CN108520252B - Road sign identification method based on generalized Hough transform and wavelet transform - Google Patents
Road sign identification method based on generalized Hough transform and wavelet transform Download PDFInfo
- Publication number
- CN108520252B CN108520252B CN201810361896.4A CN201810361896A CN108520252B CN 108520252 B CN108520252 B CN 108520252B CN 201810361896 A CN201810361896 A CN 201810361896A CN 108520252 B CN108520252 B CN 108520252B
- Authority
- CN
- China
- Prior art keywords
- image
- wavelet
- characteristic
- images
- generalized hough
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- 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
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- 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
- G06V10/52—Scale-space analysis, e.g. wavelet analysis
Abstract
The invention discloses a road sign identification method based on generalized Hough transform and wavelet transform, which comprises the following steps: step 1: selecting standard images to construct a characteristic image library; step 2: carrying out denoising operation on the acquired image; and step 3: carrying out edge extraction on the denoised image by utilizing generalized Hough transform; and 4, step 4: extracting low-frequency information of the extracted image by utilizing wavelet decomposition; and 5: establishing a corresponding characteristic image according to the image low-frequency information; step 6: and matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance. The method utilizes wavelet decomposition to obtain the low-frequency information of the image, thereby constructing a characteristic image library, retaining the most essential characteristic information and the most effective characteristic information in the image, and having higher accuracy compared with the traditional identification method.
Description
Technical Field
The invention relates to the technical field of unmanned driving, in particular to a road sign identification method based on generalized Hough transform and wavelet transform.
Background
In recent years, Advanced Driving Assistance System (ADAS) and unmanned technology have been rapidly developed, wherein landmark recognition is one of the hot spots of research. The common road signs include deceleration, stop, speed limit and the like. The probability of accidents can be reduced by accurately detecting and identifying the road signs, and the driving safety is improved. For human beings, after visual information is transmitted to the brain, the brain can process the information according to the existing knowledge so as to recognize and judge the information.
For a machine, due to the fact that the types of road signs are multiple, the distinguishing degree is not large, and the influences of illumination, collection, shielding and dimension are caused, the road signs are difficult to accurately identify from a complex scene.
The prior art comprises the following steps: due to the effect of different color spaces on the segmentation of traffic signs with colors of red, yellow, blue and the like, the road signs can be extracted by using threshold segmentation on the color spaces; aiming at the identification system of the red road sign, the method of red filtering, edge and closed curve is adopted for detection, and an expert system and a neural network are used for feature extraction and target classification, so that an expected target can be achieved; converting an image from an RGB color space to a Gaussian color space, using two quantities which are unchanged for a viewing angle as color representation of the image, and then clustering colors by using K-means to extract a road sign; a system for identifying a stoproad sign is developed, which employs a color segmentation method of HSV space in terms of detection and a neural network method in terms of identification. However, these methods always do not guarantee high accuracy and have limitations in use.
The method is an accurate and universal road sign identification method, and has important significance for improving an assistant driving system and unmanned driving safety.
Disclosure of Invention
The invention aims to provide a method for efficiently and accurately identifying road signs by combining Gaussian smooth filtering and generalized Hough transform.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the method for identifying the road sign based on the generalized Hough transform and the wavelet transform comprises the following steps of:
step 1: selecting standard images to construct a characteristic image library;
step 2: carrying out denoising operation on the acquired image;
and step 3: carrying out edge extraction on the denoised image by utilizing generalized Hough transform;
and 4, step 4: extracting low-frequency information of the extracted image by utilizing wavelet decomposition;
and 5: establishing a corresponding characteristic image according to the image low-frequency information;
step 6: and matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance.
In connection with the above technical scheme, the step 1 specifically comprises: and selecting at least 20 processed standard images as a feature image library of the road sign.
In the technical scheme, the step 2 is specifically as follows: and (3) weakening or removing irrelevant information in the image by using Gaussian smooth filtering, enhancing relevant information in the image and realizing image denoising. In order to verify the effectiveness of the method, artificial noise is added into the standard image, and the denoised image is compared with the standard image;
the Gaussian smoothing filter uses a two-dimensional Gaussian function, and the two-dimensional Gaussian function is divided into two steps: 1) convolving the image with a one-dimensional Gaussian function; 2) the convolution result is convolved with the same one-dimensional gaussian function with a vertical direction.
According to the technical scheme, in the step 3, the denoised image is subjected to edge detection by adopting generalized Hough transform, the generalized Hough transform aiming at all curves is mainly adopted for detecting the edges of geometric shapes, and the algorithm speed is accelerated by utilizing the graphic gradient quantity; the method specifically comprises the following steps:
3-1, creating a template, selecting a reference point in an image area, calculating a gradient direction angle phi of each edge point, expressing the relative positions of the test point and each boundary point by using a connecting line direction angle alpha and a connecting line length R of the test point and each boundary point, determining an edge point by using R, alpha and phi together, arranging the edge points into an R table according to the size of the gradient direction phi once, storing the R table with the stored content as edge point information (R, alpha) and forming the R table.
3-2, edge detection is carried out according to the principle of generalized Hough transform and a preset threshold value, edge point information of an image to be detected is recorded, in a statistical stage, a R table is changed to reposition a reference point and count matching times according to different multi-party rotation changes, and when the matching times of the positioning points are larger than the set threshold value aiming at a multi-target object in the image, the current positioning point is judged to be an image reference point; due to rotation and scaling, a plurality of reference points can appear on the image, and then the effective image needs to be restored according to the coordinate relation.
In connection with the above technical scheme, step 4 specifically comprises:
4-1, analyzing by using a Haar wavelet, defining a scale function, and calculating a corresponding wavelet function;
4-2, performing wavelet transformation on the signals, and obtaining a weighted sum by using different scale functions;
4-3 wavelet basis functions are reduced in scale by the power of 2, wherein small-scale wavelets represent details of signals, and large-scale wavelets represent outlines of signals; wavelet decomposition covers the whole frequency domain, and correlation among the extracted different characteristics is reduced or removed by selecting a proper filter;
4-4 for two-dimensional image data, a fast Mallat wavelet decomposition algorithm is adopted, and the wavelet transformation of each layer decomposes the image into four parts, which respectively represent the low-frequency approximate part, the vertical high-frequency part, the horizontal high-frequency part and the diagonal high-frequency part of the image.
According to the technical scheme, the step 5 specifically comprises the following steps: wavelet transformation is carried out to obtain low-frequency and high-frequency information of the images, a plurality of road sign images are taken to create a training set image library, all the images are subjected to wavelet decomposition, and the images generated by low-frequency wavelet coefficients of all the images are stored as characteristic images.
According to the technical scheme, the step 6 specifically comprises the following steps: and (3) forming a test set by the collected images, matching each low-frequency characteristic image in the test set with a low-frequency image in a characteristic image library, classifying the low-frequency characteristic images into characteristic images with the minimum Euclidean distance, indicating that the identification is successful if the two images represent the same road sign information, and otherwise indicating that the identification is failed.
The invention also provides a landmark identification system based on generalized Hough transform and wavelet transform, which comprises:
the characteristic image library establishing module is used for selecting standard images to establish a characteristic image library;
the denoising module is used for carrying out denoising operation on the acquired image;
the edge extraction module is used for extracting the edge of the denoised image by utilizing generalized Hough transform;
the image low-frequency information extraction module is used for extracting the image low-frequency information from the extracted image by utilizing wavelet decomposition;
the characteristic image establishing module is used for establishing a corresponding characteristic image according to the low-frequency information of the image;
and the identification module is used for matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance.
The invention also provides a computer readable storage medium, which comprises a computer program executable by a processor, wherein the computer program specifically executes the road sign identification system based on the generalized hough transform and the wavelet transform.
The invention has the following beneficial effects: the method can detect the curve with any shape by utilizing the generalized Hough transform, and has wider application compared with the traditional Hough transform detection. The wavelet decomposition is utilized to obtain the low-frequency information of the image, so that a characteristic image library is constructed, the most essential characteristic information and the best identification effect characteristic information in the image are reserved, and compared with the traditional identification method, the accuracy is higher.
Furthermore, the collected image is processed and identified by combining Gaussian smooth filtering, generalized Hough transform and wavelet transform, so that the method is an efficient and accurate identification method, and the identification effect of the method is greatly improved compared with that of the traditional method.
Furthermore, the two-dimensional Gaussian function is utilized for smooth filtering, so that the method has the following advantages:
1) the two-dimensional gaussian function has rotational symmetry, which means that the gaussian smoothing filter is not biased to either direction in the subsequent edge detection.
2) The gaussian function is a single-valued function, which indicates that the gaussian filter replaces the pixel value of the point with the weighted mean of the pixel neighborhood, and the weight of the pixel point of each neighborhood is monotonically increased or decreased with the distance between the point and the central point.
3) The fourier transform spectrum of the gaussian function is single-lobed, which means that the smoothed image is not contaminated by unwanted high frequency signals, while most of the wanted signal is retained.
4) The width of the Gaussian filter is characterized by a parameter sigma, sigma and the smoothing degree are very simple, the larger the sigma is, the wider the bandwidth of the Gaussian filter is, and the better the smoothing degree is.
5) Due to the separability of the gaussian function, large gaussian filters can be effectively implemented.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flowchart of a road sign recognition method based on generalized Hough transform and wavelet transform according to the present invention;
FIG. 2 is a gallery of training sets of the present invention;
FIG. 3 is a comparison diagram after denoising by Gaussian smoothing filtering according to the present invention;
FIG. 4 is an edge image obtained by generalized Hough transform according to the present invention
FIG. 5 is an image obtained using wavelet decomposition in accordance with the present invention;
FIG. 6 is a library of feature images of the present invention;
fig. 7 and 8 are graphs showing the test results of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The embodiment of the invention discloses a road sign identification calculation method based on generalized Hough transform and wavelet transform, which comprises the following steps as shown in figure 1:
step 1: selecting standard images to construct a characteristic image library;
step 2: carrying out denoising operation on the acquired image;
and step 3: carrying out edge extraction on the denoised image by utilizing generalized Hough transform;
and 4, step 4: extracting low-frequency information of the extracted image by utilizing wavelet decomposition;
and 5: establishing a corresponding characteristic image according to the obtained low-frequency information;
step 6: matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance;
the step 1 specifically comprises the following steps: 20 standard images are selected to construct a feature map library.
The step 2 specifically comprises the following steps:
2-1, firstly, denoising the image by adopting Gaussian smoothing filtering. In order to verify the effectiveness of the method, artificial noise is added into the standard image, and the denoised image is compared with the standard image.
2-2 this patentThe gaussian templates used were:
2-3 the Gaussian smoothing filter of this patent uses a two-dimensional Gaussian function, which is divided into two steps: 1) convolving the image with a one-dimensional Gaussian function; 2) the convolution result is convolved with the same one-dimensional gaussian function with a vertical direction.
Step 3, performing edge extraction on the denoised image by utilizing generalized Hough transform, and specifically comprising the following steps:
3-1 first create a template, select a reference point (typically the centroid) in the image region, then calculate the gradient direction angle phi for each edge point, and the relative positions of the test point and each boundary point can be represented by their link direction angle alpha and link length r. R, alpha and phi jointly determine an edge point, arrange the edge points into an R table according to the size of the gradient direction phi once and store the R table, and store the edge point information (R, alpha) to form the R table.
And 3-2, performing edge detection according to the principle of generalized Hough transform and a preset threshold, recording edge point information of the image to be detected, and changing the R table to reposition a reference point and count the matching times according to different multi-party rotation changes in a counting stage. And when the matching times of the positioning points are larger than a set threshold value aiming at the multi-target object in the image, judging that the current positioning points are image reference points. Due to rotation and scaling, a plurality of reference points can appear on the image, and then the effective image needs to be restored according to the coordinate relation.
In the step 4, extracting the low-frequency information of the extracted image by using wavelet decomposition specifically comprises the following steps:
4-1 using Haar wavelets for analysis, defining a scale function as
4-2 calculating the corresponding wavelet function as
4-3, wavelet transform is carried out on the signals, and different scale functions are used for obtaining the weighted sum.
The 4-4 wavelet basis functions are scaled down by a power of 2, with small-scale wavelets representing the details of the signal and large-scale wavelets representing the contours of the signal. Wavelet decomposition can cover the whole frequency domain, and by selecting a proper filter, the correlation among different extracted features can be greatly reduced or removed.
4-5 for two-dimensional image data, a fast Mallat wavelet decomposition algorithm is used. The wavelet transform of each layer decomposes the image into four parts, representing the low frequency approximation, the vertical high frequency, the horizontal high frequency, and the diagonal high frequency parts of the image.
4-6 represent different information with the scale function:
where HH, LH, and HL represent diagonal, vertical, and horizontal high frequency information, respectively, LL represents low frequency information,
is a wavelet basis function, ψ (x) is a scale function, and g (x) is a weighted sum of wavelet basis functions.
The step 5 of establishing a corresponding feature image according to the obtained low-frequency information specifically includes:
the 5-1 wavelet transform obtains low-frequency and high-frequency information of an image, the high-frequency information is generally interference information in the image, and the low-frequency information is intrinsic characteristic information with good identification effect. Therefore, a training set image library is created by taking a plurality of road sign images, all the images are subjected to wavelet decomposition, the images generated by low-frequency wavelet coefficients of the images are stored as characteristic images, and a characteristic image library is established.
In step 6, the matching and identification of the established characteristic image and the characteristic image library by using the Euclidean distance specifically comprises the following steps:
6-1, forming a test set by the shot images, matching each low-frequency image in the test set with a low-frequency image in a characteristic image library, classifying the low-frequency images into characteristic images with the minimum Euclidean distance, wherein if the two images represent the same road sign information, the identification is successful, and otherwise, the identification is failed.
The invention adopts a wavelet transform-based feature extraction and identification method for feature extraction and identification, overcomes the defects of the traditional Fourier-based feature extraction method, and ensures that neither high-frequency information nor low-frequency information is lost; the experimental results compared to the standard images show that: the algorithm can overcome the influence of Gaussian white noise caused by weather change and illumination change and noise caused by micro rotation, scale change and translation of the collected road sign image. Compared with the prior art, the method can effectively identify the road sign information of the road surface in real time, has high precision and strong reliability, has small judgment error rate in the identification process, and can be widely used in an advanced driving auxiliary system of the unmanned vehicle.
The invention also provides a landmark identification system based on generalized Hough transform and wavelet transform, which comprises:
the characteristic image library establishing module is used for selecting standard images to establish a characteristic image library;
the denoising module is used for carrying out denoising operation on the acquired image;
the edge extraction module is used for extracting the edge of the denoised image by utilizing generalized Hough transform;
the image low-frequency information extraction module is used for extracting the image low-frequency information from the extracted image by utilizing wavelet decomposition;
the characteristic image establishing module is used for establishing a corresponding characteristic image according to the low-frequency information of the image;
and the identification module is used for matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance.
The invention also provides a computer readable storage medium, which comprises a computer program executable by a processor, wherein the computer program specifically executes the road sign identification system based on the generalized hough transform and the wavelet transform.
In a specific embodiment of the invention, the landmark identification method based on generalized Hough transform and wavelet transform comprises the following steps:
step 1: constructing a training set gallery, such as the training set gallery constructed by the invention shown in FIG. 2;
step 2: carrying out denoising operation on the acquired image, such as a comparison graph obtained after denoising a certain noise image shown in FIG. 3;
and step 3: carrying out edge extraction on the denoised image by utilizing generalized Hough transform, and obtaining an edge image according to the invention as shown in FIG. 4;
and 4, step 4: extracting low-frequency information of an image by using wavelet decomposition, and as shown in fig. 5, obtaining the image by using wavelet decomposition according to the present invention;
and 5: establishing a corresponding characteristic image library according to the low-frequency information, for example, fig. 6 is a characteristic image library established by using the low-frequency information according to the present invention;
step 6: identifying by adopting Euclidean distance matching images, and obtaining test result graphs by the invention as shown in figures 7 and 8;
when the method is used for field test, the average time of each frame of the algorithm is 61ms, the accuracy rate is over 90 percent, and the method has good real-time performance and accuracy.
In summary, the two-dimensional gaussian function for smoothing filtering of the present invention has the following advantages:
1) the two-dimensional gaussian function has rotational symmetry, which means that the gaussian smoothing filter is not biased to either direction in the subsequent edge detection.
2) The gaussian function is a single-valued function, which indicates that the gaussian filter replaces the pixel value of the point with the weighted mean of the pixel neighborhood, and the weight of the pixel point of each neighborhood is monotonically increased or decreased with the distance between the point and the central point.
3) The fourier transform spectrum of the gaussian function is single-lobed, which means that the smoothed image is not contaminated by unwanted high frequency signals, while most of the wanted signal is retained.
4) The width of the Gaussian filter is characterized by a parameter sigma, sigma and the smoothing degree are very simple, the larger the sigma is, the wider the bandwidth of the Gaussian filter is, and the better the smoothing degree is.
5) Due to the separability of the gaussian function, large gaussian filters can be effectively implemented.
The method can detect the curve with any shape by utilizing the generalized Hough transform, and has wider application compared with the traditional Hough transform detection.
The method utilizes wavelet decomposition to obtain the low-frequency information of the image, thereby constructing a characteristic image library, retaining the most essential characteristic information and the most effective characteristic information in the image, and having higher accuracy compared with the traditional identification method.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.
Claims (9)
1. A road sign identification method based on generalized Hough transform and wavelet transform is characterized by comprising the following steps:
step 1: selecting standard images to construct a characteristic image library;
step 2: carrying out denoising operation on the acquired image;
and step 3: carrying out edge extraction on the denoised image by utilizing generalized Hough transform;
and 4, step 4: extracting low-frequency information of the extracted image by utilizing wavelet decomposition;
and 5: establishing a corresponding characteristic image according to the image low-frequency information;
step 6: and matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance.
2. The landmark identification method based on generalized Hough transform and wavelet transform according to claim 1, wherein the step 1 specifically comprises: and selecting at least 20 processed standard images as a feature image library of the road sign.
3. The landmark identification method based on generalized Hough transform and wavelet transform according to claim 1, wherein the step 2 is specifically as follows: denoising the collected image, and drying the image by adopting Gaussian smooth filtering, wherein artificial noise is added into the standard image for verifying the effectiveness of the method, and the denoised image is compared with the standard image;
the Gaussian smoothing filter uses a two-dimensional Gaussian function, and the two-dimensional Gaussian function is divided into two steps: 1) convolving the image with a one-dimensional Gaussian function; 2) the convolution result is convolved with the same one-dimensional gaussian function with a vertical direction.
4. The landmark identification method based on generalized Hough transform and wavelet transform as claimed in claim 1, wherein the edge detection of the denoised image by generalized Hough transform in step 3 specifically comprises:
3-1, creating a template, selecting a reference point in an image area, calculating a gradient direction angle of each edge point, representing the relative positions of the reference point and each edge point by using a connecting line direction angle and a connecting line length of the reference point and each edge point, determining an edge point by using the connecting line length, the connecting line direction angle and the gradient direction angle, arranging the edge points into a table at one time according to the size of the gradient direction angle, storing the table, and forming the table by using the stored content as edge point information;
3-2, edge detection is carried out according to the principle of generalized Hough transform and a preset threshold value, edge point information of an image to be detected is recorded, in a statistical stage, a table is changed to reposition a reference point and count matching times according to different multi-party rotation changes, and when the matching times of the positioning points are larger than the set threshold value aiming at a multi-target object in the image, the current positioning point is judged to be an image reference point; due to rotation and scaling, a plurality of reference points can appear on the image, and then the effective image needs to be restored according to the coordinate relation.
5. The landmark identification method based on generalized Hough transform and wavelet transform according to claim 1, wherein the step 4 specifically comprises:
4-1, analyzing by using a Haar wavelet, defining a scale function, and calculating a corresponding wavelet function;
4-2, performing wavelet transformation on the signals, and obtaining a weighted sum by using different scale functions;
4-3 wavelet basis functions are reduced in scale by the power of 2, wherein small-scale wavelets represent details of signals, and large-scale wavelets represent outlines of signals; wavelet decomposition covers the whole frequency domain, and correlation among the extracted different characteristics is reduced or removed by selecting a proper filter;
4-4 for two-dimensional image data, a fast Mallat wavelet decomposition algorithm is adopted, and the wavelet transformation of each layer decomposes the image into four parts, which respectively represent the low-frequency approximate part, the vertical high-frequency part, the horizontal high-frequency part and the diagonal high-frequency part of the image.
6. The landmark identification method based on generalized Hough transform and wavelet transform according to claim 1, wherein the step 5 specifically comprises: wavelet transformation is carried out to obtain low-frequency and high-frequency information of the images, a plurality of road sign images are taken to create a training set image library, all the images are subjected to wavelet decomposition, and the images generated by low-frequency wavelet coefficients of all the images are stored as characteristic images.
7. The landmark identification method based on generalized Hough transform and wavelet transform according to claim 1, wherein the step 6 specifically comprises: and (3) forming a test set by the collected images, matching each low-frequency characteristic image in the test set with a low-frequency image in a characteristic image library, classifying the low-frequency characteristic images into characteristic images with the minimum Euclidean distance, indicating that the identification is successful if the two images represent the same road sign information, and otherwise indicating that the identification is failed.
8. A road sign recognition system based on generalized Hough transform and wavelet transform is characterized by comprising:
the characteristic image library establishing module is used for selecting standard images to establish a characteristic image library;
the denoising module is used for carrying out denoising operation on the acquired image;
the edge extraction module is used for extracting the edge of the denoised image by utilizing generalized Hough transform;
the image low-frequency information extraction module is used for extracting the image low-frequency information from the extracted image by utilizing wavelet decomposition;
the characteristic image establishing module is used for establishing a corresponding characteristic image according to the low-frequency information of the image;
and the identification module is used for matching and identifying the established characteristic image and the characteristic image library by adopting the Euclidean distance.
