CN108564814B - Image-based parking lot parking space detection method and device - Google Patents

Image-based parking lot parking space detection method and device Download PDF

Info

Publication number
CN108564814B
CN108564814B CN201810573699.9A CN201810573699A CN108564814B CN 108564814 B CN108564814 B CN 108564814B CN 201810573699 A CN201810573699 A CN 201810573699A CN 108564814 B CN108564814 B CN 108564814B
Authority
CN
China
Prior art keywords
parking space
parking
image
gray
value
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
Application number
CN201810573699.9A
Other languages
Chinese (zh)
Other versions
CN108564814A (en
Inventor
王宝宗
顾会建
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou Automotive Research Institute of Tsinghua University
Original Assignee
Suzhou Automotive Research Institute of Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Suzhou Automotive Research Institute of Tsinghua University filed Critical Suzhou Automotive Research Institute of Tsinghua University
Priority to CN201810573699.9A priority Critical patent/CN108564814B/en
Publication of CN108564814A publication Critical patent/CN108564814A/en
Application granted granted Critical
Publication of CN108564814B publication Critical patent/CN108564814B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • G06V10/507—Summing image-intensity values; Histogram projection analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Traffic Control Systems (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a parking lot parking space detection method based on images, which comprises the following steps: acquiring position information of monitoring parking spaces, and expressing the position information of each parking space through coordinates of a plurality of angular points; acquiring edge information of a monitoring image, and primarily screening the parking space state; processing the image and extracting the angular point of the parking place; acquiring coordinates of angular points of each parking space, calculating the distance from each angular point coordinate to a central point of the parking space, constructing position characteristics by combining position information of the parking space, and obtaining mixed characteristics of the parking spaces by combining gray level histogram characteristics of the parking spaces; and training a parking space classifier, and judging whether a vehicle exists in the current parking space by using the trained parking space classifier. The detection method can simultaneously process a plurality of parking spaces, can reduce the equipment cost and improve the parking space management efficiency of the parking lot.

Description

Image-based parking lot parking space detection method and device
Technical Field
The invention relates to a parking lot parking space detection method, in particular to a parking lot parking space detection method and device based on monitoring image analysis.
Background
With the rapid increase of the automobile demand, the automobile production quantity in China already occupies the first place in the world. The number of public parking lots is increasingly unable to meet the increasing parking requirements of people, and the problem of difficult parking is increasingly prominent. With the larger the parking lot is built, the defects of high labor intensity and low efficiency of manual management are more prominent, and the requirements of high efficiency and fast pace of the current society cannot be met. The intelligent parking lot system has come to the end and plays an important role in medium and large-sized parking lots. The parking space detection of the parking lot is the most important part of an intelligent parking lot system.
The parking lot parking space detection technology is mainly divided into four types: the traditional manpower management system needs a great deal of manpower investment, wastes time and labor, and is easy to cause traffic jam in a parking lot; when the ground induction coil is installed, the construction needs to be carried out on the road surface, so that the operation of a parking lot is influenced, when a machine is damaged, the maintenance needs to be carried out on the road surface, so that the equipment cost and the maintenance cost are increased, and the ground induction coil is not suitable for the parking space detection of an outdoor large parking lot; when the ultrasonic detection is used for positioning a vehicle, manual positioning is needed; the intelligent parking lot system based on video analysis can guide a car owner to find an empty parking space in time, can accurately find each parking space when the car is reversely found, is high in accuracy, and reduces the time for the car owner to blindly find the parking space in the parking lot. Normalized construction of a parking lot is facilitated.
A plurality of modules are arranged in the intelligent parking lot management system, and the parking space detection module occupies an important position in the whole system. Because the video image to be detected is dynamic, the accuracy of the detection result can be reduced to a certain extent under the influence of factors such as weather change, illumination intensity, shadow, pedestrian sundries and the like.
The parking lot detection method based on video analysis aims at solving the following problems:
(1) the detection accuracy is influenced by factors such as different colors and sizes of vehicles, irregular shielding of the parking line by a vehicle owner and the like;
(2) influence of weather and illumination changes on detection results;
(3) the influence of shadows, pedestrians, sundries and the like on the parking space on the detection result.
Disclosure of Invention
In order to solve the technical problems, the invention provides an image-based parking space detection method for a parking lot, which is high in recognition rate and capable of eliminating influences of weather, illumination, pedestrians, sundries, shadows and the like on detection results by combining texture and gray level characteristics to judge the state of a parking space, can simultaneously process a plurality of parking spaces, can reduce equipment cost and improves parking space management efficiency of the parking lot.
The technical scheme of the invention is as follows:
a parking lot parking space detection method based on images comprises the following steps:
s01: acquiring position information of monitoring parking spaces, and expressing the position information of each parking space through coordinates of a plurality of angular points;
s02: acquiring edge information of a monitoring image, and primarily screening the parking space state;
s03: processing the image and extracting the angular point of the parking place;
s04: acquiring coordinates of angular points of each parking space, calculating the distance from each angular point coordinate to a central point of the parking space, constructing position characteristics by combining position information of the parking space, and obtaining mixed characteristics of the parking spaces by combining gray level histogram characteristics of the parking spaces;
s05: and training a parking space classifier, and judging whether a vehicle exists in the current parking space by using the trained parking space classifier.
In a preferred embodiment, before the step S02, the image is preprocessed by gaussian filtering.
In a preferred technical solution, the step S02 specifically includes:
s21: calculating the edge directions and local gradient amplitudes of all pixel points in the image to obtain points with larger amplitudes in the gradient direction;
s22: and carrying out non-maximum suppression on the obtained amplitude, finding out a local maximum value and obtaining an edge point.
In a preferred technical solution, the corner in step S03 is a Harris corner, and the window function E (u, v) is:
Figure BDA0001686635350000021
where (u, v) is the offset of the two-dimensional gaussian window function, w (x, y) is the pixel point of the window function, and I (x, y) represents the gray level of the image.
In a preferred technical solution, the step S04 of constructing the location feature and obtaining the mixed feature of the parking space by combining the gray histogram feature of the parking space specifically includes the following steps:
s41: constructing a feature vector Dist of the dimension A, traversing distances D (x) from all corner points to a central point, and calculating Dist features according to the following formula:
Figure BDA0001686635350000031
obtaining the position characteristics of angular point distribution;
s42: the Gray value range of the pixel points is 0-255, the feature Hist of the dimension B is constructed, each pixel point in the parking space area is traversed, the corresponding Gray value Gray is obtained, and then the Hist [ Gray ] +, the Gray histogram feature is constructed;
s43: and combining the position characteristic and the gray characteristic to construct a mixed characteristic with the dimension of A + B.
In a preferred technical solution, in the step S05, a two-class classifier is constructed by an SVM, and the training of the parking space classifier includes the following steps:
s51: collecting training set T { (x)1,y1),…,(xl,yl)}∈(X×Y)l；
Wherein x isi∈X＝Rn,yi∈Y＝{-1,1}(i＝1,2,…,l)；xiIs a feature vector, yiRepresents a label;
s52: choosing a kernel function K (x)i，xj) And parameter, ai、ajRepresenting the value of the optimal solution, C is the weight between two terms in the objective function, and l represents the dimension of the feature vector;
Figure BDA0001686635350000032
Figure BDA0001686635350000033
obtaining an optimal solution:
Figure BDA0001686635350000034
s53: selection of alpha*A component of
Figure BDA0001686635350000035
And calculating a threshold value, where K (x)i-xj) Value representing kernel function:
Figure BDA0001686635350000036
s54: constructing a decision function:
Figure BDA0001686635350000037
when the value of the function is '1', the parking space is indicated to be parked with a vehicle, and when the value is '-1', the parking space is indicated to be not parked with a vehicle.
The invention also discloses a parking lot parking space detection device based on the image, which comprises the following components:
the parking place position acquisition module is used for acquiring position information of the monitored parking places and expressing the position information of each parking place through coordinates of a plurality of angular points;
the parking space state preliminary screening module is used for acquiring edge information of the monitoring image and preliminarily screening the parking space state;
the parking space angular point extraction module is used for processing the image and extracting the angular point of the parking space;
the mixed feature construction module is used for acquiring coordinates of angular points of each parking space, calculating the distance from each angular point coordinate to a central point of the parking space, constructing a position feature by combining position information of the parking space, and obtaining mixed features of the parking spaces by combining gray histogram features of the parking spaces;
and a parking space classifier training and judging module is used for training a parking space classifier and judging whether a vehicle exists in the current parking space or not by utilizing the trained parking space classifier.
In an optimized technical scheme, the step position state preliminary screening module specifically processes the steps and comprises the following steps:
s21: calculating the edge directions and local gradient amplitudes of all pixel points in the image to obtain points with larger amplitudes in the gradient direction;
s22: and carrying out non-maximum suppression on the obtained amplitude, finding out a local maximum value and obtaining an edge point.
In an optimal technical scheme, the position feature is constructed in the mixed feature construction module, and the mixed feature of the parking space is obtained by combining the gray histogram feature of the parking space, and the method specifically comprises the following steps:
s41: constructing a feature vector Dist of the dimension A, traversing distances D (x) from all corner points to a central point, and calculating Dist features according to the following formula:
