CN107169991A - A kind of moving target detecting method of multilayer background model - Google Patents
A kind of moving target detecting method of multilayer background model Download PDFInfo
- Publication number
- CN107169991A CN107169991A CN201710328438.6A CN201710328438A CN107169991A CN 107169991 A CN107169991 A CN 107169991A CN 201710328438 A CN201710328438 A CN 201710328438A CN 107169991 A CN107169991 A CN 107169991A
- Authority
- CN
- China
- Prior art keywords
- code word
- model
- pixel
- background model
- background
- 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.)
- Withdrawn
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
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/251—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving models
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of moving target detecting method of multilayer background model, comprise the following steps:S1:Moving Targets Based on Video Streams image is gathered, and video image is pre-processed;S2:The neighborhood sample of each pixel is extracted from the first frame of video sequence, for initial background model, only needs a two field picture to complete the initialization of background model;S3:Random sampling technique is introduced, a unmatched code word is randomly selected, is replaced using new background pixel, it is to avoid the code word of mistake classification is resided in background model for a long time;S4:Using multilayer background model strategy, successively checking of each pixel Jing Guo multilayer background model, it is ensured that the accuracy of background model.
Description
Technical field
The present invention relates to field, and in particular to a kind of moving target detecting method of multilayer background model.
Background technology
Intelligent video monitoring for today's society security administration and maintain social stability and serve important function, answered extensively
For in the industry of civilian and army.Following deficiency is still suffered from mobile target in complex background detection algorithm:
(1)Background model initializing overlong time;
(2)The renewal coefficient setting of background model does not have versatility;
(3)Single-layer background model is difficult to handle complex background interference problem.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of moving target detecting method of multilayer background model.
A kind of moving target detecting method of multilayer background model, comprises the following steps:
S1:Moving Targets Based on Video Streams image is gathered, and video image is pre-processed;
S2:The neighborhood sample of each pixel is extracted from the first frame of video sequence, for initial background model, a frame is only needed
Image can complete the initialization of background model;
S3:Random sampling technique is introduced, a unmatched code word is randomly selected, is replaced, kept away using new background pixel
The code word of fault-avoidance misclassification is resided in background model for a long time;
S4:Using multilayer background model strategy, successively checking of each pixel Jing Guo multilayer background model, it is ensured that background model
Accuracy.
Further, the establishment of multilayer background model and renewal process are as follows:
1)Master cast M (x) is created based on neighborhood initial method, meanwhile, empty level cache H (x) is respectively created and delays with two grades
Deposit model U (x);
2)The input pixel given for t, find in the M (x) withThe code word of matchingIf,In M
(x) in, then it is assumed thatBelong to background pixel, master cast M (x) is updated using EPAM methods, step 3 is otherwise jumped to);
3)If not finding the code word of matching, handled according to the following steps:
A)By pixelLabeled as foreground pixel;
B)Matching code word is searched in level cache model H (x)If finding the code word of matching, then using EPAM
Method is updated to H (x) and otherwise usedCreate a new code wordIt is inserted into level cache model H (x);
C)Find in the L2 cache model U (x) withThe code word of matchingIf finding the code word of matching, then make
Model U (x) is updated with EPAM algorithms, otherwise usedCreate a new code wordIt is inserted into L2 cache model
In U (x);
4)Using pixel classifier 2, the real background code word lifting of missing inspection is arrived in master cast M (x), process is as follows:
A)Remove the maximum duration interval of code word in L2 cache model U (x)It is more thanAll code words;By code in U (x)
The maximum duration interval of wordIt is more thanThe lifting of all code words arrive in H (x), while deleting the maximum of code word in H (x)
Time interval is more thanAll code words;
B)Remove the maximum duration interval of code word in level cache model H (x)It is more thanAll code words;By code in H (x)
The maximum duration interval of wordIt is more thanThe lifting of all code words arrive in M (x), while when deleting the maximum of code word in M (x)
Between be spacedIt is more thanAll code words;
5)For new pixelJump to step 2)Handled.
The beneficial effects of the invention are as follows:
The present invention uses and chooses sample from the neighborhood of each pixel of the first two field picture, the side initialized to background model
Case, the initialization procedure of algorithm is simplified and accelerated;By using EPAM background models more new strategy, it is to avoid by mistake
The pixel of classification is retained in background model for a long time;Due to the use of multilayer background model method, i.e., created in initial phase
One master cast, creates two cache models, by the successively checking of cache model, by the background pixel of missing inspection in detection-phase
Master cast is lifted, the accuracy of master cast is enhanced, robustness of the algorithm under complex background is improved.
Embodiment
The present invention is further elaborated for specific examples below, but not as a limitation of the invention.
A kind of moving target detecting method of multilayer background model, comprises the following steps:
S1:Moving Targets Based on Video Streams image is gathered, and video image is pre-processed;
S2:The neighborhood sample of each pixel is extracted from the first frame of video sequence, for initial background model, a frame is only needed
Image can complete the initialization of background model;
S3:Random sampling technique is introduced, a unmatched code word is randomly selected, is replaced, kept away using new background pixel
The code word of fault-avoidance misclassification is resided in background model for a long time;
S4:Using multilayer background model strategy, successively checking of each pixel Jing Guo multilayer background model, it is ensured that background model
Accuracy.
The establishment of multilayer background model and renewal process are as follows:
1)Master cast M (x) is created based on neighborhood initial method, meanwhile, empty level cache H (x) is respectively created and delays with two grades
Deposit model U (x);
2)The input pixel given for t, find in the M (x) withThe code word of matchingIf,In M
(x) in, then it is assumed thatBelong to background pixel, master cast M (x) is updated using EPAM methods, step 3 is otherwise jumped to);
3)If not finding the code word of matching, handled according to the following steps:
A)By pixelLabeled as foreground pixel;
B)Matching code word is searched in level cache model H (x)If finding the code word of matching, then using EPAM
Method is updated to H (x) and otherwise usedCreate a new code wordIt is inserted into level cache model H (x);
C)Find in the L2 cache model U (x) withThe code word of matchingIf finding the code word of matching, then make
Model U (x) is updated with EPAM algorithms, otherwise usedCreate a new code wordIt is inserted into L2 cache model
In U (x);
4)Using pixel classifier 2, the real background code word lifting of missing inspection is arrived in master cast M (x), process is as follows:
A)Remove the maximum duration interval of code word in L2 cache model U (x)It is more thanAll code words;By code in U (x)
The maximum duration interval of wordIt is more thanThe lifting of all code words arrive in H (x), while deleting the maximum of code word in H (x)
Time interval is more thanAll code words;
B)Remove the maximum duration interval of code word in level cache model H (x)It is more thanAll code words;By code in H (x)
The maximum duration interval of wordIt is more thanThe lifting of all code words arrive in M (x), while when deleting the maximum of code word in M (x)
Between be spacedIt is more thanAll code words;
5)For new pixelJump to step 2)Handled.
Claims (2)
1. a kind of moving target detecting method of multilayer background model, it is characterised in that comprise the following steps:
S1:Moving Targets Based on Video Streams image is gathered, and video image is pre-processed;
S2:The neighborhood sample of each pixel is extracted from the first frame of video sequence, for initial background model, a frame is only needed
Image can complete the initialization of background model;
S3:Random sampling technique is introduced, a unmatched code word is randomly selected, is replaced, kept away using new background pixel
The code word of fault-avoidance misclassification is resided in background model for a long time;
S4:Using multilayer background model strategy, successively checking of each pixel Jing Guo multilayer background model, it is ensured that background model
Accuracy.
2. moving target detecting method according to claim 1, it is characterised in that the establishment and renewal of multilayer background model
Process is as follows:
1)Master cast M (x) is created based on neighborhood initial method, meanwhile, empty level cache H (x) is respectively created and delays with two grades
Deposit model U (x);
2)The input pixel given for t, find in the M (x) withThe code word of matchingIf,In M (x)
In, then it is assumed thatBelong to background pixel, master cast M (x) is updated using EPAM methods, step 3 is otherwise jumped to);
3)If not finding the code word of matching, handled according to the following steps:
A)By pixelLabeled as foreground pixel;
B)Matching code word is searched in level cache model H (x)If finding the code word of matching, then using EPAM
Method is updated to H (x) and otherwise usedCreate a new code wordIt is inserted into level cache model H (x);
C)Find in the L2 cache model U (x) withThe code word of matchingIf finding the code word of matching, then use
EPAM algorithms are updated to model U (x), are otherwise usedCreate a new code wