CN110490194A - A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight - Google Patents
A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight Download PDFInfo
- Publication number
- CN110490194A CN110490194A CN201910673010.4A CN201910673010A CN110490194A CN 110490194 A CN110490194 A CN 110490194A CN 201910673010 A CN201910673010 A CN 201910673010A CN 110490194 A CN110490194 A CN 110490194A
- Authority
- CN
- China
- Prior art keywords
- image
- traffic sign
- feature
- pixel
- gray
- 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
Classifications
-
- 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
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- 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
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/467—Encoded features or binary features, e.g. local binary patterns [LBP]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/09—Recognition of logos
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight, and the histogram equalization including the Traffic Sign Images convert color spaces of acquired training sample set to be carried out to color lightness enhances；Traffic Sign Images carry out gray scale normalization pretreatment again；Piecemeal is carried out to pretreated image, calculates the histograms of oriented gradients HOG feature and image texture LBP feature of block image；To Normalized Grey Level image averaging piecemeal, the histograms of oriented gradients HOG of each piecemeal is extracted using information Entropy Method and image texture LBP feature carries out the fusion of piecemeal weight, obtains the HOG-LBP total characteristic of image；The feature training classifier that the HOG-LBP total characteristic input learning model training of image is merged；The fusion feature for extracting test sample is input to feature training classifier and obtains sorted recognition result.The present invention solves the problems, such as that different parts feature has the identification of different contribution amount brings to identification in traffic sign by piecemeal multiple features fusion.
Description
Technical field
The present invention relates to digital image processing fields, melt more particularly, to a kind of multiple features piecemeal of adaptive weight
Close the recognition methods of traffic sign.
Background technique
The detection and identification of traffic sign are in intelligent transportation system ITS (Intelligent Transport System)
Assist the important component driven.Environmental factor receive more and more attention in recent years, that outdoor traffic mark is subject to
Influence is very big, such as: the problems such as traffic sign is blocked by trees or billboard, and uneven illumination is even and traffic sign is stained.This
A little problems increase the difficulty of Traffic Sign Recognition.
There are two types of the main recognition methods of traffic sign: one is traditional based on artificial design features and trains classifier
Method, common feature have HOG, LBP, SIFT etc., and common classifier has SVM, KNN, AdaBoost and ELM.Another kind side
Method is by neural network come to image study feature and the method classified, for traditional characteristic, neural network
The feature learnt is more abstracted and higher-dimension, is more advantageous to classification, but the sample size that neural network learning needs is big, and the time is multiple
Miscellaneous degree is high, is not suitable for real-time scene.Since the feature at each position of traffic sign is different to the identification role of traffic sign
Sample, for same class traffic sign mark, closer to center, the discrimination of identification is higher.
Summary of the invention
The present invention in order to overcome at least one of the drawbacks of the prior art described above (deficiency), provides a kind of adaptive weight
The recognition methods of multiple features segment fusion traffic sign.
The present invention is directed to solve above-mentioned technical problem at least to a certain extent.
In order to reach above-mentioned technical effect, technical scheme is as follows:
A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight, which is characterized in that including as follows
Step:
The histogram that the Traffic Sign Images convert color spaces of acquired training sample set are carried out color lightness by S10 is equal
Weighing apparatusization enhancing；
S20 carries out gray scale normalization to the enhanced Traffic Sign Images of lightness and pre-processes to obtain Normalized Grey Level image；
The histograms of oriented gradients HOG feature and image texture LBP feature of S30 calculating Normalized Grey Level image；
S40 extracts the histograms of oriented gradients of each piecemeal using information Entropy Method to Normalized Grey Level image averaging piecemeal
HOG and image texture LBP feature carry out the fusion of piecemeal weight, obtain the HOG-LBP total characteristic of image；
The feature training classifier that S50 is merged the HOG-LBP total characteristic input learning model training of image；
The fusion feature that S60 extracts test sample is input to feature training classifier and obtains sorted recognition result.
Preferably, the histograms of oriented gradients HOG feature of each image is calculated to Normalized Grey Level image in the S40
Method are as follows:
(1) gamma correction processing is carried out to Normalized Grey Level image:
I (x, y)=I (x, y)gamma, gamma=0.5 (2), wherein I (x, y) is the gray scale of coordinate pixel (x, y)
Value；
(2) gradient magnitude and direction are calculated to the image after correction process using following formula:
Gx(x, y)=I (x+1, y)-I (x-1, y) (3)
Gy(x, y)=I (x, y+1)-I (x, y-1) (4)
Wherein I (x, y) is the gray value of coordinate pixel (x, y), Gx(x, y) be coordinate pixel (x+1, y) with (x-1,
Y) difference of gray value, the i.e. gradient value of coordinate pixel (x, y) horizontal direction, similarly, Gy(x, y) is coordinate pixel (x, y
+ 1) with the difference of the gray value of (x, y-1), i.e. coordinate pixel (x, y) vertical gradient value, G (x, y) is coordinate pixel
The gradient magnitude of (x, y), α (x, y) are the gradient direction of coordinate pixel (x, y)；
(3) image averaging is divided into the cell block of 8 × 8 pixels, and the histogram of gradients for counting each cell block obtains
The HOG feature of each cell block；
(4) 2 × 2 cell blocks are formed into a block image, the feature of all cell blocks in block image is retouched
It states series connection and obtains the HOG feature of the block image.
Preferably, the method for the image texture LBP feature of each image is calculated Normalized Grey Level image in the S40
Are as follows:
(1) Normalized Grey Level image averaging is divided into and forms subgraph by 3 × 3 sub-images, pass through following formula
Calculate the LBP value of each sub-image pixel in each subgraph:
Wherein (xc,yc) be sub-image central element, I (p) be I (p) surrounding pixel values, I (c) be center pixel value,
S (x) is sign function, otherwise it is 0 that when surrounding pixel gray value is greater than center pixel gray value, value, which is 1,；
(2) statistics with histogram is carried out to each sub-image, obtains the histogram of subgraph；
(3) histogram of all subgraphs is normalized using following formula:Wherein N is subgraph
Total pixel of picture, g are gray level, NgFor grey level histogram, PgFor the histogram of Normalized Grey Level image；
(4) normalization histogram for connecting all subgraphs, obtains the LBP feature of complete image.
Preferably, the color modular space of Traffic Sign Images collected is RGB in the S10, and the S20 is to traffic mark
The method that will image carries out the histogram equalization enhancing of color lightness are as follows:
The rgb space of Traffic Sign Images is converted into HSV space, the lightness V component for extracting HSV space carries out histogram
Equalization enhancing.
Preferably, gray scale normalization pre-processes in the S30 specifically:
S301 carries out gray proces to the enhanced Traffic Sign Images of lightness using weighted average and obtains gray level image；
S302 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
Preferably, gray scale normalization pre-processes in the S30 specifically:
The enhanced Traffic Sign Images of lightness are carried out color space inverse transformation by S301, and HSV space is converted to RGB sky
Between；
S302 is weighted and averaged to obtain gray level image using following formula:
Gray (x, y)=0.299*R+0.587*G+0.114*B,
Wherein Gray (x, y) is pixel (x, y) gray value that weighted calculation obtains, and R, G, B are respectively pixel (x, y)
Color primaries component；
S303 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
Preferably, the S50 is specifically included:
S401 is divided into m block image to Normalized Grey Level image averaging, and the entropy weight of i-th block of image is determined by following formula:
Wherein, i is any integer value in (0, m), and n is number of greyscale levels,Indicate kth grade picture
The probability that element occurs, Ei(i=1,2 ..., m) be i-th block of image comentropy, Wi(i=1 2 ..., m) is calculated
The weight of i-th block of image；
S402 to m block image extract respectively histograms of oriented gradients HOG and image texture LBP feature using following formula into
The fusion of row weight, obtains the HOG-LBP total characteristic of image:
Preferably, learning model includes linear SVM SVM, BP neural network or the study of the ELM limit in the S50
Machine.
Compared with prior art, the beneficial effect of technical solution of the present invention is: the present invention proposes a kind of based on edge and line
The traffic sign recognition method for managing characteristic block fusion, since the feature at each position of traffic sign plays the identification of traffic sign
Effect it is different, for same mark, closer to center, the discrimination of identification is higher, and the present invention is melted by multiple features
The imperfection for overcoming single features is closed, weight fusion treatment determined by the piecemeal further through image overcomes traffic sign
In each Partial Feature to identification there is the problem of different contribution amounts.
Detailed description of the invention
Fig. 1 is the method flow diagram that the present invention one is implemented.
Specific embodiment
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent；
In order to better illustrate this embodiment, the certain components of attached drawing have omission, zoom in or out, and do not represent actual product
Size；
To those skilled in the art, it is to be understood that certain known features and its explanation, which may be omitted, in attached drawing
's.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
As shown in Figure 1, the invention discloses a kind of identification sides of the multiple features segment fusion traffic sign of adaptive weight
Method includes the following steps:
The histogram that the Traffic Sign Images convert color spaces of acquired training sample set are carried out color lightness by S10 is equal
Weighing apparatusization enhancing；
S20 carries out gray scale normalization to the enhanced Traffic Sign Images of lightness and pre-processes to obtain Normalized Grey Level image；
The histograms of oriented gradients HOG feature and image texture LBP feature of S30 calculating Normalized Grey Level image；
S40 extracts the histograms of oriented gradients of each piecemeal using information Entropy Method to Normalized Grey Level image averaging piecemeal
HOG and image texture LBP feature carry out the fusion of piecemeal weight, obtain the HOG-LBP total characteristic of image；
The feature training classifier that S50 is merged the HOG-LBP total characteristic input learning model training of image；
The fusion feature that S60 extracts test sample is input to feature training classifier and obtains sorted recognition result.
In embodiments of the present invention, it is contemplated that identification role of the feature at each position of traffic sign to traffic sign
Different, for same middle mark, closer to center, the discrimination of identification is higher, and the present invention is overcome by multiple features fusion
The imperfection of single features, weight fusion treatment determined by the piecemeal further through image overcome in traffic sign each
A Partial Feature has the problem of different contribution amounts to identification.
Preferably, the histograms of oriented gradients HOG spy of each block image is calculated Normalized Grey Level image in the S40
The method of sign are as follows:
(1) gamma correction processing is carried out to Normalized Grey Level image:
I (x, y)=I (x, y)gamma, gamma=0.5 (2),
Wherein I (x, y) is the gray value of coordinate pixel (x, y)；
(2) gradient magnitude and direction are calculated to block image after correction process using following formula:
Gx(x, y)=I (x+1, y)-I (x-1, y) (3)
Gy(x, y)=I (x, y+1)-I (x, y-1) (4)
Wherein I (x, y) is the gray value of coordinate pixel (x, y), Gx(x, y) be coordinate pixel (x+1, y) with (x-1,
Y) difference of gray value, the i.e. gradient value of coordinate pixel (x, y) horizontal direction, similarly, Gy(x, y) is coordinate pixel (x, y
+ 1) with the difference of the gray value of (x, y-1), i.e. coordinate pixel (x, y) vertical gradient value, G (x, y) is coordinate pixel
The gradient magnitude of (x, y), α (x, y) are the gradient direction of coordinate pixel (x, y)；
(3) image averaging is divided into the cell block of 8 × 8 pixels, and the histogram of gradients for counting each cell block obtains
The HOG feature of each cell block；
(4) 2 × 2 cell blocks are formed into a block image, by the feature description string of all cell blocks in block image
Connection obtains the HOG feature of the block image.
In embodiments of the present invention, The present invention gives the direction gradient for calculating each image to Normalized Grey Level image is straight
A kind of specific implementation method of side's figure HOG feature, it should be appreciated that ground is that have and be not limited to such method.
Preferably, the method for the image texture LBP feature of each image is calculated Normalized Grey Level image in the S40
Are as follows:
(1) Normalized Grey Level image averaging is divided into and forms subgraph by 3 × 3 sub-images, pass through following formula
Calculate the LBP value of each sub-image pixel in each subgraph:
Wherein (xc,yc) be sub-image central element, I (p) be I (p) surrounding pixel values, I (c) be center pixel value,
S (x) is sign function, otherwise it is 0 that when surrounding pixel gray value is greater than center pixel gray value, value, which is 1,；
(2) statistics with histogram is carried out to each sub-image, obtains the histogram of subgraph；
(3) histogram of all subgraphs is normalized using following formula:Wherein N is subgraph
Total pixel of picture, g are gray level, NgFor grey level histogram, PgFor the histogram of Normalized Grey Level image；
(4) normalization histogram for connecting all subgraphs, obtains the LBP feature of complete image.
In embodiments of the present invention, The present invention gives the image textures for calculating Normalized Grey Level image each image
A kind of specific implementation method of LBP feature, it should be appreciated that ground is that have and be not limited to such method.
Preferably, the color modular space of Traffic Sign Images collected is RGB in the S10, and the S20 is to traffic mark
The method that will image carries out the histogram equalization enhancing of color lightness are as follows:
The rgb space of Traffic Sign Images is converted into HSV space, the lightness V component for extracting HSV space carries out histogram
Equalization enhancing.
In embodiments of the present invention, the data source of Traffic Sign Images acquisition of the present invention is traffic sign database, data
The source figure in library is stored mostly with the color mode of RGB or space, if the color mode of the Traffic Sign Images of acquisition is RGB, is turned
It is changed to HSV space, the lightness V component for extracting HSV space carries out histogram equalization enhancing.
Preferably, gray scale normalization pre-processes in the S30 specifically:
The enhanced Traffic Sign Images of lightness are carried out color space inverse transformation by S301, and HSV space is converted to RGB sky
Between；
S302 is weighted and averaged to obtain gray level image using following formula:
Gray (x, y)=0.299*R+0.587*G+0.114*B,
Wherein Gray (x, y) is pixel (x, y) gray value that weighted calculation obtains, and R, G, B are respectively pixel (x, y)
Color primaries component；
S303 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
Preferably, gray scale normalization pre-processes in the S30 specifically:
S301 carries out gray proces to the enhanced Traffic Sign Images of lightness using weighted average and obtains gray level image；
S302 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
Preferably, the S50 is specifically included:
S401 is divided into m block image to Normalized Grey Level image averaging, and the entropy weight of i-th block of image is determined by following formula:
Wherein, i is any integer value in (0, m), and n is number of greyscale levels,Indicate kth grade picture
The probability that element occurs, Ei(i=1,2 ..., m) be i-th block of image comentropy, Wi(i=1 2 ..., m) is calculated
The weight of i-th block of image；
S402 to m block image extract respectively histograms of oriented gradients HOG and image texture LBP feature using following formula into
The fusion of row weight, obtains the HOG-LBP total characteristic of image:
Preferably, learning model includes linear SVM SVM, BP neural network or the study of the ELM limit in the S50
Machine.
In embodiments of the present invention, learning model of the invention has and is not limited to linear SVM SVM, BP nerve net
Network or ELM extreme learning machine, experimental result at present, suitable for the optimal learning model of comprehensive performance of the invention be linearly support to
Amount machine SVM, BP neural network are preferred in the complete situation of feature.
The same or similar label correspond to the same or similar components；
Described in attached drawing positional relationship for only for illustration, should not be understood as the limitation to this patent；
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention
Protection scope within.
Claims (8)
1. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight, which is characterized in that including walking as follows
It is rapid:
The Traffic Sign Images convert color spaces of acquired training sample set are carried out the histogram equalization of color lightness by S10
Change enhancing；
S20 carries out gray scale normalization to the enhanced Traffic Sign Images of lightness and pre-processes to obtain Normalized Grey Level image；
The histograms of oriented gradients HOG feature and image texture LBP feature of S30 calculating Normalized Grey Level image；
S40 extracts the histograms of oriented gradients HOG of each piecemeal using information Entropy Method to Normalized Grey Level image averaging piecemeal
And image texture LBP feature carries out the fusion of piecemeal weight, obtains the HOG-LBP total characteristic of image；
The feature training classifier that S50 is merged the HOG-LBP total characteristic input learning model training of image；
The fusion feature that S60 extracts test sample is input to feature training classifier and obtains sorted recognition result.
2. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is, the direction gradient of each image is calculated Normalized Grey Level image (normalized image is 48x48 pixel) in the S40
The method of histogram HOG feature are as follows:
(1) gamma correction processing is carried out to Normalized Grey Level image:
I (x, y)=I (x, y)gamma, gamma=0.5 (2), wherein I (x, y) is the gray scale of coordinate pixel (x, y)
Value；
(2) gradient magnitude and direction are calculated to block image after correction process using following formula:
Gx(x, y)=I (x+1, y)-I (x-1, y) (3)
Gy(x, y)=I (x, y+1)-I (x, y-1) (4)
Wherein I (x, y) is the gray value of coordinate pixel (x, y), Gx(x, y) is coordinate pixel (x+1, y) and (x-1, y)
The difference of gray value, the i.e. gradient value of coordinate pixel (x, y) horizontal direction, similarly, Gy(x, y) is coordinate pixel (x, y+1)
With the difference of the gray value of (x, y-1), i.e. coordinate pixel (x, y) vertical gradient value, G (x, y) is coordinate pixel (x, y)
Gradient magnitude, α (x, y) be coordinate pixel (x, y) gradient direction；
(3) histogram of gradients that image averaging is divided into the cell block of 8 × 8 pixels, and counts each cell block obtains each
The HOG feature of cell block；
(4) 2 × 2 cell blocks are formed into a block image, the feature description of all cell blocks in block image is connected
To the HOG feature of the block image.
3. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is, the method for the image texture LBP feature of each image is calculated Normalized Grey Level image in the S40 are as follows:
(1) Normalized Grey Level image averaging is divided into and forms subgraph by 3 × 3 sub-images, calculated by following formula
The LBP value of each sub-image pixel in each subgraph:
Wherein (xc,yc) be sub-image central element, I (p) be surrounding pixel values, I (c) be center pixel value, s (x) be symbol
Number function, otherwise it is 0 that when surrounding pixel gray value is greater than center pixel gray value, value, which is 1,；
(2) statistics with histogram is carried out to each sub-image, obtains the histogram of subgraph；
(3) histogram of all subgraphs is normalized using following formula:Wherein N is subgraph
Total pixel, g are gray level, NgFor grey level histogram, PgFor the histogram of Normalized Grey Level image；
(4) normalization histogram for connecting all subgraphs, obtains the LBP feature of complete image.
4. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is, the color modular space of Traffic Sign Images collected is RGB in the S10, the S20 to Traffic Sign Images into
The method of the histogram equalization enhancing of row color lightness are as follows:
The rgb space of Traffic Sign Images is converted into HSV space, the lightness V component for extracting HSV space carries out histogram equalization
Change enhancing.
5. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is that gray scale normalization pre-processes in the S30 specifically:
S301 carries out gray proces to the enhanced Traffic Sign Images of lightness using weighted average and obtains gray level image；
S302 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
6. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as claimed in claim 4, special
Sign is that gray scale normalization pre-processes in the S30 specifically:
The enhanced Traffic Sign Images of lightness are carried out color space inverse transformation by S301, and HSV space is converted to rgb space；
S302 is weighted and averaged to obtain gray level image using following formula:
Gray (x, y)=0.299*R+0.587*G+0.114*B,
Wherein Gray (x, y) is pixel (x, y) gray value that weighted calculation obtains, and R, G, B are respectively the color of pixel (x, y)
Color primary color component；
S303 is normalized gray level image to obtain Normalized Grey Level image using bilinear interpolation.
7. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is that the S40 is specifically included:
S401 is divided into m block image to Normalized Grey Level image averaging, and the entropy weight of i-th block of image is determined by following formula:
Wherein, i is any integer value in (0, m), and n is number of greyscale levels,Indicate that kth grade pixel goes out
Existing probability, Ei(i=1,2 ..., m) be i-th block of image comentropy, Wi(i=1,2 ..., be m) be calculated i-th piece
The weight of image；
S402 is extracted histograms of oriented gradients HOG and image texture LBP feature to m block image respectively and is carried out using following formula
Weight fusion, obtains the HOG-LBP total characteristic of image:
8. a kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight as described in claim 1, special
Sign is that learning model includes linear SVM SVM, BP neural network or ELM extreme learning machine in the S50.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910673010.4A CN110490194A (en) | 2019-07-24 | 2019-07-24 | A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910673010.4A CN110490194A (en) | 2019-07-24 | 2019-07-24 | A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110490194A true CN110490194A (en) | 2019-11-22 |
Family
ID=68548199
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910673010.4A Pending CN110490194A (en) | 2019-07-24 | 2019-07-24 | A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110490194A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724349A (en) * | 2020-05-29 | 2020-09-29 | 同济大学 | Image smudge recognition method based on HSV and SVM |
| CN112070041A (en) * | 2020-09-14 | 2020-12-11 | 北京印刷学院 | Living body face detection method and device based on CNN deep learning model |
| CN112132081A (en) * | 2020-09-29 | 2020-12-25 | 广东工业大学 | Method, device and equipment for identifying engineering vehicle in aerial image |
| CN112232162A (en) * | 2020-10-06 | 2021-01-15 | 武汉烽火凯卓科技有限公司 | Pedestrian detection method and device based on multi-feature fusion cascade classifier |
| CN112598013A (en) * | 2020-12-30 | 2021-04-02 | 宁波职业技术学院 | Computer vision processing method based on neural network |
| CN112966782A (en) * | 2021-04-09 | 2021-06-15 | 深圳市豪恩汽车电子装备股份有限公司 | Multi-view-angle feature-fused road surface water detection and identification method |
| CN113610877A (en) * | 2021-07-08 | 2021-11-05 | 武汉工程大学 | Crop pest and disease identification method and system based on SVM multi-classification model |
| CN114529503A (en) * | 2021-12-17 | 2022-05-24 | 南京邮电大学 | Plant leaf identification method for improving self-adaptive weighting multi-feature fusion of Gabor and HOG |
| CN114723636A (en) * | 2022-04-25 | 2022-07-08 | 平安普惠企业管理有限公司 | Model generation method, device, equipment and storage medium based on multi-feature fusion |
| CN115311282A (en) * | 2022-10-12 | 2022-11-08 | 南通迪博西电子有限公司 | Wafer defect detection method based on image enhancement |
| CN115396728A (en) * | 2022-08-18 | 2022-11-25 | 维沃移动通信有限公司 | Method and device for determining video playing multiple speed, electronic equipment and medium |
| CN116958503A (en) * | 2023-09-19 | 2023-10-27 | 广东新泰隆环保集团有限公司 | Image processing-based sludge drying grade identification method and system |
| CN117291914A (en) * | 2023-11-24 | 2023-12-26 | 南昌江铃华翔汽车零部件有限公司 | Automobile part defect detection method, system, computer and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631405A (en) * | 2015-12-17 | 2016-06-01 | 谢寒 | Multistage blocking-based intelligent traffic video recognition background modeling method |
| CN106599870A (en) * | 2016-12-22 | 2017-04-26 | 山东大学 | Face recognition method based on adaptive weighting and local characteristic fusion |
| CN109086687A (en) * | 2018-07-13 | 2018-12-25 | 东北大学 | The traffic sign recognition method of HOG-MBLBP fusion feature based on PCA dimensionality reduction |
-
2019
- 2019-07-24 CN CN201910673010.4A patent/CN110490194A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105631405A (en) * | 2015-12-17 | 2016-06-01 | 谢寒 | Multistage blocking-based intelligent traffic video recognition background modeling method |
| CN106599870A (en) * | 2016-12-22 | 2017-04-26 | 山东大学 | Face recognition method based on adaptive weighting and local characteristic fusion |
| CN109086687A (en) * | 2018-07-13 | 2018-12-25 | 东北大学 | The traffic sign recognition method of HOG-MBLBP fusion feature based on PCA dimensionality reduction |
Non-Patent Citations (1)
| Title |
|---|
| 贾永红: "《数字图像处理》", 31 July 2015, 武汉大学出版社 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724349B (en) * | 2020-05-29 | 2022-09-20 | 同济大学 | Image smudge recognition method based on HSV and SVM |
| CN111724349A (en) * | 2020-05-29 | 2020-09-29 | 同济大学 | Image smudge recognition method based on HSV and SVM |
| CN112070041A (en) * | 2020-09-14 | 2020-12-11 | 北京印刷学院 | Living body face detection method and device based on CNN deep learning model |
| CN112070041B (en) * | 2020-09-14 | 2023-06-09 | 北京印刷学院 | Living body face detection method and device based on CNN deep learning model |
| CN112132081A (en) * | 2020-09-29 | 2020-12-25 | 广东工业大学 | Method, device and equipment for identifying engineering vehicle in aerial image |
| CN112232162A (en) * | 2020-10-06 | 2021-01-15 | 武汉烽火凯卓科技有限公司 | Pedestrian detection method and device based on multi-feature fusion cascade classifier |
| CN112232162B (en) * | 2020-10-06 | 2023-04-18 | 武汉烽火凯卓科技有限公司 | Pedestrian detection method and device based on multi-feature fusion cascade classifier |
| CN112598013A (en) * | 2020-12-30 | 2021-04-02 | 宁波职业技术学院 | Computer vision processing method based on neural network |
| CN112966782A (en) * | 2021-04-09 | 2021-06-15 | 深圳市豪恩汽车电子装备股份有限公司 | Multi-view-angle feature-fused road surface water detection and identification method |
| CN113610877A (en) * | 2021-07-08 | 2021-11-05 | 武汉工程大学 | Crop pest and disease identification method and system based on SVM multi-classification model |
| CN114529503A (en) * | 2021-12-17 | 2022-05-24 | 南京邮电大学 | Plant leaf identification method for improving self-adaptive weighting multi-feature fusion of Gabor and HOG |
| CN114723636A (en) * | 2022-04-25 | 2022-07-08 | 平安普惠企业管理有限公司 | Model generation method, device, equipment and storage medium based on multi-feature fusion |
| CN115396728A (en) * | 2022-08-18 | 2022-11-25 | 维沃移动通信有限公司 | Method and device for determining video playing multiple speed, electronic equipment and medium |
| CN115396728B (en) * | 2022-08-18 | 2024-08-27 | 维沃移动通信有限公司 | Method and device for determining video playing double speed, electronic equipment and medium |
| CN115311282A (en) * | 2022-10-12 | 2022-11-08 | 南通迪博西电子有限公司 | Wafer defect detection method based on image enhancement |
| CN116958503A (en) * | 2023-09-19 | 2023-10-27 | 广东新泰隆环保集团有限公司 | Image processing-based sludge drying grade identification method and system |
| CN116958503B (en) * | 2023-09-19 | 2024-03-12 | 广东新泰隆环保集团有限公司 | Image processing-based sludge drying grade identification method and system |
| CN117291914A (en) * | 2023-11-24 | 2023-12-26 | 南昌江铃华翔汽车零部件有限公司 | Automobile part defect detection method, system, computer and storage medium |
| CN117291914B (en) * | 2023-11-24 | 2024-02-09 | 南昌江铃华翔汽车零部件有限公司 | Automobile part defect detection method, system, computer and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110490194A (en) | A kind of recognition methods of the multiple features segment fusion traffic sign of adaptive weight | |
| Tabernik et al. | Deep learning for large-scale traffic-sign detection and recognition | |
| CN109657632B (en) | Lane line detection and identification method | |
| CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN107392968B (en) | The image significance detection method of Fusion of Color comparison diagram and Color-spatial distribution figure | |
| CN107066972B (en) | Natural scene Method for text detection based on multichannel extremal region | |
| CN103049763A (en) | Context-constraint-based target identification method | |
| CN104598924A (en) | Target matching detection method | |
| CN109740572A (en) | A kind of human face in-vivo detection method based on partial color textural characteristics | |
| CN110334703B (en) | Ship detection and identification method in day and night image | |
| CN112052845A (en) | Image recognition method, device, equipment and storage medium | |
| Alvarez et al. | Road geometry classification by adaptive shape models | |
| CN105844213B (en) | Green fruit recognition method | |
| CN103853724A (en) | Multimedia data sorting method and device | |
| CN104318266A (en) | Image intelligent analysis processing early warning method | |
| CN103544504A (en) | Scene character recognition method based on multi-scale map matching core | |
| CN112233173A (en) | Method for searching and positioning indoor articles of people with visual impairment | |
| CN103065126A (en) | Re-identification method of different scenes on human body images | |
| Vil’kin et al. | Algorithm for segmentation of documents based on texture features | |
| CN111274964A (en) | Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle | |
| CN109977882A (en) | A kind of half coupling dictionary is to the pedestrian of study again recognition methods and system | |
| CN108805139A (en) | A kind of image similarity computational methods based on frequency-domain visual significance analysis | |
| CN112800968B (en) | HOG blocking-based feature histogram fusion method for identifying identity of pigs in drinking area |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191122 |
|
| RJ01 | Rejection of invention patent application after publication |