Figure BDA0001686635350000041
obtaining the position characteristics of angular point distribution;
s42: the Gray value range of the pixel points is 0-255, the feature Hist of the dimension B is constructed, each pixel point in the parking space area is traversed, the corresponding Gray value Gray is obtained, and then the Hist [ Gray ] +, the Gray histogram feature is constructed;
s43: and combining the position characteristic and the gray characteristic to construct a mixed characteristic with the dimension of A + B.
In an optimal technical scheme, a two-classifier is constructed in the parking space classifier training judgment module through an SVM, and the parking space classifier training comprises the following steps:
s51: collecting training set T { (x)1,y1),…,(xl,yl)}∈(X×Y)l；
Wherein x isi∈X＝Rn,yi∈Y＝{-1,1}(i＝1,2,…,l)；xiIs a feature vector, yiRepresents a label;
s52: choosing a kernel function K (x)i，xj) And parameter, ai、ajRepresenting the value of the optimal solution, C is the weight between two terms in the objective function, and l represents the dimension of the feature vector;
Figure BDA0001686635350000051
Figure BDA0001686635350000052
obtaining an optimal solution:
Figure BDA0001686635350000053
s53: selection of alpha*A component of
Figure BDA0001686635350000054
And calculating a threshold value, where K (x)i-xj) Value representing kernel function:
Figure BDA0001686635350000055
s54: constructing a decision function:
Figure BDA0001686635350000056
when the value of the function is '1', the parking space is indicated to be parked with a vehicle, and when the value is '-1', the parking space is indicated to be not parked with a vehicle.
Compared with the prior art, the invention has the advantages that:
1. on the basis of Gaussian filtering, Harris angular point characteristics are extracted by combining with parking space coordinates, Harris characteristics of parking spaces are acquired, and states of the parking spaces are determined through SVM classification prediction. Compared with the prior art, the method has high detection rate, can judge the parking places in various complicated parking lot environments, can well eliminate the influence of weather, illumination, pedestrians, sundries, shadows and the like on the parking places on the parking place detection, and has the parking place detection accuracy of 98.15 percent through a large number of tests, thereby meeting the requirement of large outdoor parking lots on the parking place detection.
2. The detection method can simultaneously process a plurality of parking spaces, can reduce the equipment cost and improve the parking space management efficiency of the parking lot.
Drawings
The invention is further described with reference to the following figures and examples:
fig. 1 is a flowchart of a parking space detection method based on images according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
Example (b):
the preferred embodiments of the present invention will be further described with reference to the accompanying drawings.
As shown in fig. 1, the method for detecting parking spaces in a parking lot based on images of the present invention includes the following steps:
the method comprises the following steps: and acquiring the position information of the parking places to be monitored from the monitoring image, and representing and storing the position information of each parking place by coordinates of four points (upper left, lower left, upper right and lower right) to prepare for monitoring the subsequent parking places.
Step two: obtaining an image from a monitoring camera video, and preprocessing the image by Gaussian filtering;
Figure BDA0001686635350000061
x2and y2Respectively representing the distances between other pixels in the neighborhood and the central pixel in the neighborhood, and sigma represents a targetAnd (4) tolerance. h (x, y) represents a two-dimensional Gaussian filter, is a convolution template calculated by using normal distribution, and can perform fuzzy processing on the image by performing convolution operation on the Gaussian filter and the image.
Step three: the parking stall state is tentatively screened, utilizes Canny algorithm to acquire the marginal information of parking stall, and the parking stall has the state that the car was parked, and marginal information is abundant, and marginal information is sparse when no car state, utilizes this characteristic to carry out preliminary judgement.
(1) Denoising through the Gaussian filtering of the previous step;
(2) calculating the edge directions and local gradient amplitudes of all pixel points in the image, and using Sobel
Figure BDA0001686635350000062
The operator obtains the point with the maximum amplitude value in the gradient direction, namely the edge point;
the three matrixes of the above formula are respectively an x-direction convolution template, a y-direction convolution template of the operator and a neighborhood point mark matrix of the point to be processed, and accordingly, the gradient amplitude of each point can be expressed by a mathematical formula as follows:
Figure BDA0001686635350000071
sx＝(a2+2a3+a4)-(a0+2a7+a6)
sy＝(a0+2a1+a2)-(a6+2a5+a4)
the gradient direction can then be calculated with the following formula:
Figure BDA0001686635350000072
if the angle theta is equal to zero, this indicates that the image has a longitudinal edge there, and is darker to the left and to the right.
(3) The amplitude obtained through the steps is subjected to non-maximum suppression, the situation that a plurality of larger amplitudes are close to each other may exist in the obtained amplitude image, but only one true edge point exists, and for the situation, the non-maximum suppression is carried out, the local maximum value is found out, and therefore most of non-edge points can be removed.
Step four: extracting Harris angular points of the parking spaces from the processed images;
Figure BDA0001686635350000073
wherein, E (u, v) represents a window function, (u, v) is an offset of a two-dimensional Gaussian window function, w (x, y) is a pixel point of the window function, and I (x, y) represents the gray level of an image.
Step five: acquiring coordinates of the angular points of each parking space, calculating the distance from each angular point coordinate to a parking space central point, and constructing position characteristics by combining position information of the parking spaces: calculating by using the central point of the parking space as the origin
Figure BDA0001686635350000074
Euclidean distance D (x) of each corner point coordinate to the center point:
wherein h (x), h (y) represent the x, y coordinates of the corner point, c (x), c (y) represent the x, y coordinates of the central point;
constructing a feature vector Dist with the dimension of 27, traversing distances D (x) from all corner points to a central point, and calculating Dist features according to the following formula:
Figure BDA0001686635350000075
and obtaining the position characteristics of the corner point distribution. Constructing a gray level histogram feature: the Gray value range of the pixel points is 0-255, a feature Hist with the dimensionality of 256 is constructed, each pixel point in the parking space area is traversed, a corresponding Gray value Gray is obtained, then the Hist [ Gray ] + +, a mixed feature with the dimensionality of 256+27 is constructed by combining the position feature and the Gray feature, and an SVM classifier is combined to train the parking space classifier and carry out classification prediction;
and selecting different kernel functions by using the SVM, and carrying out error correction on the related data. The kernel function K mainly includes:
(1) linear kernel function: k (x, x)i)＝xTxi；
(2) Polynomial kernel function: k (x, x)i)＝(γxTxi+r)P,γ＞0；
(3) Radial basis kernel function: k (x, x)i)＝exp(-γ||x-xi||2),γ＞0；
(4) Two layers of perceptual kernels: k (x, x)i)＝tanh(γxTxi+r)。
The SVM algorithm is used for solving the problem of two-classification, and if the actual problem belongs to multi-classification, a multi-classifier can be constructed through the SVM. In addition, multi-classification problems can also be handled by recombining multiple binary classifiers. Whether a vehicle exists on the parking space is judged, and actually, the two-classification problem is solved, so that a two-classification SVM model is used for judging whether the vehicle stops on the parking space, and the form is as follows:
(1) collecting a training set:
T＝{(x1,y1),…,(xl,yl)}∈(X×Y)l
wherein x isi∈X＝Rn,yi∈Y＝{-1,1}(i＝1,2,…,l)；xiIs a feature vector, T is a training data set, yiRepresenting a label.
(2) Selecting proper kernel function K (x, x)r) And parameter, ai、ajA value representing an optimal solution, C is
Figure BDA0001686635350000081
Figure BDA0001686635350000082
The weight between two terms in the objective function, l represents the dimension of the feature vector:
obtaining an optimal solution:
Figure BDA0001686635350000083
(3) selection of alpha*A component of
Figure BDA0001686635350000084
Wherein K (x)i-xj) Represents the value of the kernel function, and calculates a threshold value:
Figure BDA0001686635350000091
(4) constructing a decision function:
Figure BDA0001686635350000092
when the value of the function is '1', the parking space is indicated to be parked with a vehicle, and when the value is '-1', the parking space is indicated to be not parked with a vehicle.
And after the training is finished, storing the XML template file, directly loading the XML file during subsequent use, judging whether the current parking space has a car or not by using the prediction function, outputting a result and finishing the judgment.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.

Claims (8)

1. The image-based parking space detection method for the parking lot is characterized by comprising the following steps of:
s01: acquiring position information of monitoring parking spaces, and expressing the position information of each parking space through coordinates of a plurality of angular points;
s02: acquiring edge information of a monitoring image, and primarily screening the parking space state;
s03: processing the image and extracting the angular point of the parking place;
s04: acquiring coordinates of angular points of each parking space, calculating the distance from each angular point coordinate to a central point of the parking space, constructing position characteristics by combining position information of the parking space, and obtaining mixed characteristics of the parking spaces by combining gray level histogram characteristics of the parking spaces; the method for constructing the position features and obtaining the mixed features of the parking spaces by combining the gray histogram features of the parking spaces specifically comprises the following steps:
s41: constructing a feature vector Dist of the dimension A, traversing distances D (x) from all corner points to a central point, and calculating Dist features according to the following formula:
Figure FDA0002675376240000011
obtaining the position characteristics of angular point distribution;
s42: the Gray value range of the pixel points is 0-255, the feature Hist of the dimension B is constructed, each pixel point in the parking space area is traversed, the corresponding Gray value Gray is obtained, and then the Hist [ Gray ] +, the Gray histogram feature is constructed;
s43: combining the position characteristic and the gray characteristic to construct a mixed characteristic with dimension A + B;
s05: and training a parking space classifier, and judging whether a vehicle exists in the current parking space by using the trained parking space classifier.
2. The image-based parking space detection method of claim 1, wherein before the step S02, the image is preprocessed by gaussian filtering.
3. The image-based parking space detection method of the parking lot according to claim 1 or 2, wherein the step S02 specifically comprises:
s21: calculating the edge directions and local gradient amplitudes of all pixel points in the image to obtain points with larger amplitudes in the gradient direction;
s22: and carrying out non-maximum suppression on the obtained amplitude, finding out a local maximum value and obtaining an edge point.
4. The image-based parking space detection method of claim 1, wherein the angular point in step S03 is a Harris angular point, and the window function E (u, v) is:
Figure FDA0002675376240000021
where (u, v) is the offset of the two-dimensional gaussian window function, w (x, y) is the pixel point of the window function, and I (x, y) represents the gray level of the image.
5. The image-based parking space detection method of claim 1, wherein in step S05, a two-classifier is constructed by SVM, and training the space classifier comprises the following steps:
s51: collecting training set T { (x)1,y1),…,(xl,yl)}∈(X×Y)l；
Wherein x isi∈X＝Rn,yi∈Y＝{-1,1}(i＝1,2,…,l)；xiIs a feature vector, yiRepresents a label;
s52: choosing a kernel function K (x)i，xj) And parameter, ai、ajRepresenting the value of the optimal solution, C is the weight between two terms in the objective function, and l represents the dimension of the feature vector;
Figure FDA0002675376240000022
Figure FDA0002675376240000023
obtaining an optimal solution:
Figure FDA0002675376240000024
s53: selection of alpha*A component of
Figure FDA0002675376240000025
And calculating a threshold value, where K (x)i-xj) Value representing kernel function:
Figure FDA0002675376240000026
s54: constructing a decision function:
Figure FDA0002675376240000027
when the value of the function is '1', the parking space is indicated to be parked with a vehicle, and when the value is '-1', the parking space is indicated to be not parked with a vehicle.
6. The utility model provides a parking area parking stall detection device based on image which characterized in that includes:
the parking place position acquisition module is used for acquiring position information of the monitored parking places and expressing the position information of each parking place through coordinates of a plurality of angular points;
the parking space state preliminary screening module is used for acquiring edge information of the monitoring image and preliminarily screening the parking space state;
the parking space angular point extraction module is used for processing the image and extracting the angular point of the parking space;
the mixed feature construction module is used for acquiring coordinates of angular points of each parking space, calculating the distance from each angular point coordinate to a central point of the parking space, constructing a position feature by combining position information of the parking space, and obtaining mixed features of the parking spaces by combining gray histogram features of the parking spaces; the method for constructing the position features and obtaining the mixed features of the parking spaces by combining the gray histogram features of the parking spaces specifically comprises the following steps:
s41: constructing a feature vector Dist of the dimension A, traversing distances D (x) from all corner points to a central point, and calculating Dist features according to the following formula:
Figure FDA0002675376240000031
obtaining the position characteristics of angular point distribution;
s42: the Gray value range of the pixel points is 0-255, the feature Hist of the dimension B is constructed, each pixel point in the parking space area is traversed, the corresponding Gray value Gray is obtained, and then the Hist [ Gray ] +, the Gray histogram feature is constructed;
s43: combining the position characteristic and the gray characteristic to construct a mixed characteristic with dimension A + B;
and a parking space classifier training and judging module is used for training a parking space classifier and judging whether a vehicle exists in the current parking space or not by utilizing the trained parking space classifier.
7. The image-based parking space detection device for parking lot of claim 6, wherein the specific processing of the space status preliminary screening module comprises:
s21: calculating the edge directions and local gradient amplitudes of all pixel points in the image to obtain points with larger amplitudes in the gradient direction;
s22: and carrying out non-maximum suppression on the obtained amplitude, finding out a local maximum value and obtaining an edge point.
8. The image-based parking space detection device of claim 6, wherein the space classifier training judgment module constructs a two-classifier by SVM, and the training of the space classifier comprises the following steps:
s51: collecting training set T { (x)1,y1),…,(xl,yl)}∈(X×Y)l；
Wherein x isi∈X＝Rn,yi∈Y＝{-1,1}(i＝1,2,…,l)；xiIs a feature vector, yiRepresents a label;
s52: choosing a kernel function K (x)i，xj) And parameter, ai、ajRepresenting the value of the optimal solution, C is the weight between two terms in the objective function, and l represents the dimension of the feature vector;
Figure FDA0002675376240000041
Figure FDA0002675376240000042
obtaining an optimal solution:
Figure FDA0002675376240000043
s53: selection of alpha*A component of
Figure FDA0002675376240000044
And calculating a threshold value, where K (x)i-xj) Value representing kernel function:
Figure FDA0002675376240000045
s54: constructing a decision function:
Figure FDA0002675376240000046
when the value of the function is '1', the parking space is indicated to be parked with a vehicle, and when the value is '-1', the parking space is indicated to be not parked with a vehicle.
CN201810573699.9A 2018-06-06 2018-06-06 Image-based parking lot parking space detection method and device Active CN108564814B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810573699.9A CN108564814B (en) 2018-06-06 2018-06-06 Image-based parking lot parking space detection method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810573699.9A CN108564814B (en) 2018-06-06 2018-06-06 Image-based parking lot parking space detection method and device

Publications (2)

Publication Number Publication Date
CN108564814A CN108564814A (en) 2018-09-21
CN108564814B true CN108564814B (en) 2020-11-17

Family

ID=63553024

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810573699.9A Active CN108564814B (en) 2018-06-06 2018-06-06 Image-based parking lot parking space detection method and device

Country Status (1)

Country Link
CN (1) CN108564814B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110132278B (en) * 2019-05-14 2021-07-02 驭势科技（北京）有限公司 Method and device for instant positioning and mapping
CN110348297B (en) * 2019-05-31 2023-12-26 纵目科技（上海）股份有限公司 Detection method, system, terminal and storage medium for identifying stereo garage
CN112115741B (en) * 2019-06-20 2024-02-20 上海汽车集团股份有限公司 Parking garage position detection method and device
CN110490079B (en) * 2019-07-19 2022-06-03 万翼科技有限公司 Patrol data processing method and device, computer equipment and storage medium
CN110969655B (en) * 2019-10-24 2023-08-18 百度在线网络技术（北京）有限公司 Method, device, equipment, storage medium and vehicle for detecting parking space
CN110796063B (en) * 2019-10-24 2022-09-09 百度在线网络技术（北京）有限公司 Method, device, equipment, storage medium and vehicle for detecting parking space
CN111428616B (en) * 2020-03-20 2023-05-23 东软睿驰汽车技术（沈阳）有限公司 Parking space detection method, device, equipment and storage medium
CN111539278A (en) * 2020-04-14 2020-08-14 浙江吉利汽车研究院有限公司 Detection method and system for target vehicle
CN112598922B (en) * 2020-12-07 2023-03-21 安徽江淮汽车集团股份有限公司 Parking space detection method, device, equipment and storage medium
CN112634650B (en) * 2020-12-18 2021-10-26 中标慧安信息技术股份有限公司 Parking lot management method and system based on audio and video monitoring
CN113658272B (en) * 2021-08-19 2023-11-17 亿咖通(湖北)技术有限公司 Vehicle-mounted camera calibration method, device, equipment and storage medium
CN114782928A (en) * 2022-06-22 2022-07-22 苏州魔视智能科技有限公司 Parking space detection method and device, electronic equipment and storage medium
CN115791791B (en) * 2022-11-14 2024-06-18 中国科学院沈阳自动化研究所 Visual detection method for liquid crystal display panel boxing scrap

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101196996A (en) * 2007-12-29 2008-06-11 北京中星微电子有限公司 Image detection method and device
CN101807352A (en) * 2010-03-12 2010-08-18 北京工业大学 Method for detecting parking stalls on basis of fuzzy pattern recognition
DE102010019464A1 (en) * 2010-05-05 2011-11-10 Gisela Toussaint Data center for receiving, processing, and transmission of special signals, is provided for detecting all public park spaces occupied, by which precise geographical coordinates of their four corner points are detected
WO2013076365A1 (en) * 2011-11-22 2013-05-30 Nokia Corporation Method for image processing and an apparatus
CN104916163A (en) * 2015-06-29 2015-09-16 惠州华阳通用电子有限公司 Parking space detection method
CN106599921A (en) * 2016-12-14 2017-04-26 深圳市捷顺科技实业股份有限公司 Parking stall guiding method and device
CN107610506A (en) * 2017-09-04 2018-01-19 浙江工商大学 The detection method and system of parking position state
CN107886080A (en) * 2017-11-23 2018-04-06 同济大学 One kind is parked position detecting method

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101755980B1 (en) * 2016-01-29 2017-07-10 세종대학교산학협력단 Copy-Move Forgery Detection method and apparatus based on scale space representation
CN105913685A (en) * 2016-06-25 2016-08-31 上海大学 Video surveillance-based carport recognition and intelligent guide method
CN106373426B (en) * 2016-09-29 2019-02-12 成都通甲优博科技有限责任公司 Parking stall based on computer vision and violation road occupation for parking monitoring method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101196996A (en) * 2007-12-29 2008-06-11 北京中星微电子有限公司 Image detection method and device
CN101807352A (en) * 2010-03-12 2010-08-18 北京工业大学 Method for detecting parking stalls on basis of fuzzy pattern recognition
DE102010019464A1 (en) * 2010-05-05 2011-11-10 Gisela Toussaint Data center for receiving, processing, and transmission of special signals, is provided for detecting all public park spaces occupied, by which precise geographical coordinates of their four corner points are detected
WO2013076365A1 (en) * 2011-11-22 2013-05-30 Nokia Corporation Method for image processing and an apparatus
CN104916163A (en) * 2015-06-29 2015-09-16 惠州华阳通用电子有限公司 Parking space detection method
CN106599921A (en) * 2016-12-14 2017-04-26 深圳市捷顺科技实业股份有限公司 Parking stall guiding method and device
CN107610506A (en) * 2017-09-04 2018-01-19 浙江工商大学 The detection method and system of parking position state
CN107886080A (en) * 2017-11-23 2018-04-06 同济大学 One kind is parked position detecting method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Vehicle Detection in Aerial Imagery using Eigen Features;Ajay，A； Sowmya, V；Soman, KP;《2017 INTERNATIONAL CONFERENCE ON COMMUNICATION AND SIGNAL PROCESSING (ICCSP)》;20170608;全文 *
基于Harris与SVM角点判断的车位状态检测方法;杨英杰，胡义轩，张文龙;《计算机光盘软件与应用》;20140731;文件第123-124页 *
基于LOG和Canny算子的边缘检测算法;贺强，晏立;《计算机工程》;20110331;文件第211页 *
数据挖掘在入侵检测中的应用;冯翔;《中国优秀硕士学位论文全文数据库（信息科技辑）》;20110415;文件第22-23页 *

Also Published As

Publication number Publication date
CN108564814A (en) 2018-09-21

Similar Documents

Publication Publication Date Title
CN108564814B (en) Image-based parking lot parking space detection method and device
Abedin et al. License plate recognition system based on contour properties and deep learning model
CN109993056B (en) Method, server and storage medium for identifying vehicle illegal behaviors
CN107301405A (en) Method for traffic sign detection under natural scene
CN107545239B (en) Fake plate detection method based on license plate recognition and vehicle characteristic matching
Polishetty et al. A next-generation secure cloud-based deep learning license plate recognition for smart cities
Yan et al. A method of lane edge detection based on Canny algorithm
CN112819094B (en) Target detection and identification method based on structural similarity measurement
CN105809184B (en) Method for real-time vehicle identification and tracking and parking space occupation judgment suitable for gas station
CN109190444B (en) Method for realizing video-based toll lane vehicle feature recognition system
Paunwala et al. A novel multiple license plate extraction technique for complex background in Indian traffic conditions
CN108256521B (en) Effective area positioning method for vehicle body color identification
CN105740886B (en) A kind of automobile logo identification method based on machine learning
CN110414308B (en) Target identification method for dynamic foreign matters on power transmission line
Khalifa et al. Malaysian Vehicle License Plate Recognition.
CN107092876A (en) The low-light (level) model recognizing method combined based on Retinex with S SIFT features
CN109509188B (en) Power transmission line typical defect identification method based on HOG characteristics
CN104933398A (en) vehicle identification system and method
CN109583435A (en) A kind of adaptive ballot method for detecting lane lines
CN111915583A (en) Vehicle and pedestrian detection method based on vehicle-mounted thermal infrared imager in complex scene
CN111091111A (en) Vehicle bottom dangerous target identification method
CN111652033A (en) Lane line detection method based on OpenCV
CN110516666B (en) License plate positioning method based on combination of MSER and ISODATA
CN113053164A (en) Parking space identification method using look-around image
CN111476233A (en) License plate number positioning method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant