its arest neighbors characteristic point and secondary neighbour's characteristic point
Euclidean distance,Represent given threshold value.
S62:Bi-directional matching crossing filtering method is combined on the basis of S61 to matching to screening, to each in A
Characteristic pointCorresponding characteristic point is found in B, to each characteristic point in BCorresponding characteristic point is found in A。
S63:Reverse matching mechanisms are introduced, if the characteristic point in AMatch point in B is, and the characteristic point in BMatch point in A is.Then think that this to being correct, if matching is unsuccessful, directly weeds out matching.
In the present embodiment, given threshold, sequence Hamming distance carry out secondary error hiding elimination method in the step S7
For:A smallest hamming distance threshold value is set, the Hamming distance of the remaining matching double points of step S6 is ranked up to obtain minimum
Distance, is more than smallest hamming distance certain multiple by matching to Hamming distance(Take 10)Match point reject.
In the present embodiment, RANSAC algorithm iteration methods are in the step S8:
Step S81:The set of N number of point pair to be matched remaining to step S7 carries out Unitary coordinate, i.e.,, wherein
I=1,2 ..., N;
Step S82:4 points pair are randomly selected from N number of to be matched centering, solve 8 parameters of homography matrix H;
Step S83:Remaining (N-4) a characteristic point is calculated by the characteristic point that matrix H converts and other features to be matched
Euclidean distance d, d=‖ HA-A' ‖ between point.If d < t (t is threshold value), which is interior point, otherwise is
Exterior point;
Step S84:Repeat S82~S83 steps K times, interior quantity under meter each time, wherein K is iterations;
Step S85:The most set of points is as point set in final in selection, 8 in corresponding H conversion
Parameter is as parameter Estimation optimal value.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, should all belong to the covering scope of the present invention.
Claims (10)
- A kind of 1. visual pattern characteristic point error hiding method of purification based on ORB, it is characterised in that:Comprise the following steps:Step S1:Read the left and right two shot under different viewing angles to be detected and open image, structure scalogram is carried out to image Grid processing is carried out as pyramid, and to every layer of pyramid diagram picture;Step S2:In each small grid of every layer of pyramid diagram picture feature is carried out using improved FAST feature point detection algorithms The detection of point, the characteristic point extracted using the data structure storage of Octree, and determine feature point coordinates;Step S3:The characteristic point close to image border is removed, and calculates the barycenter direction of residue character point;Step S4:Calculate ORB feature point descriptions;Step S5:The characteristic point on two images is slightly matched using based on the method for Hamming distance;Step S6:The slightly matching pair of filter method screening characteristic point is intersected using bi-directional matching;Step S7:To the remaining matching double points of step S6 by given threshold, the method for the Hamming distance that sorts, to error hiding to again Once reject;Step S8:RANSAC algorithm iterations are carried out to the remaining matching double points of step S7, and export the matching image after purification.
- 2. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:The method of scale image pyramid is built in the step S1 is:Filtered according in improved FAST feature point detection algorithms The size of template, down-sampled to the picture to be detected progress Gauss of input, the down-sampled mesoscale value of Gauss is, wherein N is the size of Filtering Template;The method of grid processing is carried out to every layer of pyramid diagram picture is:It is divided into 8 layers Pyramid diagram picture, first layer are divided into 14*20 grids;The second layer is divided into 12*16 grids;Third layer is divided into 10*13 grids;4th layer It is divided into 8*11 grids;Layer 5 is divided into 6*9 grids;Layer 6 is divided into 5*7 grids;Layer 7 is divided into 4*6 grids;8th layer is divided into 3*4 grids.
- 3. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 2, its feature exist In:The detection for being carried out characteristic point in step S1 using improved FAST feature point detection algorithms is comprised the following steps:From picture Choose a pixel p and its pixel value is set to, a suitable threshold value t is set, is chosen using the pixel as the center of circle, with 3 be 16 pixels on the circle of radius;If there is n continuous pixels on circle, their pixel value all compares+ t is big, Or all compare- t is small, then it is assumed that it is an angle point.
- 4. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 3, its feature exist In:N takes 12 or 9.
- 5. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:The barycenter direction of characteristic point is drawn in the step S3 by calculating the square of characteristic point, specific method is:Define in image and appoint (p+q) the rank square of a feature vertex neighborhood of anticipating is:, then the center-of-mass coordinate C of its neighborhood be: ()(,), pass through formulaDraw the barycenter direction θ of this feature point; The barycenter direction of all residue character points is drawn using the formula.
- 6. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:The specific method of calculating ORB feature point description is in the step S4:Step S41:M related coefficient is calculated close to 0.5 random point pair using exhaust algorithm;Step S42:Random groups point in step S41 rotates the barycenter direction of the characteristic point according to step S3, On the corresponding image pyramid layer of characteristic point, using formulaGenerate binary system description Son, in formulaFor postrotational random point pair,It is image block p in pixelThe gray value at place;According to institute State formula and show that m binary bits strings are feature point description.
- 7. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:The thick matching process of characteristic point is in the step S5:Calculate the Hamming distance between feature point description vector on two images From the similarity between the smaller expression characteristic point of Hamming distance is higher, when the similarity of feature point description is more than certain threshold value When, then it is assumed that Feature Points Matching success.
- 8. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:The specific method of the thick matching pair of screening characteristic point is in the step S6:S61:Assuming that the collection for the characteristic point that left figure detects is combined into: A = { |∈ A, i=1,2 ... k }The collection for the characteristic point that right figure detects is combined into: B = { | ∈ B, i=1,2 ... h }To some characteristic point in AIts two characteristic point with the closest and secondary neighbour of Euclidean in B is found out, if arest neighbors The ratio between Euclidean distance and time neighbour's Euclidean distance are less than given threshold value, representIt it is a pair of with this arest neighbors characteristic point With point pair;S62:Bi-directional matching crossing filtering method is combined on the basis of S61 to matching to screening, to each feature in A PointCorresponding characteristic point is found in B, to each characteristic point in BCorresponding characteristic point is found in A;S63:Reverse matching mechanisms are introduced, if the characteristic point in AMatch point in B is, and the characteristic point in BIn A In match point be;Then think that this to being correct, if matching is unsuccessful, directly weeds out matching.
- 9. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:Given threshold, the secondary error hiding elimination method of sequence Hamming distance progress are in the step S7:The minimum Hamming of setting one Distance threshold, is ranked up to obtain minimum range to the Hamming distance of the remaining matching double points of step S6, will match to Hamming distance From the match point rejecting more than smallest hamming distance certain multiple.
