CN108985374A - A kind of flame detecting method based on dynamic information model - Google Patents
A kind of flame detecting method based on dynamic information model Download PDFInfo
- Publication number
- CN108985374A CN108985374A CN201810765549.8A CN201810765549A CN108985374A CN 108985374 A CN108985374 A CN 108985374A CN 201810765549 A CN201810765549 A CN 201810765549A CN 108985374 A CN108985374 A CN 108985374A
- Authority
- CN
- China
- Prior art keywords
- flame
- image
- information model
- dynamic information
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000001514 detection method Methods 0.000 claims abstract description 10
- 238000013135 deep learning Methods 0.000 claims abstract description 7
- 230000006399 behavior Effects 0.000 claims abstract description 3
- 238000005516 engineering process Methods 0.000 claims description 5
- 230000000007 visual effect Effects 0.000 abstract description 5
- 230000003068 static effect Effects 0.000 abstract description 2
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Fire-Detection Mechanisms (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
In the common flame detecting method based on deep learning, model training mainly uses the static nature of target, therefore when existing in image with chaff interferent similar in target visual feature, can not effectively distinguish.Therefore, the invention proposes a kind of night flame detecting methods for combining deep learning with multidate information.That is, the multidate information of target is added during model training.The sample namely utilized is not from a frame image of ordinary video sequence, but includes the associated multidate information of before and after frames.Detection method entire protocol of the invention includes: to carry out binaryzation to the image of acquisition；The edge configuration of flame on image after binaryzation is sought, and saves image as training sample；Use region locating for minimum circumscribed rectangle mark flame fringe；The sample marked feeding caffe frame is trained, to obtain the model for detecting flame；Judge to monitor whether region occurs fire behavior using trained model.
Description
Technical field
The invention belongs to fire defector field, especially a kind of flame detecting method based on dynamic information model.
Background technique
With the development of deep learning technology, more and more people start to carry out target detection using this technology, such
Mode automatically extracts target signature using computer, to reach good classification and Detection effect.Utilize deep learning technology
Method to carry out flame identification mainly trains model using a large amount of flame sample image, then utilizes trained mould
Type carries out fire defector.Main problem existing for this method is that deep learning only learns to have arrived the static nature of target, therefore
When existing in image with chaff interferent similar in target visual feature, can not effectively distinguish.Such as the light in the case of night with
Flame visual signature on single-frame images is extremely similar, and under actual conditions, light and flame are moved what video sequence was shown
State feature is significantly different.
Summary of the invention
The technical solution adopted by the invention is as follows:
A kind of flame detecting method based on dynamic information model, this method are based on deep learning technology, the flame inspection
The model utilized in survey method is the dynamic information model of flame, and the sample used when training the dynamic information model is utilization
The form for the flame fringe that frame difference method is sought, complete training step include:
The image of step 1. pair acquisition carries out binaryzation；
Step 2. seeks the edge configuration of flame on image after binaryzation, and saves image as training sample；
Step 3. uses region locating for minimum circumscribed rectangle mark flame fringe；
The sample marked feeding caffe frame is trained by step 4., to obtain the model for detecting flame；
Further, the formula that binaryzation uses are as follows:
Wherein f (x, y) is the pixel value for the pixel that coordinate is (x, y).
Compared with prior art, the beneficial effects of the present invention are:
1. during model training be added target multidate information, solve when in image exist and target visual feature
When similar chaff interferent, the problem of can not effectively distinguishing.
Another object of the present invention is to design a kind of night flame detector system, which includes high-definition camera, fire
Flame detects terminal and alarm unit, and the fire defector terminal includes flame detection unit, and flame detection unit uses above-mentioned one
Flame detecting method of the kind based on dynamic information model, alarm unit are connected with buzzer and alarm lamp.
Further, the detecting step of flame detection unit includes:
The image of step 1. pair acquisition carries out binaryzation；
Step 2. seeks the edge configuration of highlight regions on image after binaryzation using frame difference method；
Step 3. is the confidence level of flame fringe form using trained model analysis edge configuration；
If step 4. confidence level is greater than 0.6, continue to analyze subsequent image, if continuous 10 times confidence level occur and are greater than
0.6 the case where, then determines that fire behavior occurs.
Beneficial effects of the present invention and a kind of above-mentioned beneficial effect phase of the flame detecting method based on dynamic information model
Together, details are not described herein.
Detailed description of the invention
Fig. 1 is flame detecting method flow chart；
Fig. 2 is the image of acquisition；
Fig. 3 is the image after binaryzation；
Fig. 4 is the flame fringe image extracted.
