CN102262728A - Road traffic sign identification method - Google Patents
Road traffic sign identification method Download PDFInfo
- Publication number
- CN102262728A CN102262728A CN2011102127342A CN201110212734A CN102262728A CN 102262728 A CN102262728 A CN 102262728A CN 2011102127342 A CN2011102127342 A CN 2011102127342A CN 201110212734 A CN201110212734 A CN 201110212734A CN 102262728 A CN102262728 A CN 102262728A
- Authority
- CN
- China
- Prior art keywords
- image
- road signs
- neural network
- interior
- external wave
- 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.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a road traffic sign identification method which comprises the following steps: carrying out image signature generation treatment on the original traffic sign by using PCNN (Pulse Coupled Neural Network) technique, and comparing the degree of similarity of the image signature by searching an image signature database, wherein the image with high degree of similarity is identified as the road traffic sign. The invention can assist vehicle traveling and management in a better way, and solves the problems of great sample demands, long training time and the like by using PCNN when solving the problem of image recognition and classification in the traditional artificial neural network.
Description
Technical field
The present invention relates to intelligent transportation field, be specifically related to road signs method for quickly identifying based on artificial neural network-Pulse Coupled Neural Network time signature.
Background technology
Nineteen ninety, people such as Eckhorn provide in the research of phenomenon at the visual cortex synchronizing pulse to cat, have proposed the mammalian nervous meta-model---the Eckhorn model.This model is called the pulse generation parts of neurn simulation device (neuromime), a modulation coupling unit and a cynapse link by one and forms.In order to remedy the deficiency of Eckhorn model, make it be applicable to Flame Image Process more, people have carried out various improvement to the Eckhorn model, are summed up by people such as Johnson at last, the Pulse Coupled Neural Network notion of (Pulse Coupled Neural Network is called for short PCNN) has been proposed.The two big characteristics that Pulse Coupled Neural Network is different from the Eckhorn model are modulation coupling and pulse generting machanism.The link input is modulated in nonlinear mode and is presented input, generates the internal activity item, and the link strength parameter can play the effect of link input to inside neurons activity influence size of regulating.Pulse Coupled Neural Network has adopted a kind of like this coupling scheme of modulating coupling rather than addition coupling, and this neuron that has just prevented from not have primary input is merely at the effect down-firing that links input, and this is very important for Flame Image Process.Pulse generates and has adopted neurn simulation device mode, and principle of work such as same step function are realized the dynamic threshold increase and decrease simultaneously, thereby with general I-N-F (integrate and fire) pattern a great difference is arranged.
Pulse Coupled Neural Network output each time can be represented with a two-dimensional matrix.When Pulse Coupled Neural Network was used for Flame Image Process, each pixel in the image was with regard to each neuron in the map network, and in fact its output be exactly a width of cloth bianry image, had represented the neuron of each iteration igniting.Calculate the neuronic sum of each iteration igniting, it was lined up by the time, just constituted the time signature of Pulse Coupled Neural Network.Because this time signature often is used to handle problems such as image recognition, therefore is also referred to as image signatures.It can be used as a characteristics of image in the image characteristics extraction.For the simple image of non-complex background, time signature has periodically and about uniqueness and distortion, rotation, translation and the flexible unchangeability of image.Discover,, only need to observe the content that the very short part of its time signature just can be known entire image for such image.
Intelligent transportation system (ITS) be with advanced person's infotech, data communication transmission, electronic sensor technology, electron controls technology and Computer Processing technology etc. integrated effectively apply to whole traffic administration and set up a kind of on a large scale in, comprehensive that play a role, in real time, comprehensive traffic transportation management system accurately and efficiently.Road signs identification is as the important step of intelligent transportation, and the information of being responsible for gathering relevant road signs makes vehicle travel by traffic rules.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of efficient recognition methods of road signs accurately, make its better service vehicle travel and manage, problems such as it is big that the utilization Pulse Coupled Neural Network solves the sample demand of traditional artificial neural network in handling image recognition and classification problem, and the training time is long.
In order to solve the problems of the technologies described above, the present invention adopts following technical scheme: a kind of recognition methods of road signs comprises the steps:
(1), utilize the hole filling technique to generate outer background image and interior background image respectively at each road signs image;
(2) outer background image pre-ignition: outer background image is imported among the corresponding Pulse Coupled Neural Network PCNN_outer, through twice iteration, generates external wave;
(3) interior background image pre-ignition: interior background image is imported among the corresponding Pulse Coupled Neural Network PCNN_inner, through twice iteration, and ripple in generating;
(4) upgrading external wave is not in that part of interior ripple in the former external wave;
(5) with new external wave as incremental record in the results set R of expression road signs housing;
(6) if new external wave is not empty, among the then new external wave input PCNN_outer, generate external wave, change step (4) over to; Not so, change step (7) over to;
(7) housing of road signs has been extracted,, extracted the interior figure of road signs again in conjunction with former figure; Should import among the corresponding Pulse Coupled Neural Network PCNN_signature by interior figure, write down the neuronic number of all previous iteration point fire, generate image signatures;
(8) searching image signature database, movement images signature similarity, identification road signs.
Compared with prior art, the present invention has following beneficial effect: with the novel artificial neural network---and Pulse Coupled Neural Network (PCNN) is as main handling implement, removes and the distinctive image signatures of production burst coupled neural network is realized identification to road signs by the traffic sign frame.
A. handle image recognition and classification problem with respect to artificial neural network in the past, do not need very big sample and training for a long time, thereby saved time overhead;
B. Pulse Coupled Neural Network is highly susceptible to hardware and realizes, its following of hardware supported obtained parallel processing speed be additive method can not compare.
C. Pulse Coupled Neural Network generates image time signature, and two-dimensional image information is converted into the one dimension time serial message, has reduced information processing amount in the identifying dramatically, has improved processing speed and efficient;
D. lines are simple, the background characteristic of simple because the road signs image has, and the time signature of its generation has the uniqueness about image.Therefore, the similarity of image time signature has just been represented the similarity degree between image.The time signature is approaching more, and the key diagram picture reaches unanimity more, otherwise then the image difference is big more.
E. the image time signature of Pulse Coupled Neural Network generation has distortion, rotation, translation and the flexible unchangeability about image, and therefore the deviation to original image has certain tolerance.
Description of drawings
Fig. 1 is identification road signs process flow diagram;
Fig. 2 is for generating the PCNN image signatures flowchart of road signs;
Fig. 3 is outer background image pre-ignition process flow diagram;
Fig. 4 is interior background image pre-ignition process flow diagram;
Fig. 5 is that figure generates PCNN image signatures process flow diagram in the road signs.
Embodiment
The invention will be further described below in conjunction with the drawings and the specific embodiments.
Referring to Fig. 1, a kind of recognition methods of road signs, initial to former figure (traffic indication map),
Handle the generation image signatures, by the searching image signature database, utilize formula again:
The similarity of movement images signature, wherein, α is an amount of token image similarity, and the value of α is big more, and the expression similarity is high more, and the images match of similarity maximum identifies road signs, at last end in last and the database.
As shown in Figure 2, former figure is handled to generate in the image signatures process comprise the steps:
(1), utilize the hole filling technique to generate outer background image and interior background image respectively at each road signs image;
(2) outer background image pre-ignition: as shown in Figure 3, outer background image is imported among the corresponding Pulse Coupled Neural Network PCNN_outer, through twice iteration, generates external wave;
(3) interior background image pre-ignition: as shown in Figure 4, interior background image is imported among the corresponding Pulse Coupled Neural Network PCNN_inner, through twice iteration, and ripple in generating;
(4) upgrading external wave is not in that part of interior ripple in the former external wave;
(5) with new external wave as incremental record in the results set R of expression road signs housing;
(6) if new external wave is not empty, among the then new external wave input PCNN_outer, generate external wave, change step (4) over to; Not so, change step (7) over to;
(7) as shown in Figure 5, the housing of road signs has been extracted,, extracted the interior figure of road signs again in conjunction with former figure; Should import among the corresponding Pulse Coupled Neural Network PCNN_signature by interior figure, write down the neuronic number of all previous iteration point fire, generate image signatures.
Claims (1)
1. the recognition methods of road signs is characterized in that, comprises the steps:
(1), utilize the hole filling technique to generate outer background image and interior background image respectively at each road signs image;
(2) outer background image pre-ignition: outer background image is imported among the corresponding Pulse Coupled Neural Network PCNN_outer, through twice iteration, generates external wave;
(3) interior background image pre-ignition: interior background image is imported among the corresponding Pulse Coupled Neural Network PCNN_inner, through twice iteration, and ripple in generating;
(4) upgrading external wave is not in that part of interior ripple in the former external wave;
(5) with new external wave as incremental record in the results set R of expression road signs housing;
(6) if new external wave is not empty, among the then new external wave input PCNN_outer, generate external wave, change step (4) over to; Not so, change step (7) over to;
(7) housing of road signs has been extracted,, extracted the interior figure of road signs again in conjunction with former figure; Should import among the corresponding Pulse Coupled Neural Network PCNN_signature by interior figure, write down the neuronic number of all previous iteration point fire, generate image signatures;
(8) searching image signature database, movement images signature similarity, identification road signs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110212734 CN102262728B (en) | 2011-07-28 | 2011-07-28 | Road traffic sign identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110212734 CN102262728B (en) | 2011-07-28 | 2011-07-28 | Road traffic sign identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102262728A true CN102262728A (en) | 2011-11-30 |
| CN102262728B CN102262728B (en) | 2012-12-19 |
Family
ID=45009349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110212734 Expired - Fee Related CN102262728B (en) | 2011-07-28 | 2011-07-28 | Road traffic sign identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102262728B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104537393A (en) * | 2015-01-04 | 2015-04-22 | 大连理工大学 | Traffic sign recognizing method based on multi-resolution convolution neural networks |
| CN105930791A (en) * | 2016-04-19 | 2016-09-07 | 重庆邮电大学 | Road traffic sign identification method with multiple-camera integration based on DS evidence theory |
| CN106529609A (en) * | 2016-12-08 | 2017-03-22 | 郑州云海信息技术有限公司 | Image recognition method and device based on neural network structure |
| CN108510472A (en) * | 2018-03-08 | 2018-09-07 | 北京百度网讯科技有限公司 | Method and apparatus for handling image |
| CN110458136A (en) * | 2019-08-19 | 2019-11-15 | 广东工业大学 | A kind of traffic sign recognition method, device and equipment |
| US11270132B2 (en) * | 2018-10-26 | 2022-03-08 | Cartica Ai Ltd | Vehicle to vehicle communication and signatures |
-
2011
- 2011-07-28 CN CN 201110212734 patent/CN102262728B/en not_active Expired - Fee Related
Non-Patent Citations (2)
| Title |
|---|
| 李建锋等: "基于交叉视觉皮层局部时间序列的图像判别方法", 《计算机应用》 * |
| 王坤明等: "基于神经网络的交通标志识别方法", 《抚顺石油学院学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104537393A (en) * | 2015-01-04 | 2015-04-22 | 大连理工大学 | Traffic sign recognizing method based on multi-resolution convolution neural networks |
| CN104537393B (en) * | 2015-01-04 | 2018-01-16 | 大连理工大学 | A kind of traffic sign recognition method based on multiresolution convolutional neural networks |
| CN105930791A (en) * | 2016-04-19 | 2016-09-07 | 重庆邮电大学 | Road traffic sign identification method with multiple-camera integration based on DS evidence theory |
| CN106529609A (en) * | 2016-12-08 | 2017-03-22 | 郑州云海信息技术有限公司 | Image recognition method and device based on neural network structure |
| CN108510472A (en) * | 2018-03-08 | 2018-09-07 | 北京百度网讯科技有限公司 | Method and apparatus for handling image |
| CN108510472B (en) * | 2018-03-08 | 2019-10-22 | 北京百度网讯科技有限公司 | Method and apparatus for handling image |
| US11270132B2 (en) * | 2018-10-26 | 2022-03-08 | Cartica Ai Ltd | Vehicle to vehicle communication and signatures |
| CN110458136A (en) * | 2019-08-19 | 2019-11-15 | 广东工业大学 | A kind of traffic sign recognition method, device and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102262728B (en) | 2012-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102262728B (en) | Road traffic sign identification method | |
| CN109145759B (en) | Vehicle attribute identification method, device, server and storage medium | |
| Bi et al. | Improved VGG model-based efficient traffic sign recognition for safe driving in 5G scenarios | |
| WO2023015922A1 (en) | Image recognition model training method and apparatus, device, and storage medium | |
| CN111275107A (en) | Multi-label scene image classification method and device based on transfer learning | |
| CN103955707B (en) | A kind of large nuber of images categorizing system based on depth level feature learning | |
| CN108376267A (en) | A kind of zero sample classification method based on classification transfer | |
| CN104809187A (en) | Indoor scene semantic annotation method based on RGB-D data | |
| CN104517122A (en) | Image target recognition method based on optimized convolution architecture | |
| CN105956626A (en) | Deep learning based vehicle license plate position insensitive vehicle license plate recognition method | |
| CN109754068A (en) | Transfer learning method and terminal device based on deep learning pre-training model | |
| CN107145559A (en) | Intelligent classroom Knowledge Management Platform and method based on semantic technology and gameization | |
| CN111783521B (en) | Pedestrian re-identification method based on low-rank prior guidance and based on domain invariant information separation | |
| CN107203752A (en) | A kind of combined depth study and the face identification method of the norm constraint of feature two | |
| CN106709528A (en) | Method and device of vehicle reidentification based on multiple objective function deep learning | |
| CN107944410A (en) | A kind of cross-cutting facial characteristics analytic method based on convolutional neural networks | |
| Yang et al. | Associative memory optimized method on deep neural networks for image classification | |
| CN110222760A (en) | A kind of fast image processing method based on winograd algorithm | |
| CN115907001B (en) | Knowledge distillation-based federal graph learning method and automatic driving method | |
| CN108960412A (en) | Image-recognizing method, device and computer readable storage medium | |
| CN108333959A (en) | A kind of energy saving method of operating of locomotive based on convolutional neural networks model | |
| CN108154156A (en) | Image Ensemble classifier method and device based on neural topic model | |
| CN110163052A (en) | Video actions recognition methods, device and machinery equipment | |
| CN103336830B (en) | Image search method based on structure semantic histogram | |
| CN110334572A (en) | The fine recognition methods of vehicle under a kind of multi-angle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121219 Termination date: 20180728 |