9. A computer-readable storage medium, comprising a computer program executable by a processor, the computer program specifically performing the landmark identification method based on the generalized hough transform and the wavelet transform according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810361896.4A CN108520252B (en) | 2018-04-20 | 2018-04-20 | Road sign identification method based on generalized Hough transform and wavelet transform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810361896.4A CN108520252B (en) | 2018-04-20 | 2018-04-20 | Road sign identification method based on generalized Hough transform and wavelet transform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108520252A CN108520252A (en) | 2018-09-11 |
| CN108520252B true CN108520252B (en) | 2022-03-01 |
Family
ID=63428942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810361896.4A Active CN108520252B (en) | 2018-04-20 | 2018-04-20 | Road sign identification method based on generalized Hough transform and wavelet transform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108520252B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858409A (en) * | 2019-01-18 | 2019-06-07 | 深圳壹账通智能科技有限公司 | Manual figure conversion method, device, equipment and medium |
| CN113688828B (en) * | 2021-07-23 | 2023-09-29 | 山东云海国创云计算装备产业创新中心有限公司 | Bad element identification method and related device |
| CN117197072B (en) * | 2023-09-07 | 2024-04-05 | 石家庄铁道大学 | Automatic object counting method based on machine vision |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1897015A (en) * | 2006-05-18 | 2007-01-17 | 王海燕 | Method and system for inspecting and tracting vehicle based on machine vision |
| CN106022218A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Palm print palm vein image layer fusion method based on wavelet transformation and Gabor filter |
| CN106324618A (en) * | 2015-06-17 | 2017-01-11 | 百利得汽车主动安全系统（苏州）有限公司 | System for detecting lane line based on laser radar and realization method thereof |
| CN107292263A (en) * | 2017-06-19 | 2017-10-24 | 深圳市创艺工业技术有限公司 | A kind of motor-driven, electric vehicle automatic driving recognition system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010117575A2 (en) * | 2009-04-07 | 2010-10-14 | Virginia Commonwealth University | Accurate pelvic fracture detection for x-ray and ct images |
-
2018
- 2018-04-20 CN CN201810361896.4A patent/CN108520252B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1897015A (en) * | 2006-05-18 | 2007-01-17 | 王海燕 | Method and system for inspecting and tracting vehicle based on machine vision |
| CN106324618A (en) * | 2015-06-17 | 2017-01-11 | 百利得汽车主动安全系统（苏州）有限公司 | System for detecting lane line based on laser radar and realization method thereof |
| CN106022218A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Palm print palm vein image layer fusion method based on wavelet transformation and Gabor filter |
| CN107292263A (en) * | 2017-06-19 | 2017-10-24 | 深圳市创艺工业技术有限公司 | A kind of motor-driven, electric vehicle automatic driving recognition system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108520252A (en) | 2018-09-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109657632B (en) | Lane line detection and identification method | |
| CN107492076B (en) | Method for suppressing vehicle shadow interference in expressway tunnel scene | |
| CN107085708B (en) | High-resolution remote sensing image change detection method based on multi-scale segmentation and fusion | |
| Paunwala et al. | A novel multiple license plate extraction technique for complex background in Indian traffic conditions | |
| CN108921813B (en) | Unmanned aerial vehicle detection bridge structure crack identification method based on machine vision | |
| CN106778551B (en) | Method for identifying highway section and urban road lane line | |
| CN108520252B (en) | Road sign identification method based on generalized Hough transform and wavelet transform | |
| Li et al. | Road lane detection with gabor filters | |
| CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
| CN107992856B (en) | High-resolution remote sensing building shadow detection method under urban scene | |
| CN109886168B (en) | Ground traffic sign identification method based on hierarchy | |
| CN109858438B (en) | Lane line detection method based on model fitting | |
| CN111476804A (en) | Method, device and equipment for efficiently segmenting carrier roller image and storage medium | |
| CN111027544A (en) | MSER license plate positioning method and system based on visual saliency detection | |
| CN112597812A (en) | Finger vein identification method and system based on convolutional neural network and SIFT algorithm | |
| Bulugu | Algorithm for license plate localization and recognition for tanzania car plate numbers | |
| CN106599891A (en) | Remote sensing image region-of-interest rapid extraction method based on scale phase spectrum saliency | |
| CN114581658A (en) | Target detection method and device based on computer vision | |
| CN117197700B (en) | Intelligent unmanned inspection contact net defect identification system | |
| CN109753880B (en) | Detection and identification method for natural scene vehicle-mounted video road speed limit sign | |
| CN109271902B (en) | Infrared weak and small target detection method based on time domain empirical mode decomposition under complex background | |
| CN111476233A (en) | License plate number positioning method and device | |
| Mol et al. | Text recognition using poisson filtering and edge enhanced maximally stable extremal regions | |
| Scharfenberger et al. | Image saliency detection via multi-scale statistical non-redundancy modeling | |
| CN115512310A (en) | Vehicle type recognition method and system based on face features under video monitoring |
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 |