- 10. a kind of visual pattern characteristic point error hiding method of purification based on ORB according to claim 1, its feature exist In:RANSAC algorithm iteration methods are in the step S8:Step S81:The set of N number of point pair to be matched remaining to step S7 carries out Unitary coordinate, i.e.,, wherein i =1,2 ..., N;Step S82:4 points pair are randomly selected from N number of to be matched centering, solve 8 parameters of homography matrix H;Step S83:Remaining (N-4) a characteristic point is calculated by the characteristic point that matrix H converts and other characteristic points to be matched Between Euclidean distance d, d=‖ HA-A' ‖;If d < t, t are threshold value, then the characteristic point to be matched is interior point, otherwise is exterior point;Step S84:Repeat S82~S83 steps K times, interior quantity under meter each time, wherein K is iterations;Step S85:The most set of points is used as point set in final in selection, 8 ginsengs in corresponding H conversion Number is used as parameter Estimation optimal value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711297537.9A CN108010045A (en) | 2017-12-08 | 2017-12-08 | Visual pattern characteristic point error hiding method of purification based on ORB |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711297537.9A CN108010045A (en) | 2017-12-08 | 2017-12-08 | Visual pattern characteristic point error hiding method of purification based on ORB |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108010045A true CN108010045A (en) | 2018-05-08 |
Family
ID=62057883
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711297537.9A Pending CN108010045A (en) | 2017-12-08 | 2017-12-08 | Visual pattern characteristic point error hiding method of purification based on ORB |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108010045A (en) |
Cited By (34)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108805812A (en) * | 2018-06-04 | 2018-11-13 | 东北林业大学 | Multiple dimensioned constant ORB algorithms for image mosaic |
CN109008909A (en) * | 2018-07-13 | 2018-12-18 | 宜宾学院 | A kind of low-power consumption capsule endoscope Image Acquisition and three-dimensional reconstruction system |
CN109325453A (en) * | 2018-09-27 | 2019-02-12 | 沈阳理工大学 | The template matching tracking of moving target |
CN109727287A (en) * | 2018-12-27 | 2019-05-07 | 江南大学 | A kind of improvement register method and its system suitable for augmented reality |
CN109766758A (en) * | 2018-12-12 | 2019-05-17 | 北京计算机技术及应用研究所 | A kind of vision SLAM method based on ORB feature |
CN109801220A (en) * | 2019-01-23 | 2019-05-24 | 北京工业大学 | Mapping parameters method in a kind of splicing of line solver Vehicular video |
CN109829459A (en) * | 2019-01-21 | 2019-05-31 | 重庆邮电大学 | Based on the vision positioning method for improving RANSAC |
CN109886213A (en) * | 2019-02-25 | 2019-06-14 | 湖北亿咖通科技有限公司 | Fatigue state judgment method, electronic equipment and computer readable storage medium |
CN110009558A (en) * | 2019-01-17 | 2019-07-12 | 柳州康云互联科技有限公司 | A kind of normalized method of easy image color |
CN110046623A (en) * | 2019-03-04 | 2019-07-23 | 青岛小鸟看看科技有限公司 | A kind of characteristics of image point extracting method and camera |
CN110084248A (en) * | 2019-04-23 | 2019-08-02 | 陕西理工大学 | A kind of ORB feature homogenization extracting method |
CN110084784A (en) * | 2019-03-30 | 2019-08-02 | 天津大学 | Corner feature real-time detection and matching process on star |
CN110334762A (en) * | 2019-07-04 | 2019-10-15 | 华南师范大学 | A kind of feature matching method combining ORB and SIFT based on quaternary tree |
CN110414534A (en) * | 2019-07-01 | 2019-11-05 | 深圳前海达闼云端智能科技有限公司 | Image feature extraction method and device, storage medium and electronic equipment |
CN110414533A (en) * | 2019-06-24 | 2019-11-05 | 东南大学 | A kind of feature extracting and matching method for improving ORB |
CN110675437A (en) * | 2019-09-24 | 2020-01-10 | 重庆邮电大学 | Image matching method based on improved GMS-ORB characteristics and storage medium |
CN110728751A (en) * | 2019-06-19 | 2020-01-24 | 武汉科技大学 | Construction method of indoor 3D point cloud semantic map |
CN110874850A (en) * | 2018-09-04 | 2020-03-10 | 湖北智视科技有限公司 | Real-time unilateral grid feature registration method oriented to target positioning |
CN111209920A (en) * | 2020-01-06 | 2020-05-29 | 桂林电子科技大学 | Airplane detection method under complex dynamic background |
CN111257588A (en) * | 2020-01-17 | 2020-06-09 | 东北石油大学 | ORB and RANSAC-based oil phase flow velocity measurement method |
CN111429493A (en) * | 2020-03-20 | 2020-07-17 | 青岛联合创智科技有限公司 | Method for matching feature points among multiple images |
CN111445389A (en) * | 2020-02-24 | 2020-07-24 | 山东省科学院海洋仪器仪表研究所 | Wide-view-angle rapid splicing method for high-resolution images |
CN111898428A (en) * | 2020-06-23 | 2020-11-06 | 东南大学 | Unmanned aerial vehicle feature point matching method based on ORB |
CN112016610A (en) * | 2020-08-25 | 2020-12-01 | 济南大学 | Image feature matching method and system |
CN112037193A (en) * | 2020-08-28 | 2020-12-04 | 长安大学 | Power line feature marking method and device |
CN112418300A (en) * | 2020-11-19 | 2021-02-26 | 重庆连芯光电技术研究院有限公司 | Method for accelerating image matching |
CN112562000A (en) * | 2020-12-23 | 2021-03-26 | 安徽大学 | Robot vision positioning method based on feature point detection and mismatching screening |
CN112580506A (en) * | 2020-12-18 | 2021-03-30 | 杭州电子科技大学 | Iris characteristic point comparison method based on bidirectional minimum Hamming distance |
CN112837223A (en) * | 2021-01-28 | 2021-05-25 | 杭州国芯科技股份有限公司 | Super-large image registration splicing method based on overlapping subregions |
CN113011498A (en) * | 2021-03-22 | 2021-06-22 | 华南理工大学 | Method, system and medium for extracting and matching feature points based on color images |
CN113103232A (en) * | 2021-04-12 | 2021-07-13 | 电子科技大学 | Intelligent equipment self-adaptive motion control method based on feature distribution matching |
CN113596401A (en) * | 2021-07-29 | 2021-11-02 | 上海应用技术大学 | Image de-similarity transmission and restoration method based on ORB similarity judgment |
CN113642573A (en) * | 2021-07-20 | 2021-11-12 | 南京红松信息技术有限公司 | Picture separation method based on grids |
CN114444564A (en) * | 2021-12-14 | 2022-05-06 | 广州锐竞信息科技有限责任公司 | High-similarity paper picture comparison method based on deep neural network model |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850851A (en) * | 2015-04-22 | 2015-08-19 | 福州大学 | ORB feature point matching method with scale invariance |
CN104851094A (en) * | 2015-05-14 | 2015-08-19 | 西安电子科技大学 | Improved method of RGB-D-based SLAM algorithm |
-
2017
- 2017-12-08 CN CN201711297537.9A patent/CN108010045A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850851A (en) * | 2015-04-22 | 2015-08-19 | 福州大学 | ORB feature point matching method with scale invariance |
CN104851094A (en) * | 2015-05-14 | 2015-08-19 | 西安电子科技大学 | Improved method of RGB-D-based SLAM algorithm |
Non-Patent Citations (4)
Title |
---|
余杰: "基于ORB关键帧闭环检测算法的SLAM方法研究", 《中国优秀硕士学位论文全文数据库(电子期刊)信息科技辑》 * |
刘威等: "一种基于ORB检测的特征点匹配算法", 《激光与红外》 * |
李小红等: "基于ORB特征的快速目标检测算法", 《电子测量与仪器学报》 * |
杨杰: "图像快速拼接方法的研究与实现", 《中国优秀硕士学位论文全文数据库(电子期刊)信息科技辑》 * |
Cited By (48)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108805812A (en) * | 2018-06-04 | 2018-11-13 | 东北林业大学 | Multiple dimensioned constant ORB algorithms for image mosaic |
CN109008909A (en) * | 2018-07-13 | 2018-12-18 | 宜宾学院 | A kind of low-power consumption capsule endoscope Image Acquisition and three-dimensional reconstruction system |
CN109008909B (en) * | 2018-07-13 | 2024-01-26 | 宜宾学院 | Low-power-consumption capsule endoscope image acquisition and three-dimensional reconstruction system |
CN110874850A (en) * | 2018-09-04 | 2020-03-10 | 湖北智视科技有限公司 | Real-time unilateral grid feature registration method oriented to target positioning |
CN109325453A (en) * | 2018-09-27 | 2019-02-12 | 沈阳理工大学 | The template matching tracking of moving target |
CN109325453B (en) * | 2018-09-27 | 2022-03-04 | 沈阳理工大学 | Template matching tracking method for moving target |
CN109766758A (en) * | 2018-12-12 | 2019-05-17 | 北京计算机技术及应用研究所 | A kind of vision SLAM method based on ORB feature |
CN109727287A (en) * | 2018-12-27 | 2019-05-07 | 江南大学 | A kind of improvement register method and its system suitable for augmented reality |
CN109727287B (en) * | 2018-12-27 | 2023-08-08 | 江南大学 | Improved registration method and system suitable for augmented reality |
CN110009558A (en) * | 2019-01-17 | 2019-07-12 | 柳州康云互联科技有限公司 | A kind of normalized method of easy image color |
CN109829459A (en) * | 2019-01-21 | 2019-05-31 | 重庆邮电大学 | Based on the vision positioning method for improving RANSAC |
CN109829459B (en) * | 2019-01-21 | 2022-05-17 | 重庆邮电大学 | Visual positioning method based on improved RANSAC |
CN109801220A (en) * | 2019-01-23 | 2019-05-24 | 北京工业大学 | Mapping parameters method in a kind of splicing of line solver Vehicular video |
CN109801220B (en) * | 2019-01-23 | 2023-03-28 | 北京工业大学 | Method for solving mapping parameters in vehicle-mounted video splicing on line |
CN109886213A (en) * | 2019-02-25 | 2019-06-14 | 湖北亿咖通科技有限公司 | Fatigue state judgment method, electronic equipment and computer readable storage medium |
CN110046623A (en) * | 2019-03-04 | 2019-07-23 | 青岛小鸟看看科技有限公司 | A kind of characteristics of image point extracting method and camera |
CN110084784A (en) * | 2019-03-30 | 2019-08-02 | 天津大学 | Corner feature real-time detection and matching process on star |
CN110084248A (en) * | 2019-04-23 | 2019-08-02 | 陕西理工大学 | A kind of ORB feature homogenization extracting method |
CN110084248B (en) * | 2019-04-23 | 2022-10-21 | 陕西理工大学 | ORB feature homogenization extraction method |
CN110728751A (en) * | 2019-06-19 | 2020-01-24 | 武汉科技大学 | Construction method of indoor 3D point cloud semantic map |
CN110414533A (en) * | 2019-06-24 | 2019-11-05 | 东南大学 | A kind of feature extracting and matching method for improving ORB |
CN110414534A (en) * | 2019-07-01 | 2019-11-05 | 深圳前海达闼云端智能科技有限公司 | Image feature extraction method and device, storage medium and electronic equipment |
CN110334762A (en) * | 2019-07-04 | 2019-10-15 | 华南师范大学 | A kind of feature matching method combining ORB and SIFT based on quaternary tree |
CN110334762B (en) * | 2019-07-04 | 2020-04-03 | 华南师范大学 | Feature matching method based on quad tree combined with ORB and SIFT |
CN110675437B (en) * | 2019-09-24 | 2023-03-28 | 重庆邮电大学 | Image matching method based on improved GMS-ORB characteristics and storage medium |
CN110675437A (en) * | 2019-09-24 | 2020-01-10 | 重庆邮电大学 | Image matching method based on improved GMS-ORB characteristics and storage medium |
CN111209920B (en) * | 2020-01-06 | 2022-09-23 | 桂林电子科技大学 | Airplane detection method under complex dynamic background |
CN111209920A (en) * | 2020-01-06 | 2020-05-29 | 桂林电子科技大学 | Airplane detection method under complex dynamic background |
CN111257588A (en) * | 2020-01-17 | 2020-06-09 | 东北石油大学 | ORB and RANSAC-based oil phase flow velocity measurement method |
CN111445389A (en) * | 2020-02-24 | 2020-07-24 | 山东省科学院海洋仪器仪表研究所 | Wide-view-angle rapid splicing method for high-resolution images |
CN111429493A (en) * | 2020-03-20 | 2020-07-17 | 青岛联合创智科技有限公司 | Method for matching feature points among multiple images |
CN111898428A (en) * | 2020-06-23 | 2020-11-06 | 东南大学 | Unmanned aerial vehicle feature point matching method based on ORB |
CN112016610B (en) * | 2020-08-25 | 2022-05-31 | 济南大学 | Image feature matching method and system |
CN112016610A (en) * | 2020-08-25 | 2020-12-01 | 济南大学 | Image feature matching method and system |
CN112037193A (en) * | 2020-08-28 | 2020-12-04 | 长安大学 | Power line feature marking method and device |
CN112418300A (en) * | 2020-11-19 | 2021-02-26 | 重庆连芯光电技术研究院有限公司 | Method for accelerating image matching |
CN112580506A (en) * | 2020-12-18 | 2021-03-30 | 杭州电子科技大学 | Iris characteristic point comparison method based on bidirectional minimum Hamming distance |
CN112562000A (en) * | 2020-12-23 | 2021-03-26 | 安徽大学 | Robot vision positioning method based on feature point detection and mismatching screening |
CN112837223A (en) * | 2021-01-28 | 2021-05-25 | 杭州国芯科技股份有限公司 | Super-large image registration splicing method based on overlapping subregions |
CN112837223B (en) * | 2021-01-28 | 2023-08-29 | 杭州国芯科技股份有限公司 | Super-large image registration splicing method based on overlapped subareas |
CN113011498B (en) * | 2021-03-22 | 2023-09-26 | 华南理工大学 | Feature point extraction and matching method, system and medium based on color image |
CN113011498A (en) * | 2021-03-22 | 2021-06-22 | 华南理工大学 | Method, system and medium for extracting and matching feature points based on color images |
CN113103232B (en) * | 2021-04-12 | 2022-05-20 | 电子科技大学 | Intelligent equipment self-adaptive motion control method based on feature distribution matching |
CN113103232A (en) * | 2021-04-12 | 2021-07-13 | 电子科技大学 | Intelligent equipment self-adaptive motion control method based on feature distribution matching |
CN113642573A (en) * | 2021-07-20 | 2021-11-12 | 南京红松信息技术有限公司 | Picture separation method based on grids |
CN113642573B (en) * | 2021-07-20 | 2023-10-13 | 南京红松信息技术有限公司 | Picture separation method based on grids |
CN113596401A (en) * | 2021-07-29 | 2021-11-02 | 上海应用技术大学 | Image de-similarity transmission and restoration method based on ORB similarity judgment |
CN114444564A (en) * | 2021-12-14 | 2022-05-06 | 广州锐竞信息科技有限责任公司 | High-similarity paper picture comparison method based on deep neural network model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108010045A (en) | Visual pattern characteristic point error hiding method of purification based on ORB | |
Song et al. | Constructing stronger and faster baselines for skeleton-based action recognition | |
CN106355577B (en) | Rapid image matching method and system based on significant condition and global coherency | |
CN106960451B (en) | Method for increasing number of feature points of image weak texture area | |
Hu et al. | Angular pattern and binary angular pattern for shape retrieval | |
Lisanti et al. | Group re-identification via unsupervised transfer of sparse features encoding | |
CN104809731B (en) | A kind of rotation Scale invariant scene matching method based on gradient binaryzation | |
CN110334762B (en) | Feature matching method based on quad tree combined with ORB and SIFT | |
CN107424161B (en) | Coarse-to-fine indoor scene image layout estimation method | |
CN110246168A (en) | A kind of feature matching method of mobile crusing robot binocular image splicing | |
Li et al. | LPSNet: a novel log path signature feature based hand gesture recognition framework | |
CN101556692A (en) | Image mosaic method based on neighborhood Zernike pseudo-matrix of characteristic points | |
CN106919944A (en) | A kind of wide-angle image method for quickly identifying based on ORB algorithms | |
CN104240231A (en) | Multi-source image registration based on local structure binary pattern | |
Liu et al. | Texture classification in extreme scale variations using GANet | |
Liu et al. | Pgfnet: Preference-guided filtering network for two-view correspondence learning | |
JP2018142189A (en) | Program, distance measuring method, and distance measuring device | |
CN112614167A (en) | Rock slice image alignment method combining single-polarization and orthogonal-polarization images | |
CN111199558A (en) | Image matching method based on deep learning | |
CN114358166B (en) | Multi-target positioning method based on self-adaptive k-means clustering | |
Liu et al. | Progressive context-dependent inference for object detection in remote sensing imagery | |
CN103336964A (en) | SIFT image matching method based on module value difference mirror image invariant property | |
CN107358244B (en) | A kind of quick local invariant feature extracts and description method | |
Liang et al. | PKLNet: Keypoint localization neural network for touchless palmprint recognition based on edge-aware regression | |
CN109919057B (en) | Multi-mode fusion gesture recognition method based on efficient convolutional neural network |
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 |
Application publication date: 20180508 |
|
RJ01 | Rejection of invention patent application after publication |