CN109993991A - Parking stall condition detection method and system - Google Patents
Parking stall condition detection method and system Download PDFInfo
- Publication number
- CN109993991A CN109993991A CN201811456794.7A CN201811456794A CN109993991A CN 109993991 A CN109993991 A CN 109993991A CN 201811456794 A CN201811456794 A CN 201811456794A CN 109993991 A CN109993991 A CN 109993991A
- Authority
- CN
- China
- Prior art keywords
- image
- parking stall
- parking
- feature
- differentiation
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 48
- 230000004069 differentiation Effects 0.000 claims abstract description 71
- 238000000605 extraction Methods 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims description 24
- 238000003709 image segmentation Methods 0.000 claims description 10
- 238000012512 characterization method Methods 0.000 claims description 9
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000005260 corrosion Methods 0.000 claims description 6
- 230000007797 corrosion Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 230000000877 morphologic effect Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 238000000034 method Methods 0.000 abstract description 8
- 238000004364 calculation method Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- 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
- G06T7/344—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving models
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/04—Detecting movement of traffic to be counted or controlled using optical or ultrasonic detectors
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30264—Parking
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of parking stall condition detection method and systems, image creation shape Matching Model under the parking lot idle state that wherein method passes through acquisition, simultaneously, according to the image information on parking stall contour line extraction parking stall, discrimination threshold of several differentiation features as actual conditions is therefrom obtained;The image in practical parking scene is registrated using the shape Matching Model being pre-created after obtaining the image in practical parking scene;The image on one parking stall of regional choice in image after registration, the corresponding discrimination threshold of differentiation feature for obtaining the image are compared;The particular state of current parking stall is determined according to comparison result.As corresponding discrimination threshold, i.e. the corresponding discrimination threshold in a parking stall efficiently avoids in actually detected the differentiation feature that the present invention is gone out by the correspondence parking stall measure of the parking field picture under idle state to greatly improve the accuracy of detection.
Description
Technical field
The present invention relates to digital image processing techniques field, in particular to a kind of parking stall condition detection method and system.
Background technique
For stopping at release pressure, while parking efficiency is promoted, Intelligent parking lot management system comes into being.Its mainly by
Three parts form: collecting vehicle information, parking space information detection and control centre.Wherein the detection of free parking space is parking stall
An important link in the task of information detecting module and the system.It can greatly save car owner and not know situation
In the state of blindly find time of parking stall.
Currently, the method for related parking space state detection mainly has two major classes, one kind is traditional sensor-based physics
Feature detection techniques, another kind of is the image detecting technique based on video.Traditional physical features detection technique specifically includes that
(1) the induction coil detection technique of changes of magnetic field situation is utilized;(2) pass through the sound wave parking stall measure technology of reflection echo;(3) sharp
With the dynamic weighing detection technique etc. of detector deformation.Although this kind of detection method is with low content of technology, due to needing on ground
It constructs under face, so the cost of installation and maintenance is bigger.
And traditional physical features detection technique is compared, the detection technique based on video image then efficiently avoids above-mentioned
Problem.But found in the experimentation of parking space state detection, if directly to the practical parking field picture stopped under scene into
The detection of row parking space state, then due to the difference of camera site and the presence etc. of vehicle, parking stall line, which can exist, to be blocked
The problem of, thus can not determine the specific parking stall band of position, large effect is generated to actual detection.Simultaneously as clapping
It acting as regent and sets the imaging difference of generation, each parking area of image has certain difference, thus when carrying out state judgement, if setting
Fixed discrimination threshold is set, then can generate biggish detection error.
Summary of the invention
The present invention provides a kind of parking stall condition detection method and system, solves the problems, such as existing above-mentioned.
To solve the above problems, the embodiment of the present invention provides a kind of parking stall condition detection method, comprising the following steps:
Obtain the image under the idle state of parking lot;
According to the parking stall for extracting several only in the image of parking stall contour line in an idle state and include a parking stall
The image of information obtains several differentiation features from each only image comprising a parking stall, and is obtained according to differentiation feature
Often from the discrimination threshold of feature;
According to the image creation shape Matching Model under idle state,
Obtain the image in practical parking scene;
The image in practical parking scene is registrated using the shape Matching Model being pre-created;
One is selected in image after registration and only includes the image on a parking stall, and obtains several differentiations of the image
Feature;
The corresponding discrimination threshold of the differentiation feature that will acquire is compared;
It is each to differentiate that feature includes the state feature of at least two characterization parking stall states, it is based on using ballot thought,
It votes each differentiation feature, using the more state feature of poll as the particular state of current parking stall.
As an implementation, it is described according to parking stall contour line extraction out only comprising the parking space information on parking stall
Image, specifically includes the following steps:
Gray processing processing is carried out to the RGB image of acquisition;
Noise is removed to gray level image using gaussian filtering;
Image enhancement is carried out using grey linear transformation, is translated into bianry image;
Bianry image is optimized using the morphological operation of expansion and corrosion, image point is carried out to the image after optimization
It cuts;
The selection for carrying out characteristic area to the image after segmentation using feature histogram, filters out parking area;
Skeletal extraction is carried out to parking area, and is translated into sub-pix profile;
Sub-pix profile is split, and carries out feature selecting using direction and length as parameter, obtains final parking stall
Line profile;
Straight line fitting is carried out to parking stall line profile, and obtains corresponding starting point coordinate parameter information, is only wrapped to generate
The image of parking space information containing a parking stall;
Several differentiation features are obtained from the image on parking stall, and are obtained according to differentiation feature often from the differentiation threshold of feature
Value.
As an implementation, the image creation shape Matching Model according under idle state, specifically include with
Lower step:
Shape conversion, and selected section region are carried out to the image of parking area, according to the corresponding grayscale image in partial region
As creation shape Matching Model.
As an implementation, further comprising the steps of:
Gray processing processing is carried out to the image obtained in practical parking scene, is translated into gray level image.
As an implementation, it is described differentiate feature quantity be three, respectively comentropy, grey-scale contrast and
Harris angle point number.
As an implementation, the corresponding discrimination threshold of the differentiation feature that will acquire is compared, each
Differentiate that feature includes the state feature of at least two characterization parking stall states, based on ballot thought is utilized, to each differentiation spy
Sign is voted, using the more state feature of poll as the particular state of current parking stall, specifically includes the following steps:
Comentropy that No. i-th parking stall of the image under the idle state of parking lot is extracted, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy [i], Contrast [i] and Points [i], corresponding
Discrimination threshold E [i], C [i] and P [i] are respectively as follows:
E [i]=(Entropy [i]+2) * 0.9;
C [i]=(Contrast [i]+10) * 0.9;
P [i]=(Points [i]+15) * 0.95;
Actually parking scene in image No. i-th parking stall extract comentropy, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy1 [I], Contrast1 [I] and Points1 [I], are corresponded to
Comparison formula are as follows:
Entropy1[I]>E[i];
Contrast1[I]>C[i];
Points1[I]>P[i];
When the comparison formula for differentiating feature is set up, then it is determined as occupied state, is otherwise determined as idle state;
When two and when more than two relatively formula establishments, then determine that the parking stall for occupied state, is otherwise determined as this
Parking stall is idle state.
The present invention also provides a kind of parking stall condition detecting systems, comprising:
Module is obtained, for obtaining the image in the image under the idle state of parking lot and practical parking scene;
Discrimination threshold module includes the parking space information on a parking stall for going out several only according to parking stall contour line extraction
Image, obtain several differentiation features from each only image comprising a parking stall, and according to differentiate feature obtain often from
The discrimination threshold of feature;
Model creation module, for according to the image creation shape Matching Model under idle state;
Registration module, for being matched using the shape Matching Model being pre-created to the image in practical parking scene
It is quasi-;
Characteristic extracting module only includes the image on a parking stall for selecting one in image after registration, and obtains
Take several differentiation features of the image;
Vote module, the corresponding discrimination threshold of differentiation feature for will acquire are compared, each differentiation feature
State feature comprising at least two characterization parking stall states throws each differentiation feature based on using ballot thought
Ticket, using the more state feature of poll as the particular state of current parking stall.
As an implementation, the discrimination threshold module includes:
Gray processing unit, for carrying out gray processing processing to the RGB image of acquisition;
Unit is denoised, for being removed noise to gray level image using gaussian filtering;
Binarization unit is translated into bianry image for carrying out image enhancement using grey linear transformation;
Image segmentation unit, for being optimized using the morphological operation of expansion and corrosion to bianry image, to optimization
Image afterwards carries out image segmentation;
Screening unit filters out vehicle for carrying out the selection of characteristic area to the image after segmentation using feature histogram
Position region;
Skeletal extraction unit for carrying out skeletal extraction to parking area, and is translated into sub-pix profile;
Cutting unit carries out feature selecting for being split to sub-pix profile, and using direction and length as parameter, obtains
To final parking stall line profile;
Parking space information unit for carrying out straight line fitting to parking stall line profile, and obtains corresponding starting point coordinate parameter
Information, to generate the image of the only parking space information comprising a parking stall;
Threshold cell obtains often certainly for obtaining several differentiation features from the image on parking stall, and according to differentiation feature
The discrimination threshold of feature.
As an implementation, the model creation module includes:
Creating unit carries out shape conversion, and selected section region for the image to parking area, according to partial region
Corresponding gray level image creates shape Matching Model.
As an implementation, it is described differentiate feature quantity be three, respectively comentropy, grey-scale contrast and
Harris angle point number.
The beneficial effect of the present invention compared with the prior art is: passing through the correspondence vehicle of the parking field picture under idle state
The differentiation feature that position detects is as corresponding discrimination threshold, the i.e. corresponding discrimination threshold in a parking stall, to greatly mention
Risen detection accuracy, efficiently avoid it is actually detected present in parking stall line occlusion issue.Meanwhile to practical parking
In the case of parking field picture carry out state-detection before, utilize parking stall location information detection-phase creation shape Matching Model
Image registration is carried out to it, the discrimination threshold of the parking stall location information extracted in advance in this way and setting just has practical significance.
In addition, in order to promote the Stability and veracity of condition discrimination, this algorithm is extracted comentropy, grey-scale contrast and Harris
Three differentiation features of angle point number, using ballot thought, each feature possesses a ticket, when " occupancy " or " free time " state obtain
When two tickets or more, then it is assumed that the state is the corresponding specific state in parking stall.
Detailed description of the invention
Fig. 1 is the flow chart of parking stall condition detection method of the invention;
Fig. 2 is the parking field picture in parking stall condition detection method of the invention under idle state;
Fig. 3 is the parking field picture in parking stall condition detection method of the invention under actual conditions;
Fig. 4 is the partial process view of the step S200 of parking stall condition detection method of the invention;
Fig. 5 is the parking area image in the step S501 of parking stall condition detection method of the invention;
Fig. 6 is the module connection figure of parking stall condition detecting system of the invention;
Fig. 7 is the unit connection figure of the discrimination threshold module of parking stall condition detecting system of the invention.
Attached drawing mark: 1, module is obtained;2, discrimination threshold module;21, gray processing unit;22, unit is denoised;23, two-value
Change unit;24, image segmentation unit;25, screening unit;26, skeletal extraction unit;27, cutting unit;28, parking space information list
Member;29, threshold cell;3, model creation module;4, registration module;5, characteristic extracting module;6, vote module.
Specific embodiment
Below in conjunction with attached drawing, the technical characteristic and advantage above-mentioned and other to the present invention are clearly and completely described,
Obviously, described embodiment is only section Example of the invention, rather than whole embodiments.
As shown in Figure 1, a kind of parking stall condition detection method, comprising the following steps:
S100: the image under the idle state of parking lot is obtained, sees Fig. 2;
S200: including a parking stall according to several are extracted only in parking stall contour line image in an idle state
The image of parking space information obtains several differentiation features from each only image comprising a parking stall, and according to differentiation feature
It obtains often from the discrimination threshold of feature;
S300: according to the image creation shape Matching Model under idle state,
S400: the image in practical parking scene is obtained, sees Fig. 3;
S500: being registrated the image in practical parking scene using the shape Matching Model being pre-created, i.e., will be real
Border Image Adjusting is at the image as shape Matching Model size, angle;
S600: one is selected in image after registration only comprising the image on a parking stall, and if obtaining the image
It is dry to differentiate feature;
S700: the corresponding discrimination threshold of the differentiation feature that will acquire is compared;
S800: it is each to differentiate that feature includes the state feature of at least two characterization parking stall states, it is voted based on utilizing
Thought votes to each differentiation feature, using the more state feature of poll as the particular state of current parking stall, at this
In embodiment, state feature mainly includes occupying and two kinds of the free time.It is i.e. each to differentiate that feature includes " accounting for for characterization parking stall state
With " and " free time " two states feature, after each differentiation feature is compared with its discrimination threshold, a kind of state feature can be generated,
After all differentiation features are completeer, determine a fairly large number of state feature for the parking space state of current parking stall.
It is further comprising the steps of between step S400 and step S500:
Gray processing processing is carried out to the image obtained in practical parking scene, is translated into gray level image.
Wherein, as shown in figure 4, according to parking stall contour line extraction out only comprising a parking stall parking space information image,
Specifically includes the following steps:
S101: gray processing processing is carried out to the RGB image of acquisition;
S201: noise is removed to gray level image using gaussian filtering;
S301: image enhancement is carried out using grey linear transformation, is translated into bianry image;
S401: optimizing bianry image using the morphological operation of expansion and corrosion, carries out to the image after optimization
Image segmentation;
S501: the selection of characteristic area is carried out to the image after segmentation using feature histogram, parking area is filtered out, sees
Fig. 5;
S601: skeletal extraction is carried out to parking area, and is translated into sub-pix profile;
S701: being split sub-pix profile, and carries out feature selecting using direction and length as parameter, obtains final
Parking stall line profile;
S801: straight line fitting is carried out to parking stall line profile, and obtains corresponding starting point coordinate parameter information, to generate
The only image of the parking space information comprising a parking stall;
S901: several differentiation features are obtained from the image on parking stall, and often sentencing from feature is obtained according to differentiation feature
Other threshold value.
According to the image creation shape Matching Model under idle state, specifically includes the following steps:
Shape conversion, and selected section region are carried out to the image of parking area, according to the corresponding grayscale image in partial region
As creation shape Matching Model.
Imaging difference caused by due to shooting orientation and object distance etc. are different, the shape of each parking stall in the picture
It is all different, so as to cause its corresponding characteristic parameter, there is also differences, so needing to be determined according to specific parking stall
Specific discrimination threshold, to reduce differentiation error.In the present embodiment, differentiate feature quantity be three, respectively comentropy,
Grey-scale contrast and Harris angle point number.It is specific to differentiate that process is as follows:
Comentropy that No. i-th parking stall of the image under the idle state of parking lot is extracted, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy [i], Contrast [i] and Points [i], corresponding
Discrimination threshold E [i], C [i] and P [i] are respectively as follows:
E [i]=(Entropy [i]+2) * 0.9;
C [i]=(Contrast [i]+10) * 0.9;
P [i]=(Points [i]+15) * 0.95;
Actually parking scene in image No. i-th parking stall extract comentropy, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy1 [I], Contrast1 [I] and Points1 [I], are corresponded to
Comparison formula are as follows:
Entropy1[I]>E[i];
Contrast1[I]>C[i];
Points1[I]>P[i];
When the comparison formula for differentiating feature is set up, then it is determined as occupied state, is otherwise determined as idle state;
When two and when more than two relatively formula establishments, then determine that the parking stall for occupied state, is otherwise determined as this
Parking stall is idle state.
Comentropy: a kind of statistical form of feature is the important indicator measured image information and enrich degree, calculates
Formula isWherein, piIt is the probability for the pixel appearance that gray value is i in image, L is that gray level is total
Number (usual value 256).Comentropy D (f) is bigger, illustrates that amount of image information is bigger.By the definition of comentropy it is found that when on parking stall
In the presence of having vehicle, information content is larger, and comentropy is also larger.It is tested and is shown by many experiments, the Entropy of idle state parking space
For the difference of [i] and E [i] less than 1, the difference of the Entropy [i] and E [i] of occupied state parking space are greater than 2.Therefore, comentropy
Compare and Entropy [i]+2 is set in formula for initial discrimination threshold.In view of error present in detection, setting 0.9 is permission
Error carry out discrimination threshold adjustment.
Grey-scale contrast: the ratio of image black and white, i.e. the gradual change level from black to white, calculation formula are as follows: C=∑δδ
(i,j)2pδ(i, j), wherein δ (i, j)=| i-j |, i.e. gray scale difference between adjacent pixel i and j;pδ(i, j) is δ (i's, j)
Pixel distribution probability, C is bigger, shows that the level from black to white is more, color representation is abundanter.When the parking stall free time, in parking stall
Pixel grey scale variation is smaller, and contrast is smaller;When parking stall is occupied state, level of the image from black to white is more, color table
Now abundanter, grey-scale contrast is bigger.Show the grey-scale contrast of idle state parking space less than 5 by experiment test;Work as parking stall
When occupancy, the difference of Contrast [i] and C [i] are generally higher than 15.Therefore, grey-scale contrast compares formula setting Contrast
[i]+10 is initial threshold, it is contemplated that error when detection, setting 0.9 are that the error allowed carries out the adjustment of discrimination threshold.
Harris angle point: there are two the characteristic point of principal direction, harris angle points for tool in the intersection point or neighborhood at two edges
Calculation formula it is as follows:
Corner=det (M)-λ * trace (M)
Wherein, G indicates Gaussian smoothing, Ix,cAnd Iy,cIt is image in the gradient of X and Y-direction, det (M) is the ranks of matrix M
Formula, trace (M) are the mark of matrix M, and λ is constant.In the presence of having vehicle in parking stall, the harris angle point arrived to vehicle detection is a
The harris angle point number of parking stall when number is much larger than idle state.It is found by abundant experimental results, the parking stall detected when idle
Harris angle point number less than 10, the harris angle point number of parking stall is generally higher than 20 when occupancy.For the free time to parking stall
There is a specific differentiation with occupied state, setting 15 is differentiation difference.Simultaneously in view of the error of detection, setting 0.95 is fair
Perhaps error carries out the adjustment of discrimination threshold.
As shown in fig. 6, a kind of parking stall condition detecting system, including obtain module 1, model creation module 3, discrimination threshold
Module 2, registration module 4, characteristic extracting module 5 and vote module 6 obtain module 1 for obtaining under the idle state of parking lot
Image and practical parking scene in image;Discrimination threshold module 2 is only wrapped for going out several according to parking stall contour line extraction
The image of parking space information containing a parking stall obtains several differentiation features from each only image comprising a parking stall,
And it is obtained according to differentiation feature often from the discrimination threshold of feature;Model creation module 3 is used to be created according to the image under idle state
Build shape Matching Model;Registration module 4 is used for using the shape Matching Model being pre-created to the image in practical parking scene
It is registrated;Characteristic extracting module 5 only includes the image on a parking stall for selecting one in image after registration, and obtains
Take several differentiation features of the image;The corresponding discrimination threshold of the differentiation feature that vote module 6 is used to will acquire is compared
Compared with, it is each to differentiate that feature includes the state feature of at least two characterization parking stall states, based on ballot thought is utilized, to each
Differentiate that feature is voted, using the more state feature of poll as the particular state of current parking stall.
Wherein, as shown in fig. 7, discrimination threshold module 2 include gray processing unit 21, denoising unit 22, binarization unit 23,
Image segmentation unit 24, screening unit 25, skeletal extraction unit 26, cutting unit 27, parking space information unit 28 and threshold value list
Member 29, gray processing unit 21 are used to carry out gray processing processing to the RGB image of acquisition;Unit 22 is denoised to be used to utilize gaussian filtering
Noise is removed to gray level image;Binarization unit 23 is used to carry out image enhancement using grey linear transformation, is converted
For bianry image;Image segmentation unit 24 is used to optimize bianry image using the morphological operation of expansion and corrosion, right
Image after optimization carries out image segmentation;Screening unit 25 is used to carry out characteristic area to the image after segmentation using feature histogram
The selection in domain, filters out parking area;Skeletal extraction unit 26 is used to carry out skeletal extraction to parking area, and is translated into
Sub-pix profile;Cutting unit 27 carries out feature choosing for being split to sub-pix profile, and using direction and length as parameter
It selects, obtains final parking stall line profile;Parking space information unit 28 is used to carry out parking stall line profile straight line fitting, and obtains corresponding
Starting point coordinate parameter information, thus generate only comprising a parking stall parking space information image;Threshold cell 29 is used for
Several differentiation features are obtained from the image on parking stall, and are obtained according to differentiation feature often from the discrimination threshold of feature.
Model creation module 3 includes creating unit, and creating unit is used to carry out shape conversion to the image of parking area, and
Selected section region creates shape Matching Model according to the corresponding gray level image in partial region;
The quantity for differentiating feature is three, respectively comentropy, grey-scale contrast and Harris angle point number.
Imaging difference caused by due to shooting orientation and object distance etc. are different, the shape of each parking stall in the picture
It is all different, so as to cause its corresponding characteristic parameter, there is also differences, so needing to be determined according to specific parking stall
Specific discrimination threshold, to reduce differentiation error.In the present embodiment, differentiate feature quantity be three, respectively comentropy,
Grey-scale contrast and Harris angle point number.The specific differentiation process of vote module 6 is as follows:
Comentropy that No. i-th parking stall of the image under the idle state of parking lot is extracted, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy [i], Contrast [i] and Points [i], corresponding
Discrimination threshold E [i], C [i] and P [i] are respectively as follows:
E [i]=(Entropy [i]+2) * 0.9;
C [i]=(Contrast [i]+10) * 0.9;
P [i]=(Points [i]+15) * 0.95;
Actually parking scene in image No. i-th parking stall extract comentropy, grey-scale contrast and
Three differentiation features of Harris angle point number are respectively Entropy1 [I], Contrast1 [I] and Points1 [I], are corresponded to
Comparison formula are as follows:
Entropy1[I]>E[i];
Contrast1[I]>C[i];
Points1[I]>P[i];
When the comparison formula for differentiating feature is set up, then it is determined as occupied state, is otherwise determined as idle state;
When two and when more than two relatively formula establishments, then determine that the parking stall for occupied state, is otherwise determined as this
Parking stall is idle state.
Comentropy: a kind of statistical form of feature is the important indicator measured image information and enrich degree, calculates
Formula isWherein, piIt is the probability for the pixel appearance that gray value is i in image, L is that gray level is total
Number (usual value 256).Comentropy D (f) is bigger, illustrates that amount of image information is bigger.By the definition of comentropy it is found that when on parking stall
In the presence of having vehicle, information content is larger, and comentropy is also larger.It is tested and is shown by many experiments, the Entropy of idle state parking space
For the difference of [i] and E [i] less than 1, the difference of the Entropy [i] and E [i] of occupied state parking space are greater than 2.Therefore, comentropy
Compare and Entropy [i]+2 is set in formula for initial discrimination threshold.In view of error present in detection, setting 0.9 is permission
Error carry out discrimination threshold adjustment.
Grey-scale contrast: the ratio of image black and white, i.e. the gradual change level from black to white, calculation formula are as follows: C=∑δδ
(i,j)2pδ(i, j), wherein δ (i, j)=| i-j |, i.e. gray scale difference between adjacent pixel i and j;pδ(i, j) is δ (i's, j)
Pixel distribution probability, C is bigger, shows that the level from black to white is more, color representation is abundanter.When the parking stall free time, in parking stall
Pixel grey scale variation is smaller, and contrast is smaller;When parking stall is occupied state, level of the image from black to white is more, color table
Now abundanter, grey-scale contrast is bigger.Show the grey-scale contrast of idle state parking space less than 5 by experiment test;Work as parking stall
When occupancy, the difference of Contrast [i] and C [i] are generally higher than 15.Therefore, grey-scale contrast compares formula setting Contrast
[i]+10 is initial threshold, it is contemplated that error when detection, setting 0.9 are that the error allowed carries out the adjustment of discrimination threshold.
Harris angle point: there are two the characteristic point of principal direction, harris angle points for tool in the intersection point or neighborhood at two edges
Calculation formula it is as follows:
Corner=det (M)-λ * trace (M)
Wherein, G indicates Gaussian smoothing, Ix,cAnd Iy,cIt is image in the gradient of X and Y-direction, det (M) is the ranks of matrix M
Formula, trace (M) are the mark of matrix M, and λ is constant.In the presence of having vehicle in parking stall, the harris angle point arrived to vehicle detection is a
The harris angle point number of parking stall when number is much larger than idle state.It is found by abundant experimental results, the parking stall detected when idle
Harris angle point number less than 10, the harris angle point number of parking stall is generally higher than 20 when occupancy.For the free time to parking stall
There is a specific differentiation with occupied state, setting 15 is differentiation difference.Simultaneously in view of the error of detection, setting 0.95 is fair
Perhaps error carries out the adjustment of discrimination threshold.
The differentiation feature that the correspondence parking stall measure of parking field picture under method and system idle state of the invention goes out
As corresponding discrimination threshold, i.e. the corresponding discrimination threshold in a parking stall has to greatly improve the accuracy of detection
Effect avoid it is actually detected present in parking stall line occlusion issue.Meanwhile to the parking field picture under practical Parking situation
Before carrying out state-detection, image registration is carried out to it using the shape Matching Model that parking stall location information detection-phase creates,
The parking stall location information extracted in advance in this way and the discrimination threshold of setting just have practical significance.In addition, in order to promote state
The Stability and veracity of differentiation, this algorithm are extracted comentropy, grey-scale contrast and Harris angle point number three differentiations
Feature, using ballot thought, each feature possesses a ticket, when " occupancy " or " free time " state two tickets of acquisition or more, then
Think the state for the corresponding specific state in parking stall.
Particular embodiments described above has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that the above is only a specific embodiment of the present invention, the protection being not intended to limit the present invention
Range.It particularly points out, to those skilled in the art, all within the spirits and principles of the present invention, that is done any repairs
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of parking stall condition detection method, which comprises the following steps:
Obtain the image under the idle state of parking lot;
According to the parking space information for extracting several only in the image of parking stall contour line in an idle state and include a parking stall
Image, obtain several differentiation features from each only image comprising a parking stall, and according to differentiate feature obtain often from
The discrimination threshold of feature;
According to the image creation shape Matching Model under idle state,
Obtain the image in practical parking scene;
The image in practical parking scene is registrated using the shape Matching Model being pre-created;
One is selected in image after registration and only includes the image on a parking stall, and several differentiations for obtaining the image are special
Sign;
The corresponding discrimination threshold of the differentiation feature that will acquire is compared;
It is each to differentiate that feature includes the state feature of at least two characterization parking stall states, based on ballot thought is utilized, to every
A differentiation feature is voted, using the more state feature of poll as the particular state of current parking stall.
2. parking stall condition detection method according to claim 1, which is characterized in that described according to parking stall contour line extraction
It out only include the image of the parking space information on a parking stall, specifically includes the following steps:
Gray processing processing is carried out to the RGB image of acquisition;
Noise is removed to gray level image using gaussian filtering;
Image enhancement is carried out using grey linear transformation, is translated into bianry image;
Bianry image is optimized using the morphological operation of expansion and corrosion, image segmentation is carried out to the image after optimization;
The selection for carrying out characteristic area to the image after segmentation using feature histogram, filters out parking area;
Skeletal extraction is carried out to parking area, and is translated into sub-pix profile;
Sub-pix profile is split, and carries out feature selecting using direction and length as parameter, obtains final parking stall line wheel
It is wide;
Straight line fitting is carried out to parking stall line profile, and obtains corresponding starting point coordinate parameter information, so that generating only includes one
The image of the parking space information on a parking stall;
Several differentiation features are obtained from the image on parking stall, and are obtained according to differentiation feature often from the discrimination threshold of feature.
3. parking stall condition detection method according to claim 2, which is characterized in that the figure according under idle state
As creation shape Matching Model, specifically includes the following steps:
Shape conversion, and selected section region are carried out to the image of parking area, created according to the corresponding gray level image in partial region
Build shape Matching Model.
4. parking stall condition detection method according to claims 1 to 3, which is characterized in that further comprising the steps of:
Gray processing processing is carried out to the image obtained in practical parking scene, is translated into gray level image.
5. parking stall condition detection method according to claims 1 to 3, which is characterized in that the quantity for differentiating feature
It is three, respectively comentropy, grey-scale contrast and Harris angle point number.
6. parking stall condition detection method according to claim 5, which is characterized in that the differentiation feature that will acquire with
Its corresponding discrimination threshold is compared, each to differentiate that feature includes the state feature of at least two characterization parking stall states,
Based on using thought of voting, vote each differentiation feature, using the more state feature of poll as current parking stall
Particular state, specifically includes the following steps:
Comentropy, grey-scale contrast and the Harris that No. i-th parking stall of the image under the idle state of parking lot is extracted
Three differentiation features of angle point number are respectively Entropy [i], Contrast [i] and Points [i], corresponding differentiation threshold
Value E [i], C [i] and P [i] are respectively as follows:
E [i]=(Entropy [i]+2) * 0.9;
C [i]=(Contrast [i]+10) * 0.9;
P [i]=(Points [i]+15) * 0.95;
Comentropy, grey-scale contrast and the angle Harris that No. i-th parking stall of the image in actually parking scene is extracted
Point three differentiation features of number are respectively Entropy1 [I], Contrast1 [I] and Points1 [I], and corresponding comparison is public
Formula are as follows:
Entropy1[I]>E[i];
Contrast1[I]>C[i];
Points1[I]>P[i];
When the comparison formula for differentiating feature is set up, then it is determined as occupied state, is otherwise determined as idle state;
When two and when more than two relatively formula establishments, then determine that the parking stall for occupied state, is otherwise determined as the parking
Position is idle state.
7. a kind of parking stall condition detecting system characterized by comprising
Module is obtained, for obtaining the image in the image under the idle state of parking lot and practical parking scene;
Discrimination threshold module includes the figure of the parking space information on a parking stall for going out several only according to parking stall contour line extraction
Picture obtains several differentiation features from each only image comprising a parking stall, and is obtained according to differentiation feature often from feature
Discrimination threshold;
Model creation module, for according to the image creation shape Matching Model under idle state;
Registration module, for being registrated using the shape Matching Model being pre-created to the image in practical parking scene;
Characteristic extracting module only includes the image on a parking stall for selecting one in image after registration, and obtaining should
Several differentiation features of image;
Vote module, the corresponding discrimination threshold of differentiation feature for will acquire are compared, and each differentiation feature is wrapped
State feature containing at least two characterization parking stall states votes to each differentiation feature based on using ballot thought, will
Particular state of the more state feature of poll as current parking stall.
8. parking stall condition detection method according to claim 7, which is characterized in that the discrimination threshold module includes:
Gray processing unit, for carrying out gray processing processing to the RGB image of acquisition;
Unit is denoised, for being removed noise to gray level image using gaussian filtering;
Binarization unit is translated into bianry image for carrying out image enhancement using grey linear transformation;
Image segmentation unit, for being optimized using the morphological operation of expansion and corrosion to bianry image, after optimization
Image carries out image segmentation;
Screening unit filters out parking stall area for carrying out the selection of characteristic area to the image after segmentation using feature histogram
Domain;
Skeletal extraction unit for carrying out skeletal extraction to parking area, and is translated into sub-pix profile;
Cutting unit carries out feature selecting for being split to sub-pix profile, and using direction and length as parameter, obtains most
Whole parking stall line profile;
Parking space information unit for carrying out straight line fitting to parking stall line profile, and obtains corresponding starting point coordinate parameter information,
To generate the image of the only parking space information comprising a parking stall;
Threshold cell is obtained for obtaining several differentiation features from the image on parking stall, and according to differentiation feature often from feature
Discrimination threshold.
9. parking stall condition detection method according to claim 8, which is characterized in that the model creation module includes:
Creating unit carries out shape conversion, and selected section region for the image to parking area, corresponding according to partial region
Gray level image create shape Matching Model.
10. the parking stall condition detection method according to claim 7~9, which is characterized in that the quantity for differentiating feature
It is three, respectively comentropy, grey-scale contrast and Harris angle point number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811456794.7A CN109993991A (en) | 2018-11-30 | 2018-11-30 | Parking stall condition detection method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811456794.7A CN109993991A (en) | 2018-11-30 | 2018-11-30 | Parking stall condition detection method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109993991A true CN109993991A (en) | 2019-07-09 |
Family
ID=67129000
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811456794.7A Pending CN109993991A (en) | 2018-11-30 | 2018-11-30 | Parking stall condition detection method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109993991A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112669615A (en) * | 2020-12-09 | 2021-04-16 | 上汽大众汽车有限公司 | Parking space detection method and system based on camera |
CN113139526A (en) * | 2021-05-21 | 2021-07-20 | 一汽奔腾轿车有限公司 | Parking berth line identification method based on OpenCV |
CN113158728A (en) * | 2020-12-31 | 2021-07-23 | 杭州图歌科技有限公司 | Parking space state detection method based on gray level co-occurrence matrix |
CN113255405A (en) * | 2020-02-12 | 2021-08-13 | 广州汽车集团股份有限公司 | Parking space line identification method and system, parking space line identification device and storage medium |
CN114274948A (en) * | 2021-12-15 | 2022-04-05 | 武汉光庭信息技术股份有限公司 | Automatic parking method and device based on 360-degree panorama |
EP4138038A1 (en) * | 2021-08-20 | 2023-02-22 | Harman International Industries, Incorporated | Method and apparatus for automatic parking system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105783723A (en) * | 2016-04-26 | 2016-07-20 | 广东技术师范学院 | Machine vision-based precise die surface processing precision detection device and method |
KR101653385B1 (en) * | 2016-04-07 | 2016-09-01 | (주)서광시스템 | System for cracking down on parking violation of automatic movable using tablet pc and method therefor |
CN107610506A (en) * | 2017-09-04 | 2018-01-19 | 浙江工商大学 | The detection method and system of parking position state |
-
2018
- 2018-11-30 CN CN201811456794.7A patent/CN109993991A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101653385B1 (en) * | 2016-04-07 | 2016-09-01 | (주)서광시스템 | System for cracking down on parking violation of automatic movable using tablet pc and method therefor |
CN105783723A (en) * | 2016-04-26 | 2016-07-20 | 广东技术师范学院 | Machine vision-based precise die surface processing precision detection device and method |
CN107610506A (en) * | 2017-09-04 | 2018-01-19 | 浙江工商大学 | The detection method and system of parking position state |
Non-Patent Citations (3)
Title |
---|
杨英杰,胡义轩,张文龙: "基于Harris与SVM角点判断的车位状态检测算法", 《计算机光盘软件与应用》 * |
杨英杰,胡义轩,张文龙: "基于图像多特征识别的空闲车位检测", 《辽宁大学学报自然科学版》 * |
王楚,王亚刚: "基于拟合的亚像素边缘检测的算法比较", 《软件导刊》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113255405A (en) * | 2020-02-12 | 2021-08-13 | 广州汽车集团股份有限公司 | Parking space line identification method and system, parking space line identification device and storage medium |
CN112669615A (en) * | 2020-12-09 | 2021-04-16 | 上汽大众汽车有限公司 | Parking space detection method and system based on camera |
CN113158728A (en) * | 2020-12-31 | 2021-07-23 | 杭州图歌科技有限公司 | Parking space state detection method based on gray level co-occurrence matrix |
CN113139526A (en) * | 2021-05-21 | 2021-07-20 | 一汽奔腾轿车有限公司 | Parking berth line identification method based on OpenCV |
EP4138038A1 (en) * | 2021-08-20 | 2023-02-22 | Harman International Industries, Incorporated | Method and apparatus for automatic parking system |
CN114274948A (en) * | 2021-12-15 | 2022-04-05 | 武汉光庭信息技术股份有限公司 | Automatic parking method and device based on 360-degree panorama |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109993991A (en) | Parking stall condition detection method and system | |
CN106373426B (en) | Parking stall based on computer vision and violation road occupation for parking monitoring method | |
CN102364496B (en) | Method and system for identifying automobile license plates automatically based on image analysis | |
CN102542660B (en) | Bill anti-counterfeiting identification method based on bill watermark distribution characteristics | |
CN101807352A (en) | Method for detecting parking stalls on basis of fuzzy pattern recognition | |
CN106683076B (en) | The method of train wheel tread damage detection based on textural characteristics cluster | |
CN110414355A (en) | The right bit sky parking stall of view-based access control model and parking stall line detecting method during parking | |
CN102609701B (en) | Remote sensing detection method based on optimal scale for high-resolution SAR (synthetic aperture radar) | |
CN105632015B (en) | A kind of bank money fingerprint characteristic anti false authentication method based on cloud platform | |
CN107591005B (en) | Parking area management method, server and system combining dynamic and static detection | |
CN101763505B (en) | Vehicle license character feature extracting and classifying method based on projection symmetry | |
CN103729856B (en) | A kind of Fabric Defects Inspection detection method utilizing S-transformation signal extraction | |
CN110033431A (en) | Non-contact detection device and detection method for detecting corrosion area on surface of steel bridge | |
CN110502973A (en) | A kind of roadmarking automation extraction and recognition methods based on vehicle-mounted laser point cloud | |
CN101533466B (en) | Image processing method for positioning eyes | |
CN110321855A (en) | A kind of greasy weather detection prior-warning device | |
CN111027544B (en) | MSER license plate positioning method and system based on visual saliency detection | |
CN106815583A (en) | A kind of vehicle at night license plate locating method being combined based on MSER and SWT | |
CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
CN104036485A (en) | Method about image resampling tampering detection | |
CN109635722B (en) | Automatic identification method for high-resolution remote sensing image intersection | |
Cheng et al. | Image segmentation technology and its application in digital image processing | |
CN101931792A (en) | Multi-plate number locating method in high definition video steaming | |
Wang et al. | License plate location algorithm based on edge detection and morphology | |
CN105335688B (en) | A kind of aircraft model recognition methods of view-based access control model image |
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: 20190709 |
|
RJ01 | Rejection of invention patent application after publication |