CN103927526B - Vehicle detecting method based on Gauss difference multi-scale edge fusion - Google Patents
Vehicle detecting method based on Gauss difference multi-scale edge fusion Download PDFInfo
- Publication number
- CN103927526B CN103927526B CN201410181851.0A CN201410181851A CN103927526B CN 103927526 B CN103927526 B CN 103927526B CN 201410181851 A CN201410181851 A CN 201410181851A CN 103927526 B CN103927526 B CN 103927526B
- Authority
- CN
- China
- Prior art keywords
- image
- edge
- difference
- gaussian
- vehicle
- 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.)
- Active
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a vehicle detecting method based on Gauss difference multi-scale edge fusion. The method includes the steps that Gauss scale transformation is performed on images to obtain four Gauss images in adjacent scales; according to the four Gauss images in the adjacent scales, the difference operation is performed between the images in the adjacent scales to obtain three Gauss difference images different in scale, edge detection is performed on the obtained three Gauss difference images through a Sobel operator, then edge fusion with the scale upward searching is performed to remove a lot of background edges while edge information of a vehicle is obtained as much as possible, and expansion, closed operation, hole filling and other series of morphological operation are performed on the fused edge images to obtain a connected domain image representing the vehicle; an outside rectangle of the position where the vehicle is located is determined in the original image according to the position information of a connected domain to detect the vehicle. The images in multiple scales are processed, so that algorithm complexity is reduced, operation amount is reduced, efficiency of vehicle detection is effectively improved, and a good detection result is obtained.
Description
Technical field
The invention belongs to field of video detection is and in particular to a kind of examined based on the vehicle that difference of Gaussian multi-scale edge merges
Survey method.
Background technology
Moving object detection is computer vision and a key technology of image steganalysis.The vehicle inspection of view-based access control model
Survey technology is the study hotspot of intelligent transportation image procossing, has a wide range of applications in intelligent transportation field, and such as vehicle auxiliary is driven
Sail system, traffic parameter statistical system etc..
Three classes are broadly divided into based on the vehicle checking method of computer vision：Based on model, based on neural network learning,
The method of feature based.Detection method based on model is built in advance by the candidate detecting vehicle region and Computer Database
Vertical auto model is mated thus being detected vehicle, but the shortcoming of the method is to place one's entire reliance upon to all variety classes cars
Carry out Geometric Modeling, this is difficult to realize.
Based on study detection method by using sample, neutral net is trained, carried out with the network training
Vehicle identification, the method is frequently used for verifying the testing result of additive method.The method of feature based is passed through to detect the office of vehicle
Portion's feature such as symmetrical parts (wheel, head lamp, taillight etc.), edge and shade etc., thus position vehicle.The advantage of the method is
Using vehicle, all recognizable feature detects vehicle it is adaptable to the vehicle detection at sleet sky or even night in most circumstances
Problem.By detect wheel realize the method for vehicle detection easily by vehicle driving posture, block the problems such as affected, and lead to
The method crossing detection car light is also disturbed by the street lamp in night scenes and urban lighting, affects testing result.And it is based on edge
The vehicle checking method of detection (including vehicle shadow detection) is due to the presence of background edge (as lane line, railing, trees etc.)
Lead to the inaccurate of testing result, therefore, how to detect that vehicle edge suppresses figure viewed from behind edge, becomes and carries to greatest extent simultaneously
The key issue of high the method Detection accuracy.
Content of the invention
It is an object of the invention to overcoming the problems of the prior art, one kind is provided to melt based on difference of Gaussian multi-scale edge
The vehicle checking method closing, this method reduces algorithm complex, decreases amount of calculation, obtain preferable testing result, effectively
Improve detection efficiency.
To achieve these goals, the present invention takes following technical solution to be achieved：
Step one, gathers certain road section traffic volume video, to the piece image gray processing in video and to carry out gaussian pyramid many
Change of scale, gaussian kernel and image using four adjacent scale parameters carry out convolution algorithm, obtain the height of four adjacent yardsticks
This image Gl, wherein l represents four adjacent yardsticks, l=1,2,3,4；
Step 2, Gaussian image G to this four adjacent yardstickslCarry out adjacent yardstick image difference partite transport to calculate, obtain three width
The difference of Gaussian image D of adjacent yardstickl, the yardstick of three this difference image of panel height is respectively：σ,2×σ,2×2×σ；Wherein l=
1,2,3, σ is smoothing parameter；
Step 3, three panel heights this difference image D that step 2 is obtainedlRim detection is carried out using Sobel operator, calculates
Gradient magnitude in level, vertical both direction for each pixel in difference image, and threshold value T is set1, retain gradient magnitude
More than T1Pixel, this pixel is marginal point set its gray value as 255, is otherwise set to 0, obtains three adjacent chis of correspondence
Rim detection binary map E of degreel, wherein l=1,2,3；
Step 4, the three width binary edge figure E to corresponding three different scaleslCarry out multi-scale edge fusion, wherein, l
=1,2,3, concretely comprise the following steps：
(1) search for the difference of Gaussian image D of three adjacent yardsticks under four adjacent yardstick llEdge image ElIn every
One edge pixel, due to adjacent yardstick between edge dislocation be less than 1, in difference Gaussian image D for l-1 for the yardstickl-1Side
The region that corresponding area is 3 × 3 is searched for, all marginal points occurring in this region are all labeled as marginal point, obtain in edge image
To candidate edge image；
(2) l=l-1；If l>1 jumps to step (1), otherwise execution step (3)；
(3) during l=1, edge image ElIt is then the edge image after merging；
Step 5, the edge image after the fusion that step 4 is obtained carries out Morphological scale-space using expanding template, sets
Threshold value T2, connect pel spacing and be less than threshold value T2Marginal point or line, obtain continuous boundary；Carry out closing operation of mathematical morphology again, make up
The hole of edge image and crack, obtain the edge image being further closed；Eventually pass image completion by enclosed region
The filling of portion cavity, forms complete connected region；
Step 6, is marked to connected region, calculates the area of each connected region, setting area threshold value T3, pick
Except area is less than area threshold T3Connected domain；Determine the coordinate of the minimum enclosed rectangle of each connected domain according to connected domain coordinate,
Finally show in original-gray image, complete the detection to vehicle.
Described step one mesoscale is together decided on by smoothing parameter σ and smoothing parameter k, and k is determined by parameter s, and k=2^
(1/S), S+3=N, wherein N are the Gauss picture number in each layer of gaussian pyramid, take N=4, S=1, σ=0.5, k=2, then
Four adjacent yardsticks are respectively σ, k σ, 2k σ, 3k σ.
In described step 3, the template of Sobel operator horizontal direction is (- 1,0,1；-2,0,2；- 1,0,1), vertical direction
Template is (- 1, -2, -1；0,0,0；1,2,1), threshold value T1Obtained by maximum variance between clusters.
Expanding template size in described step 5 is 3 × 3, and form closed operation template size is 8 × 8, T2Span is 1
～8, and in units of pixel.
In described step 6, the setting up procedure of area threshold is：The area of n connected domain of labelling is ranked up, will
The 1/4 of big connected domain area is as area threshold.
Compared with prior art, the device have the advantages that：The invention discloses a kind of be based on the many chis of difference of Gaussian
The vehicle checking method of degree Fusion Edges, gaussian kernel and image first with four adjacent scale parameters carry out convolution algorithm,
Obtain the Gaussian image of four adjacent yardsticks.Then difference is carried out to the Gaussian image of this four adjacent yardsticks, obtain three phases
The difference of Gaussian image of adjacent yardstick simultaneously carries out rim detection to it using Sobel operator.Then three width detection being obtained are different
The edge image of yardstick carries out the Fusion Edges that yardstick is searched for upwards, obtains vehicle edge information as much as possible and removes greatly simultaneously
Amount background edge.Again the edge image merging is carried out expanding, closed operation, a series of morphological operations such as holes filling, obtain
Represent the connection area image of vehicle.Positional information finally according to connected domain determines the outer of vehicle position in original image
Boundary's rectangle, realizes vehicle detection.The present invention processes to the multi-scale image of non-sampled, it is to avoid image interpolation arithmetic causes
Marginal information disappearance or pseudo-edge occurs；More multiple edge is being obtained using the method that multi-scale edge image searches for fusion upwards
Suppress background edge while information.This method reduce algorithm complex, decrease amount of calculation, vehicle detection can be effectively improved
Efficiency, obtain preferable testing result.
Brief description
Fig. 1 is gray level image to be detected；
Fig. 2～4 are the difference of Gaussian images of three adjacent yardsticks；
Fig. 5 is Sobel operator template；
Fig. 6 is the edge detection results under yardstick 1；
Fig. 7 is the edge image after merging；
The connection area image that Fig. 8 obtains for morphology closed operation；
Fig. 9 is the complete connection area image obtaining after holes filling；
Figure 10 is the connection area image representing vehicle；
Figure 11 is rainy day vehicle detection result.
Figure 12 is fine day vehicle detection result.
The vehicle detecting algorithm process schematic that Figure 13 merges for difference of Gaussian multi-scale edge.
Specific embodiment
The present invention provides a kind of vehicle checking method merging based on difference of Gaussian multi-scale edge, is W to a width size
× H vehicle image carries out rim detection under multiple dimensioned, and the edge position information after Multiscale Fusion to determine vehicle location,
Thus realizing vehicle detection.
The present invention is described in detail below in conjunction with the accompanying drawings, and referring to Figure 13, the method for the present invention specifically adopts following several
Individual step is realized：
Step one, gathers certain road section traffic volume video, gathers the vehicle image I that a width size is 517 × 363 in video,
Gray scale figure is obtained to its gray processing, then carries out gaussian pyramid multi-scale transform.Four Gausses using adjacent scale parameter
Core G and image I carries out convolution algorithm, obtains non-sampled Gaussian image G of four adjacent yardsticksl(l=1,2,3,4), wherein l table
Show four adjacent yardsticks.
Yardstick is together decided on by smoothing parameter σ and smoothing parameter k, and smoothing parameter k is determined by parameter s.K=2^ (1/S),
S+3=N, wherein N are the Gauss picture number (according to Lowe paper) in each layer of gaussian pyramid, take N=4, S=in the present invention
1, σ=0.5, k=2, take σ=0.5, S=1, k=2, and four adjacent yardsticks are respectively σ, k σ, 2k σ, 3k σ, i.e. σ, 2 × σ, and 2 × 2
×σ,3×2×σ.
Step 2, yardstick different Gaussian image G identical without the size of down-sampling to this four widthl(l=1,2,3,4)
Carry out adjacent yardstick image difference partite transport to calculate, obtain the difference of Gaussian image D of three adjacent yardsticksl, wherein, l=1,2,3 and three width
The yardstick of difference of Gaussian image is respectively：σ,2×σ,2×2×σ.
Step 3, three panel heights this difference image D that step 2 is obtainedlIt is respectively adopted Sobel operator and carry out rim detection.
Sobel operator passes through to calculate gradient magnitude in level, vertical both direction for each pixel in difference image, using maximum
Inter-class variance (OTSU) method arranges threshold value T1, retain gradient magnitude and be more than T1Pixel be marginal point be set to 1, otherwise put
For 0, obtain edge detection graph E of corresponding three different scalesl(l=1,2,3).
The template of Sobel operator horizontal direction is (- 1,0,1；-2,0,2；- 1,0,1), vertical direction template be (- 1 ,-
2,-1；0,0,0；1,2,1).
Step 4, to three width edge-detected image El(l=1,2,3) carries out multi-scale edge fusion, and step is：
(1) search for difference Gaussian image D under yardstick llEdge image ElEach of edge pixel, due to adjacent
Edge dislocation between yardstick is less than 1, in difference Gaussian image D for l-1 for the yardstickl-1Edge image in search for corresponding face
Amass the region for 3 × 3, all marginal points occurring in this region are all labeled as marginal point, obtain candidate edge image；
(2) l=l-1；If l>1 jumps to step (1), otherwise execution step (3)；
(3) during l=1, edge image ElIt is the edge image obtaining after merging.
Step 5, the edge image that step 4 is obtained carries out Morphological scale-space using expanding template, and referred to as lines expand
Process, connect pel spacing and be less than threshold value T2Marginal point or line, obtain continuous boundary.Carry out closing operation of mathematical morphology again, make up
The hole of edge image and crack, obtain the edge image being further closed.Eventually pass image completion by enclosed region
The filling of portion cavity is got up, and forms complete connected region.
Wherein, expanding template size is 3 × 3, and form closed operation template size is 8 × 8, T2=1～8 (with pixel as list
Position).
Step 6, is marked and calculates the area of each connected region, setting area threshold value T to connected region3, little
In area threshold T3Connected domain be considered as background area and disallowable.Determine that according to connected domain coordinate the minimum of each connected domain is external
The coordinate of rectangle, finally shows in original-gray image, completes the detection to vehicle.
Wherein, area threshold T3Setting：The area of N number of connected domain of labelling is ranked up, takes largest connected domain face
Long-pending 1/4 is as area threshold.
The specific embodiment of the present invention given below, implements in detail below it should be noted that the invention is not limited in
Example, every equal conversion done on the basis of application scheme each falls within protection scope of the present invention.
Embodiment：
Gather the gray processing that a width size is 517 × 363 rainy day traffic images and obtain gray-scale maps, as shown in figure 1, choosing chi
Degree parameter σ=0.5, k=2, be respectively σ with yardstick, four gaussian kernel of 2 × σ, 2 × 2 × σ, 3 × 2 × σ respectively with gray-scale maps
Carry out convolution algorithm, obtain the non-sampled Gaussian Blur image of four adjacent yardsticks.
To four adjacent yardsticks without down-sampling size is identical and Gaussian Blur image that yardstick is different carry out adjacent
Yardstick calculus of differences obtains the difference of Gaussian image of three adjacent yardsticks, and that is, yardstick 1 and the Gaussian image difference of yardstick 2 obtain chi
Spend the Gaussian difference component being, figure yardstick 2 obtains the difference diagram that yardstick is, yardstick 3 and yardstick 4 with the Gaussian image difference of yardstick 3
Gaussian image difference obtain the difference diagram that yardstick is.Gaussian difference component is as shown in figs. 2 to 4.
Using horizontal direction template (- 1,0,1 as shown in Figure 5；-2,0,2；- 1,0,1), vertical direction template (- 1 ,-
2,-1；0,0,0；1,2,1) Sobel operator carries out rim detection to this difference diagram of three panel heights, obtains the side of three adjacent yardsticks
Edge detection figure, the edge graph of yardstick 1 is as shown in Figure 6.Centered on each of yardstick 3 edge graph marginal point, in yardstick 2
In edge graph, its 3 × 3 neighborhood is scanned for, retain the marginal point in neighborhood, terminate to search, after obtaining merging for the first time
Edge graph.To each of this edge graph marginal point, yardstick be 1 edge graph in proceed above-mentioned same operation, obtain
To final Fusion Edges image, as shown in Figure 7.
Take T2=3, with 3 × 3 templates, morphological dilations are carried out to edge fusion image, then carries out morphology with 8 × 8 templates
Closed operation, obtains the connected domain of edge closure, as shown in Figure 8.Image completion is carried out to Fig. 8, makes the hole in connected domain up, obtain
To complete connection area image (as shown in Figure 9).
Fig. 9 is carried out with connected component labeling (labelling result is 16), calculates the area of each connected domain and by from big to small
Order sorts, and with the 1/4 of largest connected domain area as threshold value (being 2300 in this example), rejects the connected domain that area is less than threshold value,
The connection area image (as shown in Figure 10) obtaining is considered as vehicle position.Relevant position draws the external square of connected domain in FIG
Shape, obtains rainy day vehicle detection result, as shown in figure 11.Figure 12 is processed using the present invention to another scene (fine day) and obtains
Testing result.
From Figure 11 and 12 as can be seen that carrying out vehicle detection according to the method described above it is achieved that preferable testing result.This reality
Example shows, the solution of the present invention algorithm is simple, decreases amount of calculation, is simultaneously achieved preferable vehicle detection.
The invention discloses a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge, first with four
The gaussian kernel of adjacent scale parameter and image carry out convolution algorithm, obtain the Gaussian image of four adjacent yardsticks.Then to this four
The Gaussian image of individual adjacent yardstick carries out difference, obtains the difference of Gaussian image of three adjacent yardsticks and it is adopted Sobel calculate
Son carries out rim detection.Then the edge image of three adjacent yardsticks detection being obtained carries out the edge that yardstick searches for upwards and melts
Close, obtain vehicle edge information as much as possible and remove a large amount of background edge simultaneously.Again the edge image merging is carried out expanding,
A series of morphological operation such as closed operation, holes filling, obtains the connection area image representing vehicle.Position finally according to connected domain
Confidence breath determines the extraneous rectangle of vehicle position in original image, realizes vehicle detection.The present invention is many to non-sampled
Scalogram picture processed, it is to avoid marginal information disappearance that image interpolation arithmetic causes or pseudo-edge；Using multiple dimensioned
The method that edge image searches for fusion upwards suppresses background edge while obtaining more marginal information.This method reduce calculation
Method complexity, decreases amount of calculation, can effectively improve the efficiency of vehicle detection, obtains preferable testing result.
Claims (5)
1. a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge is it is characterised in that comprise the following steps：
Step one, gathers certain road section traffic volume video, obtains original-gray image to the piece image gray processing in video, then carries out
Gaussian pyramid multi-scale transform, gaussian kernel and image using four adjacent scale parameters carry out convolution algorithm, obtain four width
Gaussian image G of adjacent yardstickl, wherein l represents four adjacent yardsticks, l=1,2,3,4；
Step 2, Gaussian image G to this four adjacent yardstickslCarry out adjacent yardstick image difference partite transport to calculate, obtain three adjacent chis
The difference of Gaussian image D of degreel, the yardstick of three this difference image of panel height is respectively：σ,2×σ,2×2×σ；Wherein l=1,2,3, σ
For smoothing parameter；
Step 3, three panel heights this difference image D that step 2 is obtainedlRim detection is carried out using Sobel operator, calculates difference
Gradient magnitude in level, vertical both direction for each pixel in image, and threshold value T is set1, retain gradient magnitude and be more than
Threshold value T1Pixel, this pixel is marginal point set its gray value as 255, is otherwise set to 0, obtains three adjacent chis of correspondence
Rim detection binary map E of degreel, wherein l=1,2,3；
Step 4, the three width binary edge figure E to corresponding three different scaleslCarry out multi-scale edge fusion, wherein, l=1,2,
3, concretely comprise the following steps：
(1) search for the difference of Gaussian image D of three adjacent yardsticks under four adjacent yardstick llEdge image ElEach of
Edge pixel, due to adjacent yardstick between edge dislocation be less than 1, in difference Gaussian image D for l-1 for the yardstickl-1Edge graph
The region that corresponding area is 3 × 3 is searched for, all marginal points occurring in this region are all labeled as marginal point, are waited in picture
Select edge image；
(2) l=l-1；If l ＞ 1, jump to step (1), otherwise execution step (3)；
(3) during l=1, edge image ElIt is then the edge image after merging；
Step 5, the edge image after the fusion that step 4 is obtained carries out Morphological scale-space, given threshold using expanding template
T2, connect pel spacing and be less than threshold value T2Marginal point or line, obtain continuous boundary；Carry out closing operation of mathematical morphology again, make edge up
The hole of image and crack, obtain the edge image being further closed；Eventually pass image completion the inside of enclosed region is empty
Hole is filled, and forms complete connected region；
Step 6, is marked to connected region, calculates the area of each connected region, setting area threshold value T3, reject area
Less than area threshold T3Connected region；Determine the coordinate of the minimum enclosed rectangle of each connected region according to connected region coordinate,
Finally show in original-gray image, complete the detection to vehicle.
2. a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge according to claim 1, its feature
It is, described step one mesoscale is together decided on by smoothing parameter σ and smoothing parameter k, and k is determined by parameter s, and k=2^ (1/
S), S+3=N, wherein N are the Gauss picture number in each layer of gaussian pyramid, take N=4, S=1, σ=0.5, k=2, then four
Individual adjacent yardstick is respectively σ, k σ, 2k σ, 3k σ.
3. a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge according to claim 1, its feature
It is, in described step 3, the template of Sobel operator horizontal direction is (- 1,0,1；-2,0,2；- 1,0,1), vertical direction template
For (- 1, -2, -1；0,0,0；1,2,1), threshold value T1Obtained by maximum variance between clusters.
4. a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge according to claim 1, its feature
It is, expanding template size in described step 5 is 3 × 3, and form closed operation template size is 8 × 8, T2Span be 1～
8, and in units of pixel.
5. a kind of vehicle checking method being merged based on difference of Gaussian multi-scale edge according to claim 1, its feature
It is, in described step 6, the setting up procedure of area threshold is：The area of n connected region of labelling is ranked up, will
The 1/4 of big connection region area is as area threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410181851.0A CN103927526B (en) | 2014-04-30 | 2014-04-30 | Vehicle detecting method based on Gauss difference multi-scale edge fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410181851.0A CN103927526B (en) | 2014-04-30 | 2014-04-30 | Vehicle detecting method based on Gauss difference multi-scale edge fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103927526A CN103927526A (en) | 2014-07-16 |
| CN103927526B true CN103927526B (en) | 2017-02-15 |
Family
ID=51145742
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410181851.0A Active CN103927526B (en) | 2014-04-30 | 2014-04-30 | Vehicle detecting method based on Gauss difference multi-scale edge fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103927526B (en) |
Families Citing this family (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104598906B (en) * | 2015-03-04 | 2017-08-22 | 西安工业大学 | Vehicle outline detection method and its device |
| CN105956532B (en) * | 2016-04-25 | 2019-05-21 | 大连理工大学 | A kind of traffic scene classification method based on multiple dimensioned convolutional neural networks |
| CN107730521B (en) * | 2017-04-29 | 2020-11-03 | 安徽慧视金瞳科技有限公司 | Method for rapidly detecting ridge type edge in image |
| CN107240111B (en) * | 2017-06-14 | 2021-03-26 | 郑州天迈科技股份有限公司 | Edge communication segmentation passenger flow statistical method |
| CN107492110A (en) * | 2017-07-31 | 2017-12-19 | 腾讯科技（深圳）有限公司 | A kind of method for detecting image edge, device and storage medium |
| CN108629781B (en) * | 2018-04-24 | 2022-04-22 | 成都品果科技有限公司 | Hair drawing method |
| CN108596931A (en) * | 2018-05-11 | 2018-09-28 | 浙江大学 | A kind of noise robustness hub edge detection algorithm based on Canny operators |
| CN108957530B (en) * | 2018-05-23 | 2019-08-23 | 电子科技大学 | A kind of crack automatic testing method based on Acceleration Algorithm in Seismic Coherence Cube slice |
| CN108830169B (en) * | 2018-05-24 | 2020-10-09 | 中国农业大学 | Method and system for detecting working state of aerator |
| CN109242884B (en) * | 2018-08-14 | 2020-11-20 | 西安电子科技大学 | Remote sensing video target tracking method based on JCFNet network |
| CN109389617A (en) * | 2018-08-27 | 2019-02-26 | 深圳大学 | A kind of motion estimate based on piece heterogeneous system and method for tracing and system |
| CN109559299B (en) * | 2018-11-19 | 2023-07-25 | 常州大学 | Method for separating double-fruit overlapped fruits |
| CN109543487B (en) * | 2018-11-22 | 2022-04-01 | 福州图腾易讯信息技术有限公司 | Automatic induction triggering method and system based on bar code edge segmentation |
| CN109741406A (en) * | 2019-01-03 | 2019-05-10 | 广州广电银通金融电子科技有限公司 | A kind of body color recognition methods under monitoring scene |
| CN109784322A (en) * | 2019-01-04 | 2019-05-21 | 平安科技（深圳）有限公司 | A kind of recognition methods of vin code, equipment and medium based on image procossing |
| CN109886312B (en) * | 2019-01-28 | 2023-06-06 | 同济大学 | Bridge vehicle wheel detection method based on multilayer feature fusion neural network model |
| CN111915496B (en) * | 2019-05-08 | 2024-04-23 | 北京字节跳动网络技术有限公司 | Image processing method, device and storage medium |
| CN111324145B (en) * | 2020-02-28 | 2022-08-16 | 厦门理工学院 | Unmanned aerial vehicle autonomous landing method, device, equipment and storage medium |
| CN111353446A (en) * | 2020-03-05 | 2020-06-30 | 广智微芯(扬州)有限公司 | Lane line detection method and system |
| CN111626304B (en) * | 2020-05-20 | 2023-08-04 | 中国科学院新疆理化技术研究所 | Color feature extraction method based on machine vision and application thereof |
| CN112329677B (en) * | 2020-11-12 | 2024-02-02 | 北京环境特性研究所 | Remote sensing image river channel target detection method and device based on feature fusion |
| CN112520524A (en) * | 2021-01-13 | 2021-03-19 | 大连理工大学 | Intelligent elevator control interaction device and control method based on gesture recognition technology |
| CN112819843B (en) * | 2021-01-20 | 2022-08-26 | 上海大学 | Method and system for extracting power line at night |
| CN113269800B (en) * | 2021-05-25 | 2023-05-12 | 华能海南发电股份有限公司澄迈光伏电站 | Infrared image segmentation method for solar photovoltaic module |
| CN114494318B (en) * | 2022-02-17 | 2023-05-23 | 天津市眼科医院 | Cornea contour extraction method based on cornea dynamic deformation video of Ojin algorithm |
| CN114972276A (en) * | 2022-06-05 | 2022-08-30 | 长沙烽铭智能科技有限公司 | Automatic driving distance judgment algorithm for vehicle |
| CN115775250B (en) * | 2023-02-13 | 2023-05-30 | 惠州威尔高电子有限公司 | Golden finger circuit board defect rapid detection system based on digital image analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520894A (en) * | 2009-02-18 | 2009-09-02 | 上海大学 | Method for extracting significant object based on region significance |
| CN102289806A (en) * | 2011-06-21 | 2011-12-21 | 北京航空航天大学 | Method for measuring image definition by utilizing multi-scale morphological characteristics |
| CN103714538A (en) * | 2013-12-20 | 2014-04-09 | 中联重科股份有限公司 | Road edge detection method, device and vehicle |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2071513A1 (en) * | 2007-12-10 | 2009-06-17 | Agfa HealthCare NV | Method of generating a multiscale contrast enhanced image |
-
2014
- 2014-04-30 CN CN201410181851.0A patent/CN103927526B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520894A (en) * | 2009-02-18 | 2009-09-02 | 上海大学 | Method for extracting significant object based on region significance |
| CN102289806A (en) * | 2011-06-21 | 2011-12-21 | 北京航空航天大学 | Method for measuring image definition by utilizing multi-scale morphological characteristics |
| CN103714538A (en) * | 2013-12-20 | 2014-04-09 | 中联重科股份有限公司 | Road edge detection method, device and vehicle |
Non-Patent Citations (1)
| Title |
|---|
| 基于多尺度特征的图像匹配与目标定位研究;刘立;《中国博士学位论文全文数据库 信息科技辑》;20090515(第 05 期);I138-58 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103927526A (en) | 2014-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103927526B (en) | Vehicle detecting method based on Gauss difference multi-scale edge fusion | |
| CN107507167B (en) | Cargo tray detection method and system based on point cloud plane contour matching | |
| CN105160309B (en) | Three lanes detection method based on morphological image segmentation and region growing | |
| Gong et al. | The recognition and tracking of traffic lights based on color segmentation and camshift for intelligent vehicles | |
| CN102663760B (en) | Location and segmentation method for windshield area of vehicle in images | |
| CN110598690B (en) | End-to-end optical character detection and recognition method and system | |
| Gomez et al. | Traffic lights detection and state estimation using hidden markov models | |
| Danescu et al. | Detection and classification of painted road objects for intersection assistance applications | |
| CN102509098B (en) | Fisheye image vehicle identification method | |
| CN109726717B (en) | Vehicle comprehensive information detection system | |
| WO2015147764A1 (en) | A method for vehicle recognition, measurement of relative speed and distance with a single camera | |
| CN105404857A (en) | Infrared-based night intelligent vehicle front pedestrian detection method | |
| CN107256633B (en) | Vehicle type classification method based on monocular camera three-dimensional estimation | |
| Liu et al. | ROI perspective transform based road marking detection and recognition | |
| CN107909047A (en) | A kind of automobile and its lane detection method and system of application | |
| CN104463138A (en) | Text positioning method and system based on visual structure attribute | |
| CN107704833A (en) | A kind of front vehicles detection and tracking based on machine learning | |
| CN113837094A (en) | Road condition rapid analysis method based on full-color high-resolution remote sensing image | |
| Ahmed et al. | Traffic sign detection and recognition model using support vector machine and histogram of oriented gradient | |
| Koutaki et al. | Automatic road extraction based on cross detection in suburb | |
| Hernández et al. | Lane marking detection using image features and line fitting model | |
| Coronado et al. | Detection and classification of road signs for automatic inventory systems using computer vision | |
| Huang | Research on license plate image segmentation and intelligent character recognition | |
| Getahun et al. | A robust lane marking extraction algorithm for self-driving vehicles | |
| Phu et al. | Traffic sign recognition system using feature points |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |