CN105844664A - Monitoring video vehicle detection tracking method based on improved TLD - Google Patents
Monitoring video vehicle detection tracking method based on improved TLD Download PDFInfo
- Publication number
- CN105844664A CN105844664A CN201610159169.0A CN201610159169A CN105844664A CN 105844664 A CN105844664 A CN 105844664A CN 201610159169 A CN201610159169 A CN 201610159169A CN 105844664 A CN105844664 A CN 105844664A
- Authority
- CN
- China
- Prior art keywords
- cam shift
- random forest
- result
- target
- tracker
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06V2201/07—Target detection
-
- 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
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a monitoring video vehicle detection tracking method based on an improved TLD, which is high in accuracy and good in robustness. The monitoring video vehicle detection tracking method comprises steps of adopting a sub-block Cam Shift tracer based on vehicle color characteristics to replace an L-K light flow point tracer, realizing description of a tracked object through a vehicle area color histogram obtained by the Cam Shift, realizing activity amount estimation on the tracking object moving between a prior frame and a later frame through measuring the similarity of the color histograms in the capture area, further combining with a random forest detector to obtain a rough position of a vehicle object, and observing a detector and positioning the tracker in real time through P-N study so as to realize effective vehicle detection tracking.
Description
Technical field
The present invention relates to Intelligent traffic video process field, a kind of accuracy is high, robustness is good based on changing
Enter the monitor video vehicle detecting and tracking method of TLD.
Background technology
Following the tracks of frame by frame of target is carried out in the environment of supposing driftlessness disappearance or blocking completely by L-K optical flow method,
It is also referred to as short-term tracker.Such tracker generally lacks there is the direct process after following the tracks of mistake, is difficult to when long
Between target following in the effect that obtains.At present, the research of such short-term tracking is concentrated mainly on tracking by people
The raising of accuracy and speed, and extend tracking time aspect, but in the case of tracking accuracy is undesirable, but can not effectively keep away
Exempt from accumulation and the drift phenomenon of tracking error.In recent years, a kind of new single goal long-time track algorithm TLD is occurred in that
(Tracking Learning-Detection), track algorithm and detection algorithm are combined by this algorithm, thus overcome
Target deforms upon the problem with partial occlusion during following the tracks of；Meanwhile, this algorithm has introduced a kind of on-line study mechanism, will
The result that tracker and detector are obtained inputs to study module, is given by the model feedback after study again and follows the tracks of and detection mould
Block, so that the detection of target is more stablized effectively with following the tracks of.But, unanimously connect based on motion owing to have employed at tracking module
The L-K optical flow method that coherence is assumed, this algorithm is shorter to the cycle, interframe movement is limited, seen from naked eyes in the case of target continuously
The motion of interframe has and preferably predicts the outcome, and for those significantly, the target of rapid movement, its prediction and tracking performance
The most undesirable.
Summary of the invention
The present invention is to solve the above-mentioned technical problem existing for prior art, it is provided that a kind of accuracy high, robust
Property good based on the monitor video vehicle detecting and tracking method improving TLD.
The technical solution of the present invention is: a kind of monitor video vehicle detecting and tracking method based on improvement TLD, it is special
Levy and be to follow the steps below:
Step 1. inputs the 1st frame video image, and hand labeled goes out target to be tracked, order；
Step 2. initializes random forest grader and Cam Shift tracker；
Step 3. makes, it is loaded into theFrame video image, and utilize random forest detection of classifier target, utilize Cam
Shift tracker is followed the tracks of target and obtains the adjustment yardstick of target frame；
The tracking result of the testing result of random forest grader with Cam Shift tracker is blended by Step 4.；
Step 5. utilizes P-N learning strategy to update random forest grader, it is thus achieved that the position of target；
If Step 6. video has arrived at last frame, then algorithm terminates；Otherwise, Step 3 is proceeded to.
Described Step 4 is as follows:
If Step 4.1 random forest grader and Cam Shift tracker all bounding box are as output, but random forest
Grader has multiple analogous location to be determined out, and Cam Shift tracker only finds a target location, now with space
Degree of overlapping carries out cluster segmentation to some testing results；
If Step 4.2 Cam Shift tracker does not has bounding box to export, and the output of random forest grader bounding box,
So multiple testing results are split with the cluster of space overlap degree, now use first cluster segmentation result as melting
Close result；
If the bigger Clustering Decision-Making result of 4.3 1 correlations of Step occurs, but this result of decision and Cam Shift with
Track device result differ farther out, then use this result of decision as fusion results, at the beginning of then Cam Shift tracker being re-started
Beginningization also loses the sample set originally praised；
If Step 4.4 Cam Shift tracker bounding box exports, and random forest grader borderless easel output,
So use the result of Cam Shift tracker output as fusion results；
If Step 4.5 random forest grader and the output of Cam Shift tracker equal borderless easel, then it is assumed that target disappears
Lose.
The present invention uses piecemeal Cam Shift tracker based on vehicle color feature to substitute the some tracker of L-K light stream,
Realized the description following the tracks of target by the vehicle region color histogram acquired in Cam Shift, then by capture region
Color histogram similarity tolerance realize to follow the tracks of target front and back two interframe movement amounts estimate；Further combined with random forest
Detector obtains the rough position of vehicle target, and learns to be observed detector in real time and to tracker by P-N
Position, thus realize effective automobile detecting following.Compared with prior art, the present invention improves and follows the tracks of long-time
During the accuracy followed the tracks of of moving vehicle significantly, under Rapid Variable Design and robustness.
Accompanying drawing explanation
Fig. 1 is the flow chart of the embodiment of the present invention.
Fig. 2 is the detecting and tracking Comparative result figure under the supervision of the cities scene of angle of inclination.
Fig. 3 is the detecting and tracking Comparative result figure under the supervision of the cities video scene of high-altitude.
Fig. 4 is the detecting and tracking Comparative result figure under rainy day municipal highway monitor video scene.
Fig. 5 is the detecting and tracking Comparative result figure under oblique angle freeway surveillance and control video scene.
Detailed description of the invention
As shown in Figure 1: as follows based on the monitor video automobile detecting following algorithm steps improving TLD:
Step 1. inputs the 1st frame video image, and hand labeled goes out target to be tracked in the way of artificial, order；
Step 2. initializes random forest grader and Cam Shift tracker；
Step 3. makes, it is loaded into theFrame video image, utilizes random forest detection of classifier target, utilizes Cam
Target followed the tracks of by Shift tracker, thus obtains the adjustment yardstick of target frame；
The tracking result of the testing result of random forest grader with Cam Shift tracker is blended by Step 4.；
Step 5. utilizes P-N learning strategy to update random forest grader, it is thus achieved that the position of target；
If Step 6. video has arrived at last frame, then algorithm terminates；Otherwise, Step 3 is proceeded to.
Described Step 4 comprises the steps of:
If Step 4.1 random forest grader and Cam Shift tracker all bounding box are as output, but random forest
Grader has multiple analogous location to be determined out, and Cam Shift tracker only finds a target location, now with space
Degree of overlapping carries out cluster segmentation to some testing results；
If Step 4.2 Cam Shift tracker does not has bounding box to export, and the output of random forest grader bounding box,
So multiple testing results are split with the cluster of space overlap degree, now use first cluster segmentation result as melting
Close result；
If the bigger Clustering Decision-Making result of 4.3 1 correlations of Step occurs, but this result of decision and Cam Shift with
Track device result differ farther out, then use this result of decision as fusion results, at the beginning of then Cam Shift tracker being re-started
Beginningization also loses the sample set originally praised；
If Step 4.4 Cam Shift tracker bounding box exports, and random forest grader borderless easel output,
So use the result of Cam Shift tracker output as fusion results；
If Step 4.5 random forest grader and the output of Cam Shift tracker equal borderless easel, then it is assumed that target disappears
Lose.
Detecting and tracking Comparative result under the supervision of the cities scene of embodiment of the present invention angle of inclination is as shown in Figure 2.
Detecting and tracking Comparative result under the supervision of the cities video scene of embodiment of the present invention high-altitude is as shown in Figure 3.
Detecting and tracking Comparative result under embodiment of the present invention rainy day municipal highway monitor video scene is as shown in Figure 4.
Detecting and tracking Comparative result under embodiment of the present invention oblique angle freeway surveillance and control video scene is as shown in Figure 5.
The embodiment of the present invention program runtime under above-mentioned different scenes successively is to such as table 1:
Table 1
The embodiment of the present invention tracking quality under above-mentioned different scenes successively is to such as table 2:
Table 2
Claims (2)
1. a monitor video vehicle detecting and tracking method based on improvement TLD, it is characterised in that follow the steps below:
Step 1. inputs the 1st frame video image, and hand labeled goes out target to be tracked, order；
Step 2. initializes random forest grader and Cam Shift tracker；
Step 3. makes, it is loaded into theFrame video image, and utilize random forest detection of classifier target, utilize Cam
Shift tracker is followed the tracks of target and obtains the adjustment yardstick of target frame；
The tracking result of the testing result of random forest grader with Cam Shift tracker is blended by Step 4.；
Step 5. utilizes P-N learning strategy to update random forest grader, it is thus achieved that the position of target；
If Step 6. video has arrived at last frame, then algorithm terminates；Otherwise, Step 3 is proceeded to.
The most according to claim 1 based on the monitor video vehicle detecting and tracking method improving TLD, it is characterised in that described
Step 4 is as follows:
If Step 4.1 random forest grader and Cam Shift tracker all bounding box are as output, but random forest
Grader has multiple analogous location to be determined out, and Cam Shift tracker only finds a target location, now with space
Degree of overlapping carries out cluster segmentation to some testing results；
If Step 4.2 Cam Shift tracker does not has bounding box to export, and the output of random forest grader bounding box,
So multiple testing results are split with the cluster of space overlap degree, now use first cluster segmentation result as melting
Close result；
If the bigger Clustering Decision-Making result of 4.3 1 correlations of Step occurs, but this result of decision and Cam Shift with
Track device result differ farther out, then use this result of decision as fusion results, at the beginning of then Cam Shift tracker being re-started
Beginningization also loses the sample set originally praised；
If Step 4.4 Cam Shift tracker bounding box exports, and random forest grader borderless easel output,
So use the result of Cam Shift tracker output as fusion results；
If Step 4.5 random forest grader and the output of Cam Shift tracker equal borderless easel, then it is assumed that target disappears
Lose.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610159169.0A CN105844664B (en) | 2016-03-21 | 2016-03-21 | Based on the monitor video vehicle detecting and tracking method for improving TLD |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610159169.0A CN105844664B (en) | 2016-03-21 | 2016-03-21 | Based on the monitor video vehicle detecting and tracking method for improving TLD |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105844664A true CN105844664A (en) | 2016-08-10 |
| CN105844664B CN105844664B (en) | 2019-01-11 |
Family
ID=56588349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610159169.0A Expired - Fee Related CN105844664B (en) | 2016-03-21 | 2016-03-21 | Based on the monitor video vehicle detecting and tracking method for improving TLD |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105844664B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296708A (en) * | 2016-08-18 | 2017-01-04 | 宁波傲视智绘光电科技有限公司 | Car tracing method and apparatus |
| CN107403439A (en) * | 2017-06-06 | 2017-11-28 | 沈阳工业大学 | Predicting tracing method based on Cam shift |
| CN107909024A (en) * | 2017-11-13 | 2018-04-13 | 哈尔滨理工大学 | Vehicle tracking system, method and vehicle based on image recognition and infrared obstacle avoidance |
| CN108876809A (en) * | 2018-06-17 | 2018-11-23 | 天津理工大学 | A kind of TLD image tracking algorithm based on Kalman filtering |
| CN112766038A (en) * | 2020-12-22 | 2021-05-07 | 深圳金证引擎科技有限公司 | Vehicle tracking method based on image recognition |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102881024A (en) * | 2012-08-24 | 2013-01-16 | 南京航空航天大学 | Tracking-learning-detection (TLD)-based video object tracking method |
| WO2014189685A1 (en) * | 2013-05-23 | 2014-11-27 | Fastvdo Llc | Motion-assisted visual language for human computer interfaces |
| CN104331901A (en) * | 2014-11-26 | 2015-02-04 | 北京邮电大学 | TLD-based multi-view target tracking device and method |
| CN104574439A (en) * | 2014-12-25 | 2015-04-29 | 南京邮电大学 | Kalman filtering and TLD (tracking-learning-detection) algorithm integrated target tracking method |
-
2016
- 2016-03-21 CN CN201610159169.0A patent/CN105844664B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102881024A (en) * | 2012-08-24 | 2013-01-16 | 南京航空航天大学 | Tracking-learning-detection (TLD)-based video object tracking method |
| WO2014189685A1 (en) * | 2013-05-23 | 2014-11-27 | Fastvdo Llc | Motion-assisted visual language for human computer interfaces |
| CN104331901A (en) * | 2014-11-26 | 2015-02-04 | 北京邮电大学 | TLD-based multi-view target tracking device and method |
| CN104574439A (en) * | 2014-12-25 | 2015-04-29 | 南京邮电大学 | Kalman filtering and TLD (tracking-learning-detection) algorithm integrated target tracking method |
Non-Patent Citations (1)
| Title |
|---|
| ZDENEK KALAL ET AL.: "《Tracking-Learning-Detection》", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296708A (en) * | 2016-08-18 | 2017-01-04 | 宁波傲视智绘光电科技有限公司 | Car tracing method and apparatus |
| CN106296708B (en) * | 2016-08-18 | 2019-02-15 | 宁波傲视智绘光电科技有限公司 | Car tracing method and apparatus |
| CN107403439A (en) * | 2017-06-06 | 2017-11-28 | 沈阳工业大学 | Predicting tracing method based on Cam shift |
| CN107403439B (en) * | 2017-06-06 | 2020-07-24 | 沈阳工业大学 | Cam-shift-based prediction tracking method |
| CN107909024A (en) * | 2017-11-13 | 2018-04-13 | 哈尔滨理工大学 | Vehicle tracking system, method and vehicle based on image recognition and infrared obstacle avoidance |
| CN107909024B (en) * | 2017-11-13 | 2021-11-05 | 哈尔滨理工大学 | Vehicle tracking system and method based on image recognition and infrared obstacle avoidance and vehicle |
| CN108876809A (en) * | 2018-06-17 | 2018-11-23 | 天津理工大学 | A kind of TLD image tracking algorithm based on Kalman filtering |
| CN108876809B (en) * | 2018-06-17 | 2021-07-20 | 天津理工大学 | TLD image tracking algorithm based on Kalman filtering |
| CN112766038A (en) * | 2020-12-22 | 2021-05-07 | 深圳金证引擎科技有限公司 | Vehicle tracking method based on image recognition |
| CN112766038B (en) * | 2020-12-22 | 2021-12-17 | 深圳金证引擎科技有限公司 | Vehicle tracking method based on image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105844664B (en) | 2019-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105844664A (en) | Monitoring video vehicle detection tracking method based on improved TLD | |
| Fischer et al. | Event-based visual place recognition with ensembles of temporal windows | |
| CN106447680B (en) | The object detecting and tracking method that radar is merged with vision under dynamic background environment | |
| Sina et al. | Vehicle counting and speed measurement using headlight detection | |
| KR101716928B1 (en) | Image processing method for vehicle camera and image processing apparatus usnig the same | |
| CN104794731B (en) | Multi-target detection tracking for ball machine control strategy | |
| O'malley et al. | Vision-based detection and tracking of vehicles to the rear with perspective correction in low-light conditions | |
| CN101344965A (en) | Tracking system based on binocular camera shooting | |
| CN103646257B (en) | A kind of pedestrian detection and method of counting based on video monitoring image | |
| US9378556B2 (en) | Method for reducing false object detection in stop-and-go scenarios | |
| WO2011028379A2 (en) | Foreground object tracking | |
| EP2474163A2 (en) | Foreground object detection in a video surveillance system | |
| KR20010000107A (en) | System tracking and watching multi moving object | |
| WO2014135404A1 (en) | Method for tracking a target in an image sequence, taking the dynamics of the target into consideration | |
| EP2813973B1 (en) | Method and system for processing video image | |
| CN105160649A (en) | Multi-target tracking method and system based on kernel function unsupervised clustering | |
| Choi et al. | New preceding vehicle tracking algorithm based on optimal unbiased finite memory filter | |
| CN102497505A (en) | Multi-ball machine linkage target tracking method and system based on improved Meanshift algorithm | |
| Denman et al. | Multi-spectral fusion for surveillance systems | |
| CN106780567B (en) | Immune particle filter extension target tracking method fusing color histogram and gradient histogram | |
| CN106558069A (en) | A kind of method for tracking target and system based under video monitoring | |
| CN104658249A (en) | Method for rapidly detecting vehicle based on frame difference and light stream | |
| Yaghoobi Ershadi et al. | Vehicle tracking and counting system in dusty weather with vibrating camera conditions | |
| CN113763427A (en) | Multi-target tracking method based on coarse-fine shielding processing | |
| CN117036404A (en) | Monocular thermal imaging simultaneous positioning and mapping method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190111 Termination date: 20210321 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |