CN112633105A - Target tracking and counting system and method - Google Patents
Target tracking and counting system and method Download PDFInfo
- Publication number
- CN112633105A CN112633105A CN202011479985.2A CN202011479985A CN112633105A CN 112633105 A CN112633105 A CN 112633105A CN 202011479985 A CN202011479985 A CN 202011479985A CN 112633105 A CN112633105 A CN 112633105A
- Authority
- CN
- China
- Prior art keywords
- tracking
- target
- tracking target
- shielded
- judging
- 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 38
- 238000012549 training Methods 0.000 claims abstract description 21
- 239000011159 matrix material Substances 0.000 claims abstract description 8
- 125000004122 cyclic group Chemical group 0.000 claims abstract description 4
- 230000004044 response Effects 0.000 claims description 50
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000007619 statistical method Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 14
- 230000000007 visual effect Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
Abstract
The invention relates to the technical field of moving target tracking, in particular to a target tracking and counting system and a method, which comprises the following steps: a KCF tracker, a Kalman filter and a classifier; further comprising: the acquisition unit is used for acquiring a video frame sequence and positioning a tracking target; an initialization unit for initializing a KCF tracker and a Kalman filter; the training unit is used for acquiring a sample image of a tracking target by adopting a cyclic matrix method and training a classifier; the judging unit is used for judging whether the tracking target is shielded or not; and the output unit is used for outputting the tracking result and the statistical result of the tracking target. The method verifies the tracking result of the KCF tracker and the prediction result predicted by the Kalman filter mutually, continuously corrects the error, and solves the technical problem that the target is repeatedly counted in the process of tracking the moving target in the prior art.
Description
Technical Field
The invention relates to the technical field of moving target tracking, in particular to a target tracking and counting system and a target tracking and counting method.
Background
The moving target tracking is the basis of modern visual technology and is widely applied to the field of security protection. Due to the translation or rotation motion of the target in the tracking process, the distance from the camera and the visual angle of the camera can be changed, so that the scale expansion and contraction change and the deformation of the target image can be caused, and the tracking failure can be caused. Under the condition that the target is shielded, due to the movement of the target, the covered and exposed areas of the target do not have a corresponding relation in the front and back two frames of images, and irregular change of the target image is caused, so that difficulty is caused in accurate tracking.
In contrast, chinese patent CN102521840A discloses a moving object tracking method, which includes the steps of: acquiring a current frame reference template, and detecting a new target on a current frame according to the current frame reference template; determining a target template according to the detection result of the new target; establishing a target template observation model according to a target template; sampling a quasi-target corresponding to the target template on the current frame, and establishing a quasi-target observation model according to the quasi-target; calculating the similarity of the target template observation model and the quasi-target observation model, and predicting the target position according to the similarity of the target template observation model and the quasi-target observation model; and establishing a predicted target observation model according to the target position, calculating the similarity between the target template observation model and the predicted target observation model, and determining a reference template of the next frame according to the similarity between the target template observation model and the predicted target observation model.
According to the technical scheme, the prediction and reference templates of the moving target can be accurately updated in real time in the moving target tracking process, and the tracking target is not easily lost when the shape and the size of the moving target are changed. However, under the condition that the moving objects are mutually shielded, due to the fact that the real-time reference template is updated, images of the same object before shielding and the same object after shielding have extremely high similarity, the shielded moving objects are likely to be tracked twice, namely, the moving objects are tracked once before shielding and tracked once after shielding. That is to say, the prior art has the problem that the target is repeatedly counted in the process of tracking the moving target.
Disclosure of Invention
The invention provides a target tracking and counting system, which solves the technical problem that in the prior art, a target is repeatedly counted in the process of tracking a moving target.
The basic scheme provided by the invention is as follows: a target tracking, statistics system, comprising: a KCF tracker, a Kalman filter and a classifier; further comprising:
the acquisition unit is used for acquiring a video frame sequence and positioning a tracking target;
an initialization unit for initializing a KCF tracker and a Kalman filter according to a video frame sequence;
the training unit is used for acquiring a sample image of a tracking target from the video frame sequence by adopting a cyclic matrix method and training a classifier according to the sample image;
the judging unit is used for detecting a sample image of the tracking target by using the thread pool, acquiring a thread with the maximum response, taking an output position corresponding to the thread with the maximum response as a target position of the current frame, comparing the acquired maximum response with a preset threshold value, and judging whether the tracking target is shielded:
if the tracking target is not shielded, the output result of the KCF tracker is used as the tracking result and is used as the target position of the current frame, and a command of continuing tracking is sent to the KCF tracker; counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially shielded, sending an instruction for stopping training to the classifier; if the tracking target is seriously shielded, judging whether the tracking fails, sending an instruction for repositioning the tracking target to the acquisition unit when the tracking fails, and sending an instruction for continuing tracking after reinitialization to the KCF tracker;
and the output unit is used for outputting the tracking result and the statistical result of the tracking target.
The working principle and the advantages of the invention are as follows: the advantages of the KCF tracker and the Kalman filter are fused, and when the target is not shielded, the KCF tracker can quickly and accurately track the tracked target; when shielding occurs, the Kalman filter does not track based on the characteristics of the tracking target, but fully utilizes the current state of the tracking target to predict the position of the target of the next frame, and predicts based on the tracking result given by the KCF tracker before to obtain the prediction result. In the subsequent tracking process, the tracking result of the KCF tracker and the prediction result predicted by the Kalman filter are mutually verified, and the error correction is continuously carried out. In this way, if the tracked target is occluded, errors caused by occlusion can be corrected, and a good tracking result can be obtained, so that the same target cannot be repeatedly counted.
The method verifies the tracking result of the KCF tracker and the prediction result predicted by the Kalman filter mutually, continuously corrects the error, and solves the technical problem that the target is repeatedly counted in the process of tracking the moving target in the prior art.
Further, the judging unit is further configured to obtain a response peak value of the classifier, and judge, according to the response peak value and the magnitudes of the first threshold and the second threshold, an occlusion condition of the tracking target: if the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded; if the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded; if the response peak value is less than or equal to the second threshold value, judging that the tracking target is seriously shielded; wherein the first threshold is greater than the second threshold.
Has the advantages that: for the KCF tracking algorithm, when a tracking target is shielded by an object, an image in a target frame becomes a shielding object, the surface is more likely to be the tracking target when the response value of the classifier is larger, the surface tracking target is more likely to be shielded when the response value is smaller, and the judgment mode is reliable and quick.
Further, the judging unit is further configured to obtain a babbitt distance of the gray level histogram of the tracking target of the current frame and the previous frame, and judge whether the tracking fails according to the babbitt distance and a third threshold: if the Papanicolaou distance is larger than or equal to a third threshold value, judging that the tracking fails; if the pap distance < the third threshold, it is determined that tracking has not failed.
Has the advantages that: the greater the Babbitt distance value is, the greater the image difference in the target frame is, and the poor tracking result is shown; on the contrary, the smaller the Babbitt distance value is, the smaller the image difference in the target frame is, and the good tracking result is shown; the judgment result is reliable and strong in operability.
Further, the judging unit is further configured to extract a color histogram of the tracking target, and obtain the color histogram and the euclidean distance of the tracking target that are initially set; and judging whether the Euclidean distance between the two objects is greater than a distance threshold value, and if the Euclidean distance between the two objects is greater than the distance threshold value, judging that the tracking target is shielded.
Has the advantages that: the KCF tracks the tracked target by adopting the HOG characteristic and judges through the Euclidean distance, namely two layers of judgment are carried out, so that the judgment is more accurate.
Further, the acquisition unit adopts a template matching algorithm to reposition the tracking target, and the template matching algorithm is a square error matching method.
Has the advantages that: the template matching is to search the most similar area with the template image in one image, has high calculation speed and is simple to realize when being applied to the fields of target identification and target tracking.
The invention also provides a target tracking and statistical method, which comprises the following steps:
s1, acquiring a video frame sequence and positioning a tracking target;
s2, initializing a KCF tracker and a Kalman filter according to the video frame sequence;
s3, acquiring sample images of the tracking target from the video frame sequence by adopting a circular matrix method, and training a classifier according to the sample images;
s4, judging whether the tracking target is blocked:
if the tracking target is not shielded, taking the output result of the KCF tracker as the tracking result and the current frame target position, turning to S5, and counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially occluded, go to S3; if the tracking target is seriously shielded, judging whether the tracking fails, if the tracking fails, relocating the tracking target, reinitializing the KCF tracker, and then turning to S5;
s5, continuing to track the tracking target;
and S6, outputting the tracking result and the statistical result of the tracking target.
The working principle and the advantages of the invention are as follows: when shielding occurs, the Kalman filter fully utilizes the current state of the tracking target to predict the target position of the next frame, and prediction is carried out based on the tracking result given by the KCF tracker before, so as to obtain the prediction result. In the subsequent tracking process, the tracking result of the KCF tracker and the prediction result predicted by the Kalman filter are mutually verified, and error correction is continuously carried out, so that repeated counting on the tracking target is avoided.
Further, in S4, it is determined whether the tracking target is occluded, specifically as follows:
a1, obtaining a response peak value of the classifier;
a2, judging the shielding condition of the tracking target according to the response peak value, the first threshold value and the second threshold value:
if the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded;
if the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded;
and if the response peak value is less than or equal to the second threshold value, judging that the tracking target is seriously shielded.
Has the advantages that: the larger the response value of the classifier is, the more likely the surface is a tracking target, the smaller the response value is, the more likely the surface tracking target is shielded, and the judgment is reliable and quick through the response peak value.
Further, S4 includes:
a3, acquiring the Papanicolaou distance of the gray level histogram of the tracking target of the current frame and the previous frame;
a4, judging whether the tracking fails according to the Babbitt distance and the size of a third threshold value: if the Papanicolaou distance is larger than or equal to a third threshold value, judging that the tracking fails; if the pap distance < the third threshold, it is determined that tracking has not failed.
Has the advantages that: the smaller the Babbitt distance value is, the smaller the image difference in the target frame is, the better the tracking result is, and the judgment mode is strong in operability and can not increase errors.
Drawings
Fig. 1 is a block diagram of a system structure of an embodiment of a target tracking and statistics system according to the present invention.
Detailed Description
The following is further detailed by the specific embodiments:
example 1
The embodiment of the target tracking and counting system provided by the invention is basically as shown in the attached figure 1: the method comprises the following steps: a KCF tracker, a Kalman filter and a classifier; further comprising:
the acquisition unit is used for acquiring a video frame sequence and positioning a tracking target;
an initialization unit for initializing a KCF tracker and a Kalman filter according to a video frame sequence;
the training unit is used for acquiring a sample image of a tracking target from the video frame sequence by adopting a cyclic matrix method and training a classifier according to the sample image;
the judging unit is used for detecting a sample image of the tracking target by using the thread pool, acquiring a thread with the maximum response, taking an output position corresponding to the thread with the maximum response as a target position of the current frame, comparing the acquired maximum response with a preset threshold value, and judging whether the tracking target is shielded:
if the tracking target is not shielded, the output result of the KCF tracker is used as the tracking result and is used as the target position of the current frame, and a command of continuing tracking is sent to the KCF tracker; counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially shielded, sending an instruction for stopping training to the classifier; if the tracking target is seriously shielded, judging whether the tracking fails, sending an instruction for repositioning the tracking target to the acquisition unit when the tracking fails, and sending an instruction for continuing tracking after reinitialization to the KCF tracker;
and the output unit is used for outputting the tracking result and the statistical result of the tracking target.
In this embodiment, the KCF tracker, the kalman filter, the classifier, the acquisition unit, the initialization unit, the training unit, the judgment unit, and the output unit are all integrated or mounted on a server, and the functions thereof are realized by software/programs/codes.
The specific implementation process is as follows:
firstly, an acquisition unit reads in a video frame sequence, selects a tracking target to be tracked and positions the tracking target; the initialization unit initializes the KCF tracker and the Kalman filter according to the video frame sequence, the training unit acquires sample images of a tracking target from the video frame sequence by adopting a circular matrix method, and trains the classifier according to the sample images.
Then, the judgment unit judges whether the tracking target is occluded. Detecting a sample image of a tracking target by using a thread pool, acquiring a thread with the largest response, taking an output position corresponding to the thread with the largest response as a target position of a current frame, comparing the acquired maximum response with a preset threshold, judging that the tracking target of the current frame is shielded if the maximum response is smaller than the preset threshold, and judging that the tracking target of the current frame is not shielded if the maximum response is larger than or equal to the preset threshold.
If the tracking target is not shielded, the output result of the KCF tracker is used as the tracking result and is used as the target position of the current frame, the judgment unit sends a command of continuing tracking to the KCF tracker, and the KCF tracker tracks the tracking target after receiving the command of continuing tracking; and meanwhile, counting the number of the tracking targets to obtain a statistical result.
If the tracking target is occluded, the output result of the Kalman filter is used as the target position of the current frame, and the judging unit judges whether the tracking target is seriously occluded or partially occluded: if the tracking target is partially shielded, sending an instruction for stopping training to the classifier; if the tracking target is seriously shielded, judging whether the tracking fails, sending an instruction for repositioning the tracking target to the acquisition unit when the tracking fails, and sending an instruction for continuing tracking after reinitialization to the KCF tracker.
In this embodiment, a response peak of the classifier is obtained, and the shielding condition of the tracking target is determined according to the response peak, a preset first threshold and a preset second threshold: if the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded; if the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded; if the response peak value is less than or equal to the second threshold value, judging that the tracking target is seriously shielded; wherein the first threshold is greater than the second threshold.
For example, the first threshold is preset to 0.32, and the second threshold is preset to 0.21. And if the response peak value is 0.33, namely the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded, sending a command of continuing tracking to the KCF tracker, and tracking the tracking target after the KCF tracker receives the command of continuing tracking. And if the response peak value is 0.27, namely the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded, sending an instruction for stopping training to the classifier, and stopping training the classifier after the classifier receives the instruction for stopping training.
If the response peak value is 0.19, namely the response peak value is less than or equal to the second threshold value, the tracking target is judged to be seriously shielded, then the judging unit acquires the Papanicolaou distance of the gray level histogram of the current frame and the previous frame of the tracking target, and judges whether the tracking fails according to the size of the Papanicolaou distance and the preset third threshold value. For example, the third threshold is preset to be 0.25, and if the babbit distance is 0.24, that is, the babbit distance is less than the third threshold, it is determined that the tracking has not failed; if the babbit distance is 0.26, namely the babbit distance is larger than or equal to a third threshold value, the tracking failure is judged, the judging unit sends an instruction for repositioning the tracking target to the acquiring unit, and sends an instruction for continuing tracking after reinitialization to the KCF tracker; after receiving the instruction of repositioning the tracking target, the acquisition unit repositions the tracking target by adopting a template matching algorithm, namely a square error matching method in the embodiment; after the KCF tracker receives the command of continuing tracking after reinitialization, the initialization unit initializes the KCF tracker first, and then the KCF tracker continues to track the relocated tracking target.
And finally, outputting the tracking result and the statistical result of the tracked target by an output unit in a visual mode, such as a rendering graph and a track graph matched table mode after the tracking process is finished.
Based on the target tracking and counting system, the target tracking and counting method is further disclosed, and comprises the following steps:
s1, acquiring a video frame sequence and positioning a tracking target;
s2, initializing a KCF tracker and a Kalman filter according to the video frame sequence;
s3, acquiring sample images of the tracking target from the video frame sequence by adopting a circular matrix method, and training a classifier according to the sample images;
s4, judging whether the tracking target is blocked:
if the tracking target is not shielded, taking the output result of the KCF tracker as the tracking result and the current frame target position, turning to S5, and counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially occluded, go to S3; if the tracking target is seriously shielded, judging whether the tracking fails, if the tracking fails, relocating the tracking target, reinitializing the KCF tracker, and then turning to S5;
s5, continuing to track the tracking target;
and S6, outputting the tracking result and the statistical result of the tracking target.
Example 2
The difference from embodiment 1 is that, when the determination unit determines whether the tracking target is occluded, the determination unit extracts the color histogram of the tracking target, acquires the color histogram and the euclidean distance of the tracking target that are initially set, determines whether the tracking target is occluded according to the euclidean distance between the two and the distance threshold, and determines that the tracking target is occluded if the euclidean distance between the two is greater than the distance threshold. For example, the distance threshold is 0.20, and if the euclidean distance between the two is 0.22 and is greater than the distance threshold, it is determined that the tracking target is occluded; and if the Euclidean distance between the two is 0.19 and is smaller than the distance threshold, judging that the tracking target is not shielded.
Example 3
The difference from the embodiment 2 is only that the tracking target is a human, and the sample image is a human face image of the tracking target acquired from a plurality of angles in advance through a camera; and the whole collection process is not carried out simultaneously and simultaneously but is carried out in different time and places.
Specifically, during the whole pre-acquisition process, the tracking target moves slowly towards the camera, and the camera acquires the face image at a preset time interval, such as 2 seconds, during the moving process of the tracking target. The collected face images are multi-angle and dynamic, and contain dynamic characteristics of human motion, and the tracking accuracy can be greatly improved by training the classifier.
Meanwhile, if the human face is shielded in the whole pre-acquisition process, an image recognition algorithm is adopted to judge whether the human face is actively shielded or passively shielded. That is, the image recognition algorithm is adopted to recognize the type of the sheltering object, if the sheltering object is a cap, clothes and other common clothes decoration, the passive sheltering is described, the tracking target is not intentionally sheltered, and the face image is effective; on the contrary, the sheltering object is an unusual object such as a helmet, a head cover and the like, which is indicated as active sheltering, the tracking target is deliberately sheltered, the face image is invalid, and the invalid face image is deleted.
The foregoing is merely an example of the present invention, and common general knowledge in the field of known specific structures and characteristics is not described herein in any greater extent than that known in the art at the filing date or prior to the priority date of the application, so that those skilled in the art can now appreciate that all of the above-described techniques in this field and have the ability to apply routine experimentation before this date can be combined with one or more of the present teachings to complete and implement the present invention, and that certain typical known structures or known methods do not pose any impediments to the implementation of the present invention by those skilled in the art. It should be noted that, for those skilled in the art, without departing from the structure of the present invention, several changes and modifications can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicability of the patent. The scope of the claims of the present application shall be determined by the contents of the claims, and the description of the embodiments and the like in the specification shall be used to explain the contents of the claims.
Claims (8)
1. A target tracking, statistics system, comprising: a KCF tracker, a Kalman filter and a classifier; it is characterized by also comprising:
the acquisition unit is used for acquiring a video frame sequence and positioning a tracking target;
an initialization unit for initializing a KCF tracker and a Kalman filter according to a video frame sequence;
the training unit is used for acquiring a sample image of a tracking target from the video frame sequence by adopting a cyclic matrix method and training a classifier according to the sample image;
the judging unit is used for detecting a sample image of the tracking target by using the thread pool, acquiring a thread with the maximum response, taking an output position corresponding to the thread with the maximum response as a target position of the current frame, comparing the acquired maximum response with a preset threshold value, and judging whether the tracking target is shielded:
if the tracking target is not shielded, the output result of the KCF tracker is used as the tracking result and is used as the target position of the current frame, and a command of continuing tracking is sent to the KCF tracker; counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially shielded, sending an instruction for stopping training to the classifier; if the tracking target is seriously shielded, judging whether the tracking fails, sending an instruction for repositioning the tracking target to the acquisition unit when the tracking fails, and sending an instruction for continuing tracking after reinitialization to the KCF tracker;
and the output unit is used for outputting the tracking result and the statistical result of the tracking target.
2. The target tracking and statistics system of claim 1, wherein the determining unit is further configured to obtain a response peak value of the classifier, and determine an occlusion condition of the tracked target according to the response peak value and the magnitudes of the first threshold and the second threshold: if the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded; if the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded; if the response peak value is less than or equal to the second threshold value, judging that the tracking target is seriously shielded; wherein the first threshold is greater than the second threshold.
3. The target tracking and statistics system of claim 2, wherein the determining unit is further configured to obtain the babbitt distance of the gray histogram of the tracking target of the current frame and the previous frame, and determine whether the tracking fails according to the babbitt distance and the third threshold: if the Papanicolaou distance is larger than or equal to a third threshold value, judging that the tracking fails; if the pap distance < the third threshold, it is determined that tracking has not failed.
4. The target tracking and statistics system of claim 3, wherein the determination unit is further configured to extract a color histogram of the tracking target, and obtain the color histogram and the euclidean distance of the tracking target initially set; and judging whether the Euclidean distance between the two objects is greater than a distance threshold value, and if the Euclidean distance between the two objects is greater than the distance threshold value, judging that the tracking target is shielded.
5. The target tracking and statistics system of claim 4, wherein the acquisition unit relocates the tracked target using a template matching algorithm, the template matching algorithm being a squared error matching method.
6. A target tracking and statistical method is characterized by comprising the following steps:
s1, acquiring a video frame sequence and positioning a tracking target;
s2, initializing a KCF tracker and a Kalman filter according to the video frame sequence;
s3, acquiring sample images of the tracking target from the video frame sequence by adopting a circular matrix method, and training a classifier according to the sample images;
s4, judging whether the tracking target is blocked:
if the tracking target is not shielded, taking the output result of the KCF tracker as the tracking result and the current frame target position, turning to S5, and counting the number of the tracking targets to obtain a statistical result;
if the tracking target is shielded, the output result of the Kalman filter is used as the target position of the current frame, and whether the tracking target is seriously shielded or partially shielded is judged: if the tracking target is partially occluded, go to S3; if the tracking target is seriously shielded, judging whether the tracking fails, if the tracking fails, relocating the tracking target, reinitializing the KCF tracker, and then turning to S5;
s5, continuing to track the tracking target;
and S6, outputting the tracking result and the statistical result of the tracking target.
7. The target tracking and statistics method according to claim 6, wherein in S4, whether the tracked target is occluded is determined as follows:
a1, obtaining a response peak value of the classifier;
a2, judging the shielding condition of the tracking target according to the response peak value, the first threshold value and the second threshold value:
if the response peak value is larger than or equal to the first threshold value, judging that the tracking target is not shielded;
if the second threshold value is less than the response peak value and less than the first threshold value, judging that the tracking target is partially shielded;
and if the response peak value is less than or equal to the second threshold value, judging that the tracking target is seriously shielded.
8. The method for tracking and counting objects as claimed in claim 7, wherein the step S4 further comprises:
a3, acquiring the Papanicolaou distance of the gray level histogram of the tracking target of the current frame and the previous frame;
a4, judging whether the tracking fails according to the Babbitt distance and the size of a third threshold value: if the Papanicolaou distance is larger than or equal to a third threshold value, judging that the tracking fails; if the pap distance < the third threshold, it is determined that tracking has not failed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011479985.2A CN112633105A (en) | 2020-12-15 | 2020-12-15 | Target tracking and counting system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011479985.2A CN112633105A (en) | 2020-12-15 | 2020-12-15 | Target tracking and counting system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112633105A true CN112633105A (en) | 2021-04-09 |
Family
ID=75313202
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011479985.2A Pending CN112633105A (en) | 2020-12-15 | 2020-12-15 | Target tracking and counting system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112633105A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113436228A (en) * | 2021-06-22 | 2021-09-24 | 中科芯集成电路有限公司 | Anti-blocking and target recapturing method of correlation filtering target tracking algorithm |
| CN114092515A (en) * | 2021-11-08 | 2022-02-25 | 国汽智控(北京)科技有限公司 | Target tracking detection method, device, equipment and medium for obstacle blocking |
| CN116109975A (en) * | 2023-02-08 | 2023-05-12 | 广州宝立科技有限公司 | Power grid safety operation monitoring image processing method and intelligent video monitoring system |
| CN117036740A (en) * | 2023-08-04 | 2023-11-10 | 上海第二工业大学 | Anti-occlusion tracking method for moving target |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107657630A (en) * | 2017-07-21 | 2018-02-02 | 南京邮电大学 | A kind of modified anti-shelter target tracking based on KCF |
| CN108010067A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of visual target tracking method based on combination determination strategy |
| CN108765470A (en) * | 2018-06-17 | 2018-11-06 | 天津理工大学 | One kind being directed to the improved KCF track algorithms of target occlusion |
| WO2019074601A1 (en) * | 2017-10-13 | 2019-04-18 | Qualcomm Incorporated | Object tracking for neural network systems |
| CN109712171A (en) * | 2018-12-28 | 2019-05-03 | 上海极链网络科技有限公司 | A kind of Target Tracking System and method for tracking target based on correlation filter |
| EP3723046A1 (en) * | 2017-12-29 | 2020-10-14 | Huawei Technologies Co., Ltd. | Target tracking method and device |
-
2020
- 2020-12-15 CN CN202011479985.2A patent/CN112633105A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107657630A (en) * | 2017-07-21 | 2018-02-02 | 南京邮电大学 | A kind of modified anti-shelter target tracking based on KCF |
| WO2019074601A1 (en) * | 2017-10-13 | 2019-04-18 | Qualcomm Incorporated | Object tracking for neural network systems |
| CN108010067A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of visual target tracking method based on combination determination strategy |
| EP3723046A1 (en) * | 2017-12-29 | 2020-10-14 | Huawei Technologies Co., Ltd. | Target tracking method and device |
| CN108765470A (en) * | 2018-06-17 | 2018-11-06 | 天津理工大学 | One kind being directed to the improved KCF track algorithms of target occlusion |
| CN109712171A (en) * | 2018-12-28 | 2019-05-03 | 上海极链网络科技有限公司 | A kind of Target Tracking System and method for tracking target based on correlation filter |
Non-Patent Citations (2)
| Title |
|---|
| AISHWARYA S N RANI等: "Robust object tracking using kernalized correlation filters (KCF) and Kalman predictive estimates", 《2017 2ND IEEE INTERNATIONAL CONFERENCE ON RECENT TRENDS IN ELECTRONICS, INFORMATION & COMMUNICATION TECHNOLOGY (RTEICT)》 * |
| 宋俊芳等: "基于数据关联融合KCF与Kalman滤波的", 《科学技术与工程》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113436228A (en) * | 2021-06-22 | 2021-09-24 | 中科芯集成电路有限公司 | Anti-blocking and target recapturing method of correlation filtering target tracking algorithm |
| CN113436228B (en) * | 2021-06-22 | 2024-01-23 | 中科芯集成电路有限公司 | Anti-shielding and target recapturing method of related filtering target tracking algorithm |
| CN114092515A (en) * | 2021-11-08 | 2022-02-25 | 国汽智控(北京)科技有限公司 | Target tracking detection method, device, equipment and medium for obstacle blocking |
| CN114092515B (en) * | 2021-11-08 | 2024-03-05 | 国汽智控(北京)科技有限公司 | Target tracking detection method, device, equipment and medium for obstacle shielding |
| CN116109975A (en) * | 2023-02-08 | 2023-05-12 | 广州宝立科技有限公司 | Power grid safety operation monitoring image processing method and intelligent video monitoring system |
| CN116109975B (en) * | 2023-02-08 | 2023-10-20 | 广州宝立科技有限公司 | Power grid safety operation monitoring image processing method and intelligent video monitoring system |
| CN117036740A (en) * | 2023-08-04 | 2023-11-10 | 上海第二工业大学 | Anti-occlusion tracking method for moving target |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112633105A (en) | Target tracking and counting system and method | |
| WO2021196294A1 (en) | Cross-video person location tracking method and system, and device | |
| US10551854B2 (en) | Method for detecting target object, detection apparatus and robot | |
| JP6448223B2 (en) | Image recognition system, image recognition apparatus, image recognition method, and computer program | |
| Yang et al. | Robust superpixel tracking | |
| JP6273685B2 (en) | Tracking processing apparatus, tracking processing system including the tracking processing apparatus, and tracking processing method | |
| CN101593022B (en) | Method for quick-speed human-computer interaction based on finger tip tracking | |
| CN110264493B (en) | Method and device for tracking multiple target objects in motion state | |
| CN104217428B (en) | A kind of fusion feature matching and the video monitoring multi-object tracking method of data correlation | |
| JP2018523868A5 (en) | ||
| CN110569785B (en) | Face recognition method integrating tracking technology | |
| CN109766796B (en) | Deep pedestrian detection method for dense crowd | |
| CN110827321B (en) | Multi-camera collaborative active target tracking method based on three-dimensional information | |
| Pellegrini et al. | Human posture tracking and classification through stereo vision and 3d model matching | |
| Yang et al. | Multiple marker tracking in a single-camera system for gait analysis | |
| CN109902543A (en) | Target trajectory estimation method, device and Target Tracking System | |
| CN113033353A (en) | Pedestrian trajectory generation method based on overlook image, storage medium and electronic device | |
| CN111654668B (en) | Monitoring equipment synchronization method and device and computer terminal | |
| CN113033348A (en) | Overlook image correction method for pedestrian re-recognition, storage medium, and electronic device | |
| TWI641265B (en) | Mobile target position tracking system | |
| CN112307897A (en) | Pet tracking method based on local feature recognition and adjacent frame matching in community monitoring scene | |
| CN108985216B (en) | Pedestrian head detection method based on multivariate logistic regression feature fusion | |
| Sun et al. | Detecting acceleration for gait and crime scene analysis | |
| CN115767424A (en) | Video positioning method based on RSS and CSI fusion | |
| JP7375806B2 (en) | Image processing device and image processing method |
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: 20210409 |
|
| RJ01 | Rejection of invention patent application after publication |