CN112419345A - Patrol car high-precision tracking method based on echo state network - Google Patents
Patrol car high-precision tracking method based on echo state network Download PDFInfo
- Publication number
- CN112419345A CN112419345A CN202011405929.4A CN202011405929A CN112419345A CN 112419345 A CN112419345 A CN 112419345A CN 202011405929 A CN202011405929 A CN 202011405929A CN 112419345 A CN112419345 A CN 112419345A
- Authority
- CN
- China
- Prior art keywords
- patrol car
- patrol
- image
- value
- state network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 235000002566 Capsicum Nutrition 0.000 claims abstract description 24
- 239000006002 Pepper Substances 0.000 claims abstract description 24
- 235000016761 Piper aduncum Nutrition 0.000 claims abstract description 24
- 235000017804 Piper guineense Nutrition 0.000 claims abstract description 24
- 235000008184 Piper nigrum Nutrition 0.000 claims abstract description 24
- 150000003839 salts Chemical class 0.000 claims abstract description 24
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 241000722363 Piper Species 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 18
- 238000003708 edge detection Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 12
- 238000009499 grossing Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 230000001629 suppression Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims 1
- 244000203593 Piper nigrum Species 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30256—Lane; Road marking
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
A patrol car high-precision tracking method based on an echo state network. The method comprises the following steps: step 1, collecting patrol car driving image data; step 2, adding salt and pepper noise to the acquired image; step 3, detecting a lane line in the image; step 4, establishing tracking models of the patrol car and the lane lines; step 5, collecting abnormal conditions on the patrol route; and 6, starting the driving interruption of the patrol car. The method simulates the severe environment during driving, realizes the patrol car tracking function under the interference of a noise environment, enhances the stability, reliability and robustness of patrol car tracking, detects the edge line of the lane line through a Canny algorithm, can smoothly reduce the noise of the collected road surface image, effectively detects the edge line of the lane line, provides important lane line data for a tracking model, solves a nonlinear tracking model of the patrol car through an echo state network, and effectively solves the problem of nonlinear lane lines.
Description
Technical Field
The invention relates to the field of patrol car tracking, in particular to a patrol car high-precision tracking method based on an echo state network.
Background
In recent years, with the development of automobile automatic driving technology, many students worldwide spend a lot of manpower, material resources, and financial resources to research the automobile automatic driving technology. With the development of machine learning and deep learning in the fields of image processing, voice recognition, natural language processing and the like, the automatic driving technology can be perfectly combined with artificial intelligence.
Machine learning and depth can provide a good algorithm basis for environment perception and cognitive decision in an automatic driving system, and meanwhile, the excellent performance of an intelligent technology in image recognition, laser radar point cloud processing, decision planning and intelligent control greatly promotes the research and development speed of the automatic driving technology. In google, tesla, galloping, bmw and other major technical enterprises and automobile host factories, automatic driving technology research, test and other works are carried out successively. With the automated driving industry investment, technology development and experimental testing progressively regressing rationality, it is clearly recognized that: the realization difficulty of an automatic driving technology, particularly an L5 level mass production automatic driving automobile with an open road trip scene is far beyond expectation; however, under the condition of orienting to a specific scene and limiting a speed range, a safe, reliable and efficient solution can be found between the networking degree and the intelligent degree by combining with technologies such as 5G, V2X, command scheduling and the like.
The security patrol is one of the best scenes for the intelligent network patrol car to land, has the characteristics of typical high frequency, low speed, rigidity requirement, large flow and the like, and simultaneously has relatively low technical implementation difficulty.
Disclosure of Invention
In order to solve the problems, the invention provides a patrol car high-precision tracking method based on an echo state network on the basis of acquiring a road surface image by a real-time image acquisition system and simulating a noise environment. In order to reduce the influence of environmental noise on data collected by the sensor as much as possible, the method reduces the noise influence by a Canny edge detection algorithm, detects the road route, and enhances the robustness of the model obtained by training. The invention provides a patrol car high-precision tracking method based on an echo state network, which comprises the following specific steps of:
step 1, acquiring patrol car driving image data: acquiring a road surface image of the patrol car running through a real-time image acquisition system on the patrol car;
step 2, adding salt and pepper noise to the collected image: adding salt and pepper noise into the road surface image obtained in the step 1 to simulate the interference of the acquired signal in a noise environment, and adding 20-30% of salt and pepper noise;
step 3, detecting lane lines in the image: detecting lane lines in the image by using a Canny edge detection algorithm, and segmenting the lane lines;
step 4, establishing tracking models of the patrol car and the lane lines: training an echo state network by taking the lane line value detected in the step 3 as an input vector, and solving a nonlinear tracking model of the lane line by using the echo state network;
step 5, collecting abnormal conditions on the patrol route: acquiring emergency situations such as help seeking of personnel on a patrol route, illegal crime and the like through a real-time image acquisition system;
and 6, starting driving interruption of the patrol car, starting an interruption processing mechanism by the system when the patrol is abnormal within a safe distance, and recording the abnormal condition of the vehicle-mounted terminal by the log module.
Further, the process of adding salt and pepper noise to the acquired image in step 2 can be expressed as:
the noise model of the added salt and pepper is:
wherein, ImaxAnd IminIs the maximum value and the minimum value of the image pixel points, p is the probability of noise occurrence of the image, the value range of p is 20 percent to 30 percent, ixyThe actual value of the collected image pixel point (x, y) is obtained, and the f (x, y) is the image pixel point value after the salt and pepper noise is added.
Further, the process of detecting the lane line in the image in step 3 can be represented as:
step 3.1, smoothing and filtering the road surface image by using a Gaussian smoothing function to inhibit salt and pepper noise in the image:
the road surface image can be represented as f (x, y), and the formula for smoothing the image is:
g(x,y)＝h(x,y,σ)*f(x,y) (2)
where σ is the gaussian standard deviation, which represents the convolution, g (x, y) is the smooth filtered road image, and h (x, y, σ) is:
step 3.2, calculating the finite difference of the first-order partial derivatives through a sobel operator to calculate the amplitude and the direction of the gradient of the road image:
dx＝g(x,y)*Sobelx(x,y) (4)
dy＝g(x,y)*Sobely(x,y) (5)
in the formula (d)xAnd dyThe gradients of the image in the x-direction and y-direction, Sobelx(x, y) is the Sobel operator in the x-direction, Sobely(x, y) is a sobel operator in the y direction, and the gradient M (x, y) of the road surface image can be expressed as:
direction of gradient thetaMComprises the following steps:
step 3.3, carrying out non-maximum suppression on the amplitude along the gradient direction calculated in the step 3.2;
step 3.4, detecting and connecting edges by using a double-threshold algorithm: selecting two thresholds to set a threshold range, setting the gray value of the pixel less than the gradient value of the threshold range to be 0, assigning 1 to the point mark more than the threshold range, connecting the contour lines of the lane lines, and segmenting to obtain the edge detection line f of the lane line1(x,y)。
Further, the process of establishing the patrol car and lane line tracking model in step 4 can be represented as follows:
step 4.1, establishing a patrol car and lane line tracking model, wherein the patrol car is subjected to left deviation and right deviation in the driving process, when the patrol car is positioned at the right sides of the lane lines at the two sides, the patrol car needs to be deviated towards the left direction, when the patrol car is positioned at the left sides of the lane lines at the two sides, the patrol car needs to be deviated towards the right direction, and the edge detection line of the training sample road surface picture in the step 3 is marked with a left deviation value E (i) and a right deviation value E (i);
step 4.2, initializing the network, forming a sample D (i) by an edge detection line of the road surface picture and forming a training sample u (i) { D (i), E (i) } by a bias value E (i) of the left and right sides of the patrol car, and inputting a training sample characteristic D (i) into a connection weight matrix W (i)inEntering a reserve pool, E (i) connecting the weights W through feedbackbackEntering a reserve pool, and acquiring the system state and the output state according to the following sequence:
x(i+1)＝f(WinD(i+1)+Wx(i)+WbackE(i)) (8)
E(i+1)＝fout(WoutD(i+1),x(i+1),E(i)) (9)
wherein x (i) is a system parameter with an initial value of 0, f () is an excitation function of a reserve pool node, fout() Is the excitation function of the output unit of the reserve pool, W represents the connection weight matrix of the internal neurons in the reserve pool, WoutRepresenting a matrix of output values;
step 4.3 calculate output value matrix Wout：
Wherein K is the number of neurons in the input layer, N is the number of neurons in the reserve pool, L is the number of neurons in the output layer, and lambda belongs to the element+Representing the regularization factor, | | | | represents the euclidean distance.
And 4.3, taking the echo state network obtained by training as a patrol car tracking route model, accurately calculating the left and right deviation values of the patrol car through the real-time image acquisition system and the trained echo state network, and calibrating the driving route of the patrol car.
Further, the process of collecting the abnormal situation on the patrol route in step 5 can be represented as follows:
the method comprises the steps that firstly, images around a patrol route are collected through a real-time image collection system, the images are identified and classified through an event classification CNN model, if the patrol process system meets the conditions of help seeking and illegal crime of personnel, the patrol process system of the patrol car orders the patrol car to stop, meanwhile, a video is started to be left, event pictures are uploaded to a cloud, and workers are reminded to perform further intervention.
The patrol car high-precision tracking method based on the echo state network has the beneficial effects that: the invention has the technical effects that:
1. the patrol car tracking system simulates a severe environment during driving, realizes the patrol car tracking function under the interference of a noise environment, and enhances the stability, reliability and robustness of the patrol car tracking;
2. the Canny algorithm detects the edge lines of the lane lines, can smoothly reduce noise of the collected road surface images, effectively detects the edge lines of the lane lines and provides important lane line data for a tracking model;
3. the nonlinear tracking model of the patrol car is solved through the echo state network, so that the problem of a nonlinear track route is effectively solved;
4. the invention provides an important technical means for high-precision tracking of the patrol car.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The invention is described in further detail below with reference to the following detailed description and accompanying drawings:
the invention provides a patrol car high-precision tracking method based on an echo state network, which aims to improve the tracking precision of a patrol car, simultaneously add salt and pepper noise in an acquired image to simulate a severe environment during traveling and solve a patrol car tracking model by combining a Canny edge detection algorithm and the echo state network in order to improve the stability and the accuracy of solving the patrol car tracking by the model, and a flow chart of the invention is shown in figure 1. The steps of the present invention will be described in detail with reference to the flow chart.
Step 1, acquiring patrol car driving image data: acquiring a road surface image of the patrol car running through a real-time image acquisition system on the patrol car;
step 2, adding salt and pepper noise to the collected image: adding salt and pepper noise into the road surface image obtained in the step 1 to simulate the interference of the acquired signal in a noise environment, and adding 20-30% of salt and pepper noise;
the process of adding salt and pepper noise to the acquired image in the step 2 can be represented as follows:
the noise model of the added salt and pepper is:
wherein, ImaxAnd IminIs the maximum value and the minimum value of the image pixel points, p is the probability of noise occurrence of the image, the value range of p is 20 percent to 30 percent, ixyThe actual value of the collected image pixel point (x, y) is obtained, and the f (x, y) is the image pixel point value after the salt and pepper noise is added.
Step 3, detecting lane lines in the image: detecting lane lines in the image by using a Canny edge detection algorithm, and segmenting the lane lines;
the process of detecting the lane line in the image in step 3 can be represented as:
step 3.1, smoothing and filtering the road surface image by using a Gaussian smoothing function to inhibit salt and pepper noise in the image:
the road surface image can be represented as f (x, y), and the formula for smoothing the image is:
g(x,y)＝h(x,y,σ)*f(x,y) (2)
where σ is the gaussian standard deviation, which represents the convolution, g (x, y) is the smooth filtered road image, and h (x, y, σ) is:
step 3.2, calculating the finite difference of the first-order partial derivatives through a sobel operator to calculate the amplitude and the direction of the gradient of the road image:
dx＝g(x,y)*Sobelx(x,y) (4)
dy＝g(x,y)*Sobely(x,y) (5)
in the formula (d)xAnd dyThe gradients of the image in the x-direction and y-direction, Sobelx(x, y) is the Sobel operator in the x-direction, Sobely(x, y) is a sobel operator in the y direction, and the gradient M (x, y) of the road surface image can be expressed as:
direction of gradient thetaMComprises the following steps:
step 3.3, carrying out non-maximum suppression on the amplitude along the gradient direction calculated in the step 3.2;
step 3.4, detecting and connecting edges by using a double-threshold algorithm: selecting two thresholds to set a threshold range, setting the gray value of the pixel less than the gradient value of the threshold range to be 0, assigning 1 to the point mark more than the threshold range, connecting the contour lines of the lane lines, and segmenting to obtain the edge detection line f of the lane line1(x,y)。
Step 4, establishing tracking models of the patrol car and the lane lines: training an echo state network by taking the lane line value detected in the step 3 as an input vector, and solving a nonlinear tracking model of the lane line by using the echo state network;
the process of establishing the patrol car and lane line tracking model in the step 4 can be represented as follows:
step 4.1, establishing a patrol car and lane line tracking model, wherein the patrol car is subjected to left deviation and right deviation in the driving process, when the patrol car is positioned at the right sides of the lane lines at the two sides, the patrol car needs to be deviated towards the left direction, when the patrol car is positioned at the left sides of the lane lines at the two sides, the patrol car needs to be deviated towards the right direction, and the edge detection line of the training sample road surface picture in the step 3 is marked with a left deviation value E (i) and a right deviation value E (i);
step 4.2, initializing the network, forming a sample D (i) by an edge detection line of the road surface picture and forming a training sample u (i) { D (i), E (i) } by a bias value E (i) of the left and right sides of the patrol car, and inputting a training sample characteristic D (i) into a connection weight matrix W (i)inEntering a reserve pool, E (i) connecting the weights W through feedbackbackEntering a reserve pool, and acquiring the system state and the output state according to the following sequence:
x(i+1)＝f(WinD(i+1)+Wx(i)+WbackE(i)) (8)
E(i+1)＝fout(WoutD(i+1),x(i+1),E(i)) (9)
wherein x (i) is a system parameter with an initial value of 0, f () is an excitation function of a reserve pool node, fout() Is the excitation function of the output unit of the reserve pool, W represents the connection weight matrix of the internal neurons in the reserve pool, WoutRepresenting a matrix of output values;
step 4.3 calculate output value matrix Wout：
Wherein K is the number of neurons in the input layer, N is the number of neurons in the reserve pool, L is the number of neurons in the output layer, and lambda belongs to the element+Representing the regularization factor, | | | | represents the euclidean distance.
And 4.3, taking the echo state network obtained by training as a patrol car tracking route model, accurately calculating the left and right deviation values of the patrol car through the real-time image acquisition system and the trained echo state network, and calibrating the driving route of the patrol car.
Step 5, collecting abnormal conditions on the patrol route: acquiring emergency situations such as help seeking of personnel on a patrol route, illegal crime and the like through a real-time image acquisition system;
the process of collecting abnormal conditions on the patrol route in step 5 can be represented as:
the method comprises the steps that firstly, images around a patrol route are collected through a real-time image collection system, the images are identified and classified through an event classification CNN model, if the patrol process system meets the conditions of help seeking and illegal crime of personnel, the patrol process system of the patrol car orders the patrol car to stop, meanwhile, a video is started to be left, event pictures are uploaded to a cloud, and workers are reminded to perform further intervention.
And 6, starting driving interruption of the patrol car, starting an interruption processing mechanism by the system when the patrol is abnormal within a safe distance, and recording the abnormal condition of the vehicle-mounted terminal by the log module.
The echo state network training sample data used by the invention is subjected to salt and pepper noise processing to simulate the interference of environmental noise, and simultaneously, the stability and robustness of the echo state network model are improved, so that overfitting of the model is avoided to a certain extent. In addition, when the echo state network output obtained by training has errors, the echo state network model is finely adjusted.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.
Claims (5)
1. The patrol car high-precision tracking method based on the echo state network comprises the following specific steps:
step 1, acquiring patrol car driving image data: acquiring a road surface image of the patrol car running through a real-time image acquisition system on the patrol car;
step 2, adding salt and pepper noise to the collected image: adding salt and pepper noise into the road surface image obtained in the step 1 to simulate the interference of the acquired signal in a noise environment, and adding 20-30% of salt and pepper noise;
step 3, detecting lane lines in the image: detecting lane lines in the image by using a Canny edge detection algorithm, and segmenting the lane lines;
step 4, establishing tracking models of the patrol car and the lane lines: training an echo state network by taking the lane line value detected in the step 3 as an input vector, and solving a nonlinear tracking model of the lane line by using the echo state network;
step 5, collecting abnormal conditions on the patrol route: acquiring emergency situations such as help seeking of personnel on a patrol route, illegal crime and the like through a real-time image acquisition system;
and 6, starting driving interruption of the patrol car, starting an interruption processing mechanism by the system when the patrol is abnormal within a safe distance, and recording the abnormal condition of the vehicle-mounted terminal by the log module.
2. The patrol car high-precision tracking method based on the echo state network as claimed in claim 1, wherein: the process of adding salt and pepper noise to the acquired image in the step 2 can be represented as follows:
the noise model of the added salt and pepper is:
wherein, ImaxAnd IminIs the maximum value and the minimum value of the image pixel points, p is the probability of noise occurrence of the image, the value range of p is 20 percent to 30 percent, ixyThe actual value of the collected image pixel point (x, y) is obtained, and the f (x, y) is the image pixel point value after the salt and pepper noise is added.
3. The patrol car high-precision tracking method based on the echo state network as claimed in claim 1, wherein: the process of detecting the lane line in the image in step 3 can be represented as:
step 3.1, smoothing and filtering the road surface image by using a Gaussian smoothing function to inhibit salt and pepper noise in the image;
3.2, calculating the finite difference of the first-order partial derivatives through a sobel operator to calculate the amplitude and the direction of the gradient of the road surface image;
step 3.3, carrying out non-maximum suppression on the amplitude along the gradient direction calculated in the step 3.2;
step 3.4, detecting and connecting edges by using a double-threshold algorithm: selecting two thresholds to set a threshold range, setting the gray value of the pixel less than the gradient value of the threshold range to be 0, assigning 1 to the point mark more than the threshold range, connecting the contour lines of the lane lines, and segmenting to obtain the edge detection line f of the lane line1(x,y)。
4. The patrol car high-precision tracking method based on the echo state network as claimed in claim 1, wherein: the process of establishing the patrol car and lane line tracking model in the step 4 can be represented as follows:
step 4.1, establishing a patrol car and lane line tracking model, wherein the patrol car is subjected to left deviation and right deviation in the driving process, when the patrol car is positioned at the right sides of the lane lines at the two sides, the patrol car needs to be deviated towards the left direction, when the patrol car is positioned at the left sides of the lane lines at the two sides, the patrol car needs to be deviated towards the right direction, and the edge detection line of the training sample road surface picture in the step 3 is marked with a left deviation value E (i) and a right deviation value E (i);
step 4.2, initializing the network, forming a sample D (i) by an edge detection line of the road surface picture and forming a training sample u (i) { D (i), E (i) } by a bias value E (i) of the left and right sides of the patrol car, and inputting a training sample characteristic D (i) into a connection weight matrix W (i)inEntering a reserve pool, E (i) connecting the weights W through feedbackbackEntering a reserve pool, and acquiring the system state and the output state according to the following sequence:
x(i+1)＝f(WinD(i+1)+Wx(i)+WbackE(i)) (2)
E(i+1)＝fout(WoutD(i+1),x(i+1),E(i)) (3)
wherein x (i) is a system parameter with an initial value of 0, and f () is the activation of the pool nodeExcitation function, fout() Is the excitation function of the output unit of the reserve pool, W represents the connection weight matrix of the internal neurons in the reserve pool, WoutRepresenting a matrix of output values;
step 4.3 calculate output value matrix Wout：
Wherein K is the number of neurons in the input layer, N is the number of neurons in the reserve pool, L is the number of neurons in the output layer, and lambda belongs to the element+Expressing a regularization factor, | | | | represents a Euclidean distance;
and 4.3, taking the echo state network obtained by training as a patrol car tracking route model, accurately calculating the left and right deviation values of the patrol car through the real-time image acquisition system and the trained echo state network, and calibrating the driving route of the patrol car.
5. The patrol car high-precision tracking method based on the echo state network as claimed in claim 1, wherein: the process of collecting abnormal conditions on the patrol route in step 5 can be represented as:
the method comprises the steps that firstly, images around a patrol route are collected through a real-time image collection system, the images are identified and classified through an event classification CNN model, if the patrol process system meets the conditions of help seeking and illegal crime of personnel, the patrol process system of the patrol car orders the patrol car to stop, meanwhile, a video is started to be left, event pictures are uploaded to a cloud, and workers are reminded to perform further intervention.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011405929.4A CN112419345A (en) | 2020-12-04 | 2020-12-04 | Patrol car high-precision tracking method based on echo state network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011405929.4A CN112419345A (en) | 2020-12-04 | 2020-12-04 | Patrol car high-precision tracking method based on echo state network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112419345A true CN112419345A (en) | 2021-02-26 |
Family
ID=74830204
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011405929.4A Pending CN112419345A (en) | 2020-12-04 | 2020-12-04 | Patrol car high-precision tracking method based on echo state network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112419345A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113777913A (en) * | 2021-09-02 | 2021-12-10 | 盐城中科高通量计算研究院有限公司 | Patrol car high-precision tracking method based on improved extreme learning machine |
| CN113867341A (en) * | 2021-09-18 | 2021-12-31 | 盐城中科高通量计算研究院有限公司 | Patrol car path planning and tracking algorithm with high-precision tracking and control |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109462821A (en) * | 2018-11-19 | 2019-03-12 | 东软集团股份有限公司 | Method, apparatus, storage medium and the electronic equipment of predicted position |
| US20200026282A1 (en) * | 2018-07-23 | 2020-01-23 | Baidu Usa Llc | Lane/object detection and tracking perception system for autonomous vehicles |
| CN111260957A (en) * | 2020-01-17 | 2020-06-09 | 巢湖学院 | Lane departure warning system |
-
2020
- 2020-12-04 CN CN202011405929.4A patent/CN112419345A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200026282A1 (en) * | 2018-07-23 | 2020-01-23 | Baidu Usa Llc | Lane/object detection and tracking perception system for autonomous vehicles |
| CN109462821A (en) * | 2018-11-19 | 2019-03-12 | 东软集团股份有限公司 | Method, apparatus, storage medium and the electronic equipment of predicted position |
| CN111260957A (en) * | 2020-01-17 | 2020-06-09 | 巢湖学院 | Lane departure warning system |
Non-Patent Citations (2)
| Title |
|---|
| PIETRO VERZELLI等: "Echo State Networks with Self-Normalizing Activations on the Hyper-Sphere", SCIENTIFIC REPORTS, 25 September 2019 (2019-09-25) * |
| 储开斌;郭俊俊;朱栋;: "复杂道路环境下车道线快速提取方法", 实验室研究与探索, no. 07, 25 July 2020 (2020-07-25) * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113777913A (en) * | 2021-09-02 | 2021-12-10 | 盐城中科高通量计算研究院有限公司 | Patrol car high-precision tracking method based on improved extreme learning machine |
| CN113777913B (en) * | 2021-09-02 | 2023-12-08 | 盐城中科高通量计算研究院有限公司 | Patrol car high-precision tracking method based on improved extreme learning machine |
| CN113867341A (en) * | 2021-09-18 | 2021-12-31 | 盐城中科高通量计算研究院有限公司 | Patrol car path planning and tracking algorithm with high-precision tracking and control |
| CN113867341B (en) * | 2021-09-18 | 2023-12-22 | 盐城中科高通量计算研究院有限公司 | Patrol car path planning and tracking method with high-precision tracking and control |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107563372B (en) | License plate positioning method based on deep learning SSD frame | |
| CN111310583B (en) | Vehicle abnormal behavior identification method based on improved long-term and short-term memory network | |
| CN109087510B (en) | Traffic monitoring method and device | |
| CN109344753A (en) | A kind of tiny fitting recognition methods of Aerial Images transmission line of electricity based on deep learning | |
| CN108230254A (en) | A kind of full lane line automatic testing method of the high-speed transit of adaptive scene switching | |
| CN111340855A (en) | Road moving target detection method based on track prediction | |
| CN111091023B (en) | Vehicle detection method and device and electronic equipment | |
| CN108875754B (en) | Vehicle re-identification method based on multi-depth feature fusion network | |
| CN111428558A (en) | Vehicle detection method based on improved YO L Ov3 method | |
| CN112581445A (en) | Detection method and device for bolt of power transmission line, storage medium and electronic equipment | |
| CN112419345A (en) | Patrol car high-precision tracking method based on echo state network | |
| CN107545263A (en) | A kind of object detecting method and device | |
| CN111274886B (en) | Deep learning-based pedestrian red light running illegal behavior analysis method and system | |
| CN110334703B (en) | Ship detection and identification method in day and night image | |
| CN111127520B (en) | Vehicle tracking method and system based on video analysis | |
| CN108445866B (en) | LDW false negative report test method and test system based on convolutional neural network | |
| CN110751266A (en) | Unmanned aerial vehicle trajectory prediction module and prediction method thereof | |
| CN109086803A (en) | A kind of haze visibility detection system and method based on deep learning and the personalized factor | |
| CN109165602A (en) | A kind of black smoke vehicle detection method based on video analysis | |
| CN113129336A (en) | End-to-end multi-vehicle tracking method, system and computer readable medium | |
| CN110852358A (en) | Vehicle type distinguishing method based on deep learning | |
| CN114359669A (en) | Picture analysis model adjusting method and device and computer readable storage medium | |
| CN114842340A (en) | Robot binocular stereoscopic vision obstacle sensing method and system | |
| CN117215316B (en) | Method and system for driving environment perception based on cooperative control and deep learning | |
| CN103886331B (en) | Method for classifying appearances of vehicles based on multi-feature fusion of surveillance video |
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 |