CN104952082A - Rapid compressive tracking method based on classification-based three-step search strategy - Google Patents
Rapid compressive tracking method based on classification-based three-step search strategy Download PDFInfo
- Publication number
- CN104952082A CN104952082A CN201510250690.0A CN201510250690A CN104952082A CN 104952082 A CN104952082 A CN 104952082A CN 201510250690 A CN201510250690 A CN 201510250690A CN 104952082 A CN104952082 A CN 104952082A
- Authority
- CN
- China
- Prior art keywords
- target
- target detection
- tracking
- wave mixing
- tracking method
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/223—Analysis of motion using block-matching
- G06T7/238—Analysis of motion using block-matching using non-full search, e.g. three-step search
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Abstract
Disclosed is a rapid compressive tracking method based on a classification-based three-step search strategy. The rapid compressive tracking method is characterized by including: in a target detection phase, firstly, using a sliding window for target detection within a large search range with a large step length so as to obtain an initial target position; secondly, using the sliding window for target detection within a small search range centering on the initial target position with a small step length so as to obtain a current optimal position; finally, conducting target detection at the current optimal position with a minimum step length and obtaining a final target position by the compressive tracking method. The rapid compressive tracking method based on the classification-based three-step search strategy is used for further improving real-time performance of compressive tracking and achieving rapid tracking of a target in a video image and plays a guiding role in target tracking and intelligent monitoring for security and protection departments.
Description
Technical field
The present invention relates to computer vision field, specifically a kind of method of the target in video image being carried out to follow the tracks of fast.
Background technology
Performance along with computing machine improves constantly increase that is gradually cheap with camera shooting terminal and the automatic video frequency analysis market demand, object detecting and tracking in video image causes increasing concern, and all has application prospect very widely in fields such as intelligent monitoring, man-machine interaction, video frequency searching, medical treatment.
Target following is the important step of intelligent monitoring, can play directive function to the research of target following to intelligent video monitoring and national defense safety.
Summary of the invention
The object of the invention is for building a kind of method for tracking target with higher robustness and real-time.
The present invention is in the target detection stage, first in larger hunting zone, the initial position that target detection obtains target is carried out with larger step-length sliding window, then centered by this position more among a small circle in, carry out target detection again with less step-length sliding window, obtain current optimum position; Finally on the position that back obtains, do the minimum target detection of step-length, obtain final target location by compression track algorithm.
Concrete technical scheme is as follows:
(1) initialization tracking target, manually chooses target.Gather target and background sample, utilize feature extraction matrix to generate the feature of target and background sample;
(2) read a new two field picture, utilize class three-wave mixing strategy to obtain the target location of present frame;
(3) gather target and background sample in this frame, upgrade classifier parameters.
(4) new to next again frame processes equally, until process all frames.
The invention has the beneficial effects as follows:
1, the Fast Compression tracking of a kind three-wave mixing strategy is set up;
2, the present invention is simple, can robust, in real time the target in video image is followed the tracks of, have wide range of applications, protection and monitor field had to the value of reference and application, directive function is played to field of intelligent monitoring.
Accompanying drawing explanation
Fig. 1 is the tracking results of the present invention to car video sequence;
Fig. 2 is the error curve diagram that the present invention follows the tracks of car video sequence;
Fig. 3 is the real-time comparison sheet that the present invention and other two kinds of algorithms are followed the tracks of car video sequence.
Embodiment
Further illustrate flesh and blood of the present invention below in conjunction with accompanying drawing and example, but content of the present invention is not limited to this.
Embodiment 1:
Obtain david sequence of video images.Initialization tracking target, manually chooses target.Gather target and background sample, utilize feature extraction matrix to generate the feature of target and background sample, calculating parameter; Read a new two field picture, utilize class three-wave mixing strategy to obtain the target location of present frame; Gather target and background sample in this frame, upgrade classifier parameters; Next frame is processed, until process all frames, obtains in Fig. 1 the result that car is followed the tracks of.
Claims (4)
1. the Fast Compression tracking of a kind three-wave mixing strategy, for following the tracks of the target in video image, provides reference, for intelligent video monitoring to directive function can to security protection department.It is characterized in that: in the target detection stage, first in larger hunting zone, carry out target detection with larger step-length sliding window.After obtaining target location, centered by this position more among a small circle in, carry out target detection with less step-length sliding window, obtain current optimum position.Finally on the position that back obtains, do the minimum target detection of step-length, obtain final target location by compression track algorithm.
2. the Fast Compression tracking of a kind three-wave mixing strategy according to claim 1, is characterized in that: use compressive sensing theory, utilizes random measurement matrix to extract image sheet feature.
3. the Fast Compression tracking of a kind three-wave mixing strategy according to claim 1, is characterized in that: use target and foreground picture photo features training Naive Bayes Classifier.
4. the Fast Compression tracking of a kind three-wave mixing strategy according to claim 1, is characterized in that: the strategy localizing objects position quickly utilizing three-wave mixing in the target detection stage.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510250690.0A CN104952082A (en) | 2015-05-15 | 2015-05-15 | Rapid compressive tracking method based on classification-based three-step search strategy |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510250690.0A CN104952082A (en) | 2015-05-15 | 2015-05-15 | Rapid compressive tracking method based on classification-based three-step search strategy |
Publications (1)
Publication Number | Publication Date |
---|---|
CN104952082A true CN104952082A (en) | 2015-09-30 |
Family
ID=54166710
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510250690.0A Pending CN104952082A (en) | 2015-05-15 | 2015-05-15 | Rapid compressive tracking method based on classification-based three-step search strategy |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104952082A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105513095A (en) * | 2015-12-30 | 2016-04-20 | 山东大学 | Behavior video non-supervision time-sequence partitioning method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130022234A1 (en) * | 2011-07-22 | 2013-01-24 | Honeywell International Inc. | Object tracking |
CN104243916A (en) * | 2014-09-02 | 2014-12-24 | 江苏大学 | Moving object detecting and tracking method based on compressive sensing |
CN104299247A (en) * | 2014-10-15 | 2015-01-21 | 云南大学 | Video object tracking method based on self-adaptive measurement matrix |
-
2015
- 2015-05-15 CN CN201510250690.0A patent/CN104952082A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130022234A1 (en) * | 2011-07-22 | 2013-01-24 | Honeywell International Inc. | Object tracking |
CN104243916A (en) * | 2014-09-02 | 2014-12-24 | 江苏大学 | Moving object detecting and tracking method based on compressive sensing |
CN104299247A (en) * | 2014-10-15 | 2015-01-21 | 云南大学 | Video object tracking method based on self-adaptive measurement matrix |
Non-Patent Citations (2)
Title |
---|
宋劲松 等: "基于变步长增益调整的机动目标跟踪新算法", 《电光与控制》 * |
钟权 等: "一种改进的实时压缩跟踪算法", 《光电工程》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105513095A (en) * | 2015-12-30 | 2016-04-20 | 山东大学 | Behavior video non-supervision time-sequence partitioning method |
CN105513095B (en) * | 2015-12-30 | 2019-04-09 | 山东大学 | A kind of unsupervised timing dividing method of behavior video |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Singla | Motion detection based on frame difference method | |
CN110049206B (en) | Image processing method, image processing apparatus, and computer-readable storage medium | |
Eiselein et al. | Real-time multi-human tracking using a probability hypothesis density filter and multiple detectors | |
CN102314589B (en) | Fast human-eye positioning method and device | |
CN102982537B (en) | A kind of method and system detecting scene change | |
CN103646242B (en) | Extended target tracking based on maximum stable extremal region feature | |
CN104715494A (en) | Video target compressing and tracking method based on partial sensitive histogram | |
CN111191535B (en) | Pedestrian detection model construction method based on deep learning and pedestrian detection method | |
Lu et al. | Generating fluent tubes in video synopsis | |
CN105426928B (en) | A kind of pedestrian detection method based on Haar feature and EOH feature | |
Wang et al. | Violence video detection by discriminative slow feature analysis | |
CN104952082A (en) | Rapid compressive tracking method based on classification-based three-step search strategy | |
CN103400395A (en) | Light stream tracking method based on HAAR feature detection | |
NC et al. | HOG-PCA descriptor with optical flow based human detection and tracking | |
CN102881025A (en) | Method for detecting multiple moving targets | |
Li et al. | Loitering detection based on trajectory analysis | |
CN111091586A (en) | Rapid smoke dynamic shielding area detection and positioning method and application thereof | |
Dong et al. | Recognizing human interaction by multiple features | |
US20210312642A1 (en) | Continuous learning for object tracking | |
Yang et al. | Robust object tracking via online multiple instance metric learning | |
ZHANG et al. | A moving object detection method based on improved three⁃ frame difference algorithm and edge information | |
Wang et al. | Robust tracking method with drift correction | |
Chau et al. | Automatic parameter adaptation for multi-object tracking | |
Baek et al. | Cognition system of bolt hole using template matching | |
Gupta et al. | Detection and recognition of handguns in the surveillance videos using neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150930 |