CN103700112A - Sheltered target tracking method based on mixed predicting strategy - Google Patents
Sheltered target tracking method based on mixed predicting strategy Download PDFInfo
- Publication number
- CN103700112A CN103700112A CN201210364748.0A CN201210364748A CN103700112A CN 103700112 A CN103700112 A CN 103700112A CN 201210364748 A CN201210364748 A CN 201210364748A CN 103700112 A CN103700112 A CN 103700112A
- Authority
- CN
- China
- Prior art keywords
- target
- frame
- present frame
- histogram
- prediction mode
- 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
Images
Landscapes
- Image Analysis (AREA)
- Studio Devices (AREA)
Abstract
The invention relates to a sheltered target tracking method based on a mixed predicting strategy and belongs to the technical field of photoelectric product application. In order to avoid a target sheltering phenomenon in a video scene, the sheltered target tracking method comprises the following steps: processing a current frame in a threshold segmenting way, allocating histogram templates of the current frame and a previous frame in a certain proportion, and updating a histogram template required by a next frame; predicting a target in a 3-point linear predicting way and a 5-point square predicting way respectively, and mixing a weighted array to the prediction results and generating a weighting coefficient; according to the prediction results obtained by the 3-point linear predicting way and the 5-point square predicting way and the weighting coefficient, obtaining a predicting position of the target at the next frame. Tests on visible light sequence images and infrared sequence images in a realistic environment prove that the sheltered target tracking method has good interference resistance and robustness and can realize stable tracking under the circumstance that the target is sheltered.
Description
Technical field
The present invention relates to photovoltaic applied technical field, be specifically related to a kind of shelter target tracking based on hybrid predicting strategy.
Background technology
Motion target tracking based on video or image sequence is all an extremely important and active research topic in computer vision, image processing and area of pattern recognition for a long time.Motion target tracking technology is an important step that is connected moving object detection and goal behavior analysis and understanding.In real application systems, target following not only can provide the movement locus, kinematic parameter of target and position accurately, also can provide authentic data source for motion analysis and the scene analysis of target in scene, simultaneously the trace information of moving target has been conversely for the correct detection of moving target and the identification of moving target provide decision support, thereby is more conducive to the tracking of moving target.
Blocking is the common situations in target following, and target may be blocked by object static in background, also may be blocked by other moving target, or because some own information is blocked in the rotation of self, and the degree of blocking is also different.Current most video monitoring system all can not be processed more serious occlusion issue, can not provide standard to judge and when stop and when restarting following the tracks of, and in track rejection situation, there is no again to obtain accordingly the bootstrap technique of target.
Summary of the invention
(1) technical matters that will solve
The technical problem to be solved in the present invention is how for the target occlusion phenomenon in video scene, design a kind of shelter target tracking based on hybrid predicting strategy, make when tracking target is blocked, by hybrid predicting strategy pursuit movement target stably.
(2) technical scheme
In order to solve the problems of the technologies described above, the invention provides a kind of shelter target tracking based on hybrid predicting strategy, the method comprises the steps:
Step S1: gather current video sequence image under capture apparatus stationary state;
Step S2: adopt Threshold segmentation mode to process present frame, according to the result of present frame, in conjunction with current residing concrete scene, with certain proportion, distribute the histogram template of present frame and previous frame, upgrade the required histogram template of next frame;
Step S3: adopt 3 point Linear prediction mode and 5 squares of prediction mode to predict respectively target;
Step S4: the result obtaining according to described 3 point Linear prediction mode and the prediction of 5 squares of prediction mode is carried out to mixed weighting combination, generate weighting coefficient;
Step S5: the result obtaining according to described 3 point Linear prediction mode predictions, 5 squares of result and described weighting coefficients that prediction mode prediction obtains, obtain target in the predicted position of next frame.
Wherein, described step S2 comprises:
Step S201: choose the prime area of tracking target in present frame, calculate current frame image selection area at the hue histogram in HSV space;
Step S202: the back projection figure that calculates selection area according to described hue histogram;
Step S203: utilize large Tianjin method to carry out Threshold segmentation to selection area and obtain bianry image;
Step S204: the bianry image obtaining with Threshold segmentation and back projection figure carry out logical and processing, obtains revised back projection figure;
Step S205: utilize back projection figure to adopt mean shift algorithm to carry out iterative processing to tracking target and obtain new target location;
Step S206: the histogram p that calculates new target location region
t;
Step S207: calculate and upgrade the required histogram template p of next frame according to formula (1)
n, p
tthe histogram template that represents present frame, p
rbe expressed as the histogram template of previous frame; β represents scale factor, and its value presets in 0 ~ 1 according to current residing concrete scene;
p
N＝β×p
T+(1-β)×p
R （1）。
Wherein, described step S3 comprises:
Step S301: the back projection figure that chooses new histogram formwork calculation target area;
Step S302: adopt mean shift algorithm to carry out iterative processing to tracking target according to back projection figure and obtain new target location;
Step S303: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2) and calculate 3 point Linear predicted value Y of k+1 frame in conjunction with formula (2)
l(k+1):
Y
L(k+1)＝[4Y(k)+Y(k-1)-2Y(k-2)]/3 （2）；
Wherein, present frame k represents the residing frame of current time, supposes Y
l(k) for pressing 3 predictions of point Linear prediction mode to present frame target location in formula (2), Y
l(k+1) be 3 predictions of point Linear prediction mode to next frame target location, Y (k) is that target is at the actual position of present frame;
Step S304: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2), Y (k-3), Y (k-4) and calculate 5 squares of predicted value Y of k+1 frame in conjunction with formula (3)
s(k+1):
Y
S(k+1)＝[9Y(k)-4Y(k-2)-3Y(k-3)+3Y(k-4)]/5 （3）；
Wherein, present frame k represents the residing frame of current time, supposes Y
s(k) for pressing the prediction to present frame target location of 5 squares of prediction mode in formula (3), Y
s(k+1) be the predictions of 5 squares of predictions to next frame target location, Y (k) is that target is at the actual position of present frame.
Wherein, in described step S4, according to 3 point Linear predicted value Y
l(k+1) and 5 squares of predicted value Y
s(k+1), by formula (4), calculate the value of weighting coefficient α:
Wherein, in described step S5, according to 3 point Linear predicted value Y
l(k+1), 5 squares of predicted value Y
s(k+1) and the value of weighting coefficient α, by formula (5), calculate target at the predicted position Y'(k+1 of next frame):
Y'(k+1)＝αY
L(k+1)+(1-α)Y
S(k+1) （5）。
(3) beneficial effect
The shelter target tracking based on hybrid predicting strategy of technical solution of the present invention upgrades and least square hybrid predicting strategy by histogram on the basis of mean shift algorithm, can eliminate the impact of partial occlusion on motion target tracking.By the visible light sequential image under actual environment and infrared sequence image, test, proved that the method has good anti-interference and robustness, can realize the tenacious tracking being blocked in situation in target.
Accompanying drawing explanation
The process flow diagram of the shelter target tracking that Fig. 1 provides for technical solution of the present invention.
Fig. 2 for carrying out the comparison schematic diagram of motion target tracking in certain open deck to visible images according to technical solution of the present invention.
Fig. 3 for carrying out the comparison schematic diagram of motion target tracking in certain open deck to infrared sequence image according to technical solution of the present invention.
Embodiment
For making object of the present invention, content and advantage clearer, below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.
The present invention is the improvement of the target tracking algorism based on average drifting, target tracking algorism based on average drifting adopts color characteristic conventionally, with Pasteur (Bhattacharyya) coefficient, weigh the similarity between object module and histogram corresponding to candidate target region, tracking problem is converted into the pattern matching problem based on average drifting.But when processing infrared image, owing to lacking colouring information, tend to cause mean shift algorithm to occur Divergent Phenomenon when search best match position, and cause the loss of tracking target.The present invention introduces the Threshold segmentation result of present frame in method design process, colouring information while describing infrared image to make up by a changeless histogram template is lost problem, according to the testing result of present frame, with certain proportion, upgrade the required histogram template of next frame.
Particularly, concrete application of the present invention is for magnanimity video frequency searching software systems, the enforcement of the method can adopt C Plus Plus to programme to carry out under VC6.0 platform, in order to realize the tenacious tracking of target under occlusion state, as shown in Figure 1, shelter target tracking based on hybrid predicting strategy provided by the present invention, comprises the steps:
Step S1: gather current video sequence image under capture apparatus stationary state;
Step S2: adopt Threshold segmentation mode to process present frame, according to the result of present frame, in conjunction with current residing concrete scene, with certain proportion, distribute the histogram template of present frame and previous frame, upgrade the required histogram template of next frame;
Step S3: adopt 3 point Linear prediction mode and 5 squares of prediction mode to predict respectively target;
Step S4: the result obtaining according to described 3 point Linear prediction mode and the prediction of 5 squares of prediction mode is carried out to mixed weighting combination, generate weighting coefficient;
Step S5: the result obtaining according to described 3 point Linear prediction mode predictions, 5 squares of result and described weighting coefficients that prediction mode prediction obtains, obtain target in the predicted position of next frame.
Wherein, described step S2 comprises:
Step S201: choose the prime area of tracking target in present frame, calculate current frame image selection area at HSV(hue, saturation, value, hue, saturation, intensity) tone (Hue) histogram in space;
Step S202: the back projection figure that calculates selection area according to described hue histogram;
Step S203: utilize large Tianjin method to carry out Threshold segmentation to selection area and obtain bianry image;
Step S204: the bianry image obtaining with Threshold segmentation and back projection figure carry out logical and processing, obtains revised back projection figure;
Step S205: utilize back projection figure to adopt mean shift algorithm to carry out iterative processing to tracking target and obtain new target location;
Step S206: the histogram p that calculates new target location region
t;
Step S207: calculate and upgrade the required histogram template p of next frame according to formula (1)
n, p
tthe histogram template that represents present frame, p
rbe expressed as the histogram template of previous frame; β represents scale factor, its value presets in 0 ~ 1 according to current residing concrete scene, such as, when the residing moment of present frame, target is blocked completely, the histogram template of previous frame occupied large percentage in the histogram template of next frame, now the value of β is set as less; Otherwise if the residing moment of previous frame, target is blocked completely, the histogram template of present frame occupied large percentage in the histogram template of next frame, now the value of β is set as larger;
p
N＝β×p
T+(1-β)×p
R （1）。
Wherein, when blocking appears in moving target, due to the loss of target information, if directly adopt mean shift algorithm can cause the loss of target, this method is weighted 3 point Linears predictions and 5 squares of predictions to combine carries out hybrid predicting to target.
Particularly, the in the situation that of least square principle, described step S3 comprises:
Step S301: the back projection figure that chooses new histogram formwork calculation target area;
Step S302: adopt mean shift algorithm to carry out iterative processing to tracking target according to back projection figure and obtain new target location;
Step S303: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2) and calculate 3 point Linear predicted value Y of k+1 frame in conjunction with formula (2)
l(k+1):
Y
L(k+1)＝[4Y(k)+Y(k-1)-2Y(k-2)]/3 （2）；
Wherein, present frame k represents the residing frame of current time, supposes Y
l(k) for pressing 3 predictions of point Linear prediction mode to present frame target location in formula (2), Y
l(k+1) be 3 predictions of point Linear prediction mode to next frame target location, Y (k) is that target is at the actual position of present frame;
Step S304: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2), Y (k-3), Y (k-4) and calculate 5 squares of predicted value Y of k+1 frame in conjunction with formula (3)
s(k+1):
Y
S(k+1)＝[9Y(k)-4Y(k-2)-3Y(k-3)+3Y(k-4)]/5 （3）；
Wherein, present frame k represents the residing frame of current time, supposes Y
s(k) for pressing the prediction to present frame target location of 5 squares of prediction mode in formula (3), Y
s(k+1) be the predictions of 5 squares of predictions to next frame target location, Y (k) is that target is at the actual position of present frame.
Next, in described step S4, according to 3 point Linear predicted value Y
l(k+1) and 5 squares of predicted value Y
s(k+1), by formula (4), calculate the value of weighting coefficient α:
Finally, in described step S5, according to 3 point Linear predicted value Y
l(k+1), 5 squares of predicted value Y
s(k+1) and the value of weighting coefficient α, by formula (5), calculate target at the predicted position Y'(k+1 of next frame):
Y'(k+1)＝αY
L(k+1)+(1-α)Y
S(k+1) （5）。
Below, in conjunction with concrete scene, test to verify the technique effect of technical solution of the present invention.
As shown in Figure 2, for visible images being carried out the example of motion target tracking in certain open deck, left figure is the tracking mode before original series image blocks; Middle figure is the tracking mode that occurs partial occlusion; Right figure is the tracking mode after target occlusion finishes.Through Histogram Matching and the hybrid predicting tracking strategy of average drifting, can find out the tracking under can realize target circumstance of occlusion.
As shown in Figure 3, infrared sequence image is carried out the example of motion target tracking in certain open deck, left figure is the tracking mode before original series image blocks; Middle figure is that the tracking mode of intersecting under blocking appears in two moving targets; Right figure is the tracking mode after target occlusion finishes.Because infrared sequence image lacks the support of colouring information, therefore occur intersecting while blocking when target, traditional mean shift algorithm can cause the loss of tracking target, and improvement project of the present invention can realize the tenacious tracking of intersection under blocking.
The above is only the preferred embodiment of the present invention; it should be pointed out that for those skilled in the art, do not departing under the prerequisite of the technology of the present invention principle; can also make some improvement and distortion, these improvement and distortion also should be considered as protection scope of the present invention.
Claims (5)
1. the shelter target tracking based on hybrid predicting strategy, is characterized in that, the method comprises the steps:
Step S1: gather current video sequence image under capture apparatus stationary state;
Step S2: adopt Threshold segmentation mode to process present frame, according to the result of present frame, in conjunction with current residing concrete scene, with certain proportion, distribute the histogram template of present frame and previous frame, upgrade the required histogram template of next frame;
Step S3: adopt 3 point Linear prediction mode and 5 squares of prediction mode to predict respectively target;
Step S4: the result obtaining according to described 3 point Linear prediction mode and the prediction of 5 squares of prediction mode is carried out to mixed weighting combination, generate weighting coefficient;
Step S5: the result obtaining according to described 3 point Linear prediction mode predictions, 5 squares of result and described weighting coefficients that prediction mode prediction obtains, obtain target in the predicted position of next frame.
2. the shelter target tracking based on hybrid predicting strategy as claimed in claim 1, is characterized in that, described step S2 comprises:
Step S201: choose the prime area of tracking target in present frame, calculate current frame image selection area at the hue histogram in HSV space;
Step S202: the back projection figure that calculates selection area according to described hue histogram;
Step S203: utilize large Tianjin method to carry out Threshold segmentation to selection area and obtain bianry image;
Step S204: the bianry image obtaining with Threshold segmentation and back projection figure carry out logical and processing, obtains revised back projection figure;
Step S205: utilize back projection figure to adopt mean shift algorithm to carry out iterative processing to tracking target and obtain new target location;
Step S206: the histogram p that calculates new target location region
t;
Step S207: calculate and upgrade the required histogram template p of next frame according to formula (1)
n, p
tthe histogram template that represents present frame, p
rbe expressed as the histogram template of previous frame; β represents scale factor, and its value presets in 0 ~ 1 according to current residing concrete scene;
p
N＝β×p
T+(1-β)×p
R （1）。
3. the shelter target tracking based on hybrid predicting strategy as claimed in claim 1, is characterized in that, described step S3 comprises:
Step S301: the back projection figure that chooses new histogram formwork calculation target area;
Step S302: adopt mean shift algorithm to carry out iterative processing to tracking target according to back projection figure and obtain new target location;
Step S303: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2) and calculate 3 point Linear predicted value Y of k+1 frame in conjunction with formula (2)
l(k+1):
Y
L(k+1)＝[4Y(k)+Y(k-1)-2Y(k-2)]/3 （2）；
Wherein, present frame k represents the residing frame of current time, supposes Y
l(k) for pressing 3 predictions of point Linear prediction mode to present frame target location in formula (2), Y
l(k+1) be 3 predictions of point Linear prediction mode to next frame target location, Y (k) is that target is at the actual position of present frame;
Step S304: for present frame k, according to the source location Y (k) obtaining, Y (k-1), Y (k-2), Y (k-3), Y (k-4) and calculate 5 squares of predicted value Y of k+1 frame in conjunction with formula (3)
s(k+1):
Y
S(k+1)＝[9Y(k)-4Y(k-2)-3Y(k-3)+3Y(k-4)]/5 （3）；
Wherein, present frame k represents the residing frame of current time, supposes Y
s(k) for pressing the prediction to present frame target location of 5 squares of prediction mode in formula (3), Y
s(k+1) be the predictions of 5 squares of predictions to next frame target location, Y (k) is that target is at the actual position of present frame.
4. the shelter target tracking based on hybrid predicting strategy as claimed in claim 1, is characterized in that, in described step S4, according to 3 point Linear predicted value Y
l(k+1) and 5 squares of predicted value Y
s(k+1), by formula (4), calculate the value of weighting coefficient α:
5. the shelter target tracking based on hybrid predicting strategy as claimed in claim 1, is characterized in that, in described step S5, according to 3 point Linear predicted value Y
l(k+1), 5 squares of predicted value Y
s(k+1) and the value of weighting coefficient α, by formula (5), calculate target at the predicted position Y'(k+1 of next frame):
Y'(k+1)＝αY
L(k+1)+(1-α)Y
S(k+1) （5）。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210364748.0A CN103700112A (en) | 2012-09-27 | 2012-09-27 | Sheltered target tracking method based on mixed predicting strategy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210364748.0A CN103700112A (en) | 2012-09-27 | 2012-09-27 | Sheltered target tracking method based on mixed predicting strategy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103700112A true CN103700112A (en) | 2014-04-02 |
Family
ID=50361630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210364748.0A Pending CN103700112A (en) | 2012-09-27 | 2012-09-27 | Sheltered target tracking method based on mixed predicting strategy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103700112A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107452020A (en) * | 2017-08-04 | 2017-12-08 | 河北汉光重工有限责任公司 | A kind of the anti-of adaptive template matching blocks tracking |
| CN110276785A (en) * | 2019-06-24 | 2019-09-24 | 电子科技大学 | One kind is anti-to block infrared object tracking method |
| CN111526422A (en) * | 2019-02-01 | 2020-08-11 | 网宿科技股份有限公司 | Method, system and equipment for fitting target object in video frame |
| CN111654700A (en) * | 2020-06-19 | 2020-09-11 | 杭州海康威视数字技术股份有限公司 | Privacy mask processing method and device, electronic equipment and monitoring system |
| CN114022468A (en) * | 2021-11-12 | 2022-02-08 | 珠海安联锐视科技股份有限公司 | Method for detecting article leaving and losing in security monitoring |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101794385A (en) * | 2010-03-23 | 2010-08-04 | 上海交通大学 | Multi-angle multi-target fast human face tracking method used in video sequence |
| CN101976504A (en) * | 2010-10-13 | 2011-02-16 | 北京航空航天大学 | Multi-vehicle video tracking method based on color space information |
| WO2011146054A1 (en) * | 2010-05-18 | 2011-11-24 | William Bohn | Improved consumer rewards systems and methods |
-
2012
- 2012-09-27 CN CN201210364748.0A patent/CN103700112A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101794385A (en) * | 2010-03-23 | 2010-08-04 | 上海交通大学 | Multi-angle multi-target fast human face tracking method used in video sequence |
| WO2011146054A1 (en) * | 2010-05-18 | 2011-11-24 | William Bohn | Improved consumer rewards systems and methods |
| CN101976504A (en) * | 2010-10-13 | 2011-02-16 | 北京航空航天大学 | Multi-vehicle video tracking method based on color space information |
Non-Patent Citations (3)
| Title |
|---|
| 沈志熙等: "均值漂移算法中的目标模型更新方法", 《自动化学报》, 31 May 2009 (2009-05-31) * |
| 翁木云,谢宇昕: "一种改进的自适应质心跟踪算法", 《空军工程大学学报(自然科学版)》, 30 April 2009 (2009-04-30) * |
| 邹青刚: "一种遮挡情况下的目标追踪方法", 《计算机应用与软件》, 30 September 2011 (2011-09-30) * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107452020A (en) * | 2017-08-04 | 2017-12-08 | 河北汉光重工有限责任公司 | A kind of the anti-of adaptive template matching blocks tracking |
| CN107452020B (en) * | 2017-08-04 | 2021-04-06 | 河北汉光重工有限责任公司 | Anti-occlusion tracking method for adaptive template matching |
| CN111526422A (en) * | 2019-02-01 | 2020-08-11 | 网宿科技股份有限公司 | Method, system and equipment for fitting target object in video frame |
| CN110276785A (en) * | 2019-06-24 | 2019-09-24 | 电子科技大学 | One kind is anti-to block infrared object tracking method |
| CN110276785B (en) * | 2019-06-24 | 2023-03-31 | 电子科技大学 | Anti-shielding infrared target tracking method |
| CN111654700A (en) * | 2020-06-19 | 2020-09-11 | 杭州海康威视数字技术股份有限公司 | Privacy mask processing method and device, electronic equipment and monitoring system |
| CN111654700B (en) * | 2020-06-19 | 2022-12-06 | 杭州海康威视数字技术股份有限公司 | Privacy mask processing method and device, electronic equipment and monitoring system |
| CN114022468A (en) * | 2021-11-12 | 2022-02-08 | 珠海安联锐视科技股份有限公司 | Method for detecting article leaving and losing in security monitoring |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110021033A (en) | A kind of method for tracking target based on the twin network of pyramid | |
| CN101799968B (en) | Detection method and device for oil well intrusion based on video image intelligent analysis | |
| CN103105924B (en) | Man-machine interaction method and device | |
| CN103208008A (en) | Fast adaptation method for traffic video monitoring target detection based on machine vision | |
| CN103700112A (en) | Sheltered target tracking method based on mixed predicting strategy | |
| CN110705412A (en) | Video target detection method based on motion history image | |
| CN101237522A (en) | Motion detection method and device | |
| CN105335701A (en) | Pedestrian detection method based on HOG and D-S evidence theory multi-information fusion | |
| CN109711332B (en) | Regression algorithm-based face tracking method and application | |
| CN110796018A (en) | Hand motion recognition method based on depth image and color image | |
| CN111191535B (en) | Pedestrian detection model construction method based on deep learning and pedestrian detection method | |
| CN108320306A (en) | Merge the video target tracking method of TLD and KCF | |
| CN103810718A (en) | Method and device for detection of violently moving target | |
| CN101908214A (en) | Moving object detection method with background reconstruction based on neighborhood correlation | |
| CN103500456A (en) | Object tracking method and equipment based on dynamic Bayes model network | |
| CN110728700B (en) | Moving target tracking method and device, computer equipment and storage medium | |
| CN110472607A (en) | A kind of ship tracking method and system | |
| JP2021528767A (en) | Visual search methods, devices, computer equipment and storage media | |
| CN107452019B (en) | Target detection method, device and system based on model switching and storage medium | |
| CN112686122B (en) | Human body and shadow detection method and device, electronic equipment and storage medium | |
| CN111539390A (en) | Small target image identification method, equipment and system based on Yolov3 | |
| Cho et al. | FPGA-based real-time visual tracking system using adaptive color histograms | |
| CN111768427A (en) | Multi-moving-target tracking method and device and storage medium | |
| CN110633641A (en) | Intelligent security pedestrian detection method, system and device and storage medium | |
| CN112149698A (en) | Method and device for screening difficult sample data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140402 |