Specific embodiment
Flame detecting method proposed by the present invention, the sample utilized are not from a frame image of ordinary video sequence,
But include the associated multidate information of before and after frames, specifically, the sample of training is the shape for the flame fringe sought using frame difference method
State.Complete step includes:
The image of step 1. pair acquisition carries out binaryzation；
Step 2. seeks the edge configuration of flame on image after binaryzation, and saves image as training sample；
Step 3. uses region locating for minimum circumscribed rectangle mark flame fringe；
The sample marked feeding caffe frame is trained by step 4., to obtain the model for detecting flame.
Wherein, the formula that binaryzation uses are as follows:
Wherein f (x, y) is the pixel value for the pixel that coordinate is (x, y).
The multidate information of target is added in above-mentioned model training method during model training, solves when existing in image
When with chaff interferent similar in target visual feature, the problem of can not effectively distinguishing, a kind of night fire defector system is designed based on this
System, which includes high-definition camera, fire defector terminal and alarm unit, and wherein alarm unit is connected with buzzer and alarm
Lamp.High-definition camera is used to shoot the image of monitor area, and fire defector terminal receives the image of high-definition camera acquisition, goes forward side by side
Row fire defector, detecting step include: to carry out binaryzation to the image of acquisition；It is sought after binaryzation using frame difference method high on image
The edge configuration of bright area；It is the confidence level of flame fringe form using trained model analysis edge configuration；If confidence level
Greater than 0.6, then continue to analyze subsequent image, if occurring the case where confidence level is greater than 0.6 for continuous 10 times, determines that fire occurs
Feelings, the buzzer of alarm unit and alarm lamp are worked at the same time to warn the personnel for being in danger zone at this time.
The foregoing is merely the preferred embodiments of the invention, are not intended to limit the invention creation, all at this
Within the spirit and principle of innovation and creation, any modification, equivalent replacement, improvement and so on should be included in the invention
Protection scope within.
Claims (6)
1. a kind of flame detecting method based on dynamic information model, this method is based on deep learning technology, which is characterized in that institute
State the dynamic information model that the model utilized in flame detecting method is flame, the sample used when training the dynamic information model
This is the form for the flame fringe sought using frame difference method.
2. a kind of flame detecting method based on dynamic information model as described in claim 1, which is characterized in that training is described dynamic
The entire protocol of state information model includes:
The image of step 1. pair acquisition carries out binaryzation；
Step 2. seeks the edge configuration of flame on image after binaryzation, and saves image as training sample；
Step 3. uses region locating for minimum circumscribed rectangle mark flame fringe；
The sample marked feeding training aids is trained by step 4., to obtain the model for detecting flame.
3. a kind of flame detecting method based on dynamic information model as claimed in claim 2, which is characterized in that the training aids
For caffe frame.
4. a kind of flame detecting method based on dynamic information model as claimed in claim 2, which is characterized in that binaryzation uses
Formula are as follows:
Wherein f (x, y) is the pixel value for the pixel that coordinate is (x, y).
5. a kind of night flame detector system, which is characterized in that including high-definition camera, fire defector terminal and alarm unit,
The fire defector terminal includes flame detection unit, and flame detection unit uses a kind of above-mentioned fire based on dynamic information model
Flame detection method, alarm unit are connected with buzzer and alarm lamp.
6. a kind of night flame detector system as claimed in claim 5, which is characterized in that the detecting step packet of flame detection unit
It includes:
The image of step 1. pair acquisition carries out binaryzation；
Step 2. seeks the edge configuration of highlight regions on image after binaryzation using frame difference method；
Step 3. is the confidence level of flame fringe form using trained model analysis edge configuration；
If step 4. confidence level is greater than 0.6, continue to analyze subsequent image, if occurring confidence level for continuous 10 times is greater than 0.6
Situation then determines that fire behavior occurs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810765549.8A CN108985374A (en) | 2018-07-12 | 2018-07-12 | A kind of flame detecting method based on dynamic information model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810765549.8A CN108985374A (en) | 2018-07-12 | 2018-07-12 | A kind of flame detecting method based on dynamic information model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108985374A true CN108985374A (en) | 2018-12-11 |
Family
ID=64537868
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810765549.8A Pending CN108985374A (en) | 2018-07-12 | 2018-07-12 | A kind of flame detecting method based on dynamic information model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985374A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114549296A (en) * | 2022-04-21 | 2022-05-27 | 北京世纪好未来教育科技有限公司 | Training method of image processing model, image processing method and electronic equipment |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201232482A (en) * | 2011-01-27 | 2012-08-01 | Chang Jung Christian University | Machine vision flame identification system and method |
| CN103065124A (en) * | 2012-12-24 | 2013-04-24 | 成都国科海博计算机***有限公司 | Smoke detection method, device and fire detection device |
| CN103903020A (en) * | 2014-04-22 | 2014-07-02 | 天津市协力自动化工程有限公司 | Method and device for fire disaster image identification based on Code Books |
| CN105788142A (en) * | 2016-05-11 | 2016-07-20 | 中国计量大学 | Video image processing-based fire detection system and detection method |
| CN106203334A (en) * | 2016-07-11 | 2016-12-07 | 电子科技大学 | A kind of indoor scene flare detection method |
| CN106709521A (en) * | 2016-12-26 | 2017-05-24 | 深圳极视角科技有限公司 | Fire pre-warning method and fire pre-warning system based on convolution neural network and dynamic tracking |
| CN106845410A (en) * | 2017-01-22 | 2017-06-13 | 西安科技大学 | A kind of flame identification method based on deep learning model |
| CN107025652A (en) * | 2017-05-05 | 2017-08-08 | 太原理工大学 | A kind of flame detecting method based on kinetic characteristic and color space time information |
-
2018
- 2018-07-12 CN CN201810765549.8A patent/CN108985374A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201232482A (en) * | 2011-01-27 | 2012-08-01 | Chang Jung Christian University | Machine vision flame identification system and method |
| CN103065124A (en) * | 2012-12-24 | 2013-04-24 | 成都国科海博计算机***有限公司 | Smoke detection method, device and fire detection device |
| CN103903020A (en) * | 2014-04-22 | 2014-07-02 | 天津市协力自动化工程有限公司 | Method and device for fire disaster image identification based on Code Books |
| CN105788142A (en) * | 2016-05-11 | 2016-07-20 | 中国计量大学 | Video image processing-based fire detection system and detection method |
| CN106203334A (en) * | 2016-07-11 | 2016-12-07 | 电子科技大学 | A kind of indoor scene flare detection method |
| CN106709521A (en) * | 2016-12-26 | 2017-05-24 | 深圳极视角科技有限公司 | Fire pre-warning method and fire pre-warning system based on convolution neural network and dynamic tracking |
| CN106845410A (en) * | 2017-01-22 | 2017-06-13 | 西安科技大学 | A kind of flame identification method based on deep learning model |
| CN107025652A (en) * | 2017-05-05 | 2017-08-08 | 太原理工大学 | A kind of flame detecting method based on kinetic characteristic and color space time information |
Non-Patent Citations (2)
| Title |
|---|
| 刘鹏等: "基于视频的烟焰检测方法研究", vol. 16, no. 10, pages 14 - 17 * |
| 黄孝平: "当代机器深度学习方法与应用研究", 成都：电子科技大学出版社, pages: 117 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114549296A (en) * | 2022-04-21 | 2022-05-27 | 北京世纪好未来教育科技有限公司 | Training method of image processing model, image processing method and electronic equipment |
| CN114549296B (en) * | 2022-04-21 | 2022-07-12 | 北京世纪好未来教育科技有限公司 | Training method of image processing model, image processing method and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110044486B (en) | Method, device and equipment for avoiding repeated alarm of human body inspection and quarantine system | |
| CN106373320B (en) | Method for recognizing fire disaster based on flame color dispersion and sequential frame image similarity | |
| JP5569990B2 (en) | Attribute determination method, attribute determination apparatus, program, recording medium, and attribute determination system | |
| CN105426828B (en) | Method for detecting human face, apparatus and system | |
| CN101738394B (en) | Method and system for detecting indoor smog | |
| CN110837784A (en) | Examination room peeping cheating detection system based on human head characteristics | |
| CN102881106A (en) | Dual-detection forest fire identification system through thermal imaging video and identification method thereof | |
| US20190197313A1 (en) | Monitoring device | |
| CN104598895A (en) | Method and device for flame detection based on video image analysis | |
| CN104504369A (en) | Wearing condition detection method for safety helmets | |
| JP2016218760A5 (en) | ||
| CN105205437B (en) | Side face detection method and device based on contouring head verifying | |
| CN109635758A (en) | Wisdom building site detection method is dressed based on the high altitude operation personnel safety band of video | |
| CN101916372B (en) | Video-based smoke detection device and method according to multi-feature fusion | |
| CN101803928A (en) | Video-based driver fatigue detection device | |
| CN105930798A (en) | Tongue image quick detection and segmentation method based on learning and oriented to handset application | |
| CN107067412A (en) | A kind of video flame smog detection method of Multi-information acquisition | |
| CN109190475A (en) | A kind of recognition of face network and pedestrian identify network cooperating training method again | |
| CN109741565B (en) | Coal mine fire disaster recognition system and method | |
| CN108492517A (en) | A kind of fire monitoring system for building | |
| CN103336948A (en) | Video tracking method based on face recognition | |
| CN102609728B (en) | Method for detecting special pornographic image | |
| CN110136172A (en) | The detection method that safeguard is worn before a kind of miner goes into the well | |
| CN108898782B (en) | Smoke detection method and system for infrared image temperature information identification for tunnel fire prevention | |
| CN110929608A (en) | Safety helmet wearing detection method and system |
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: 20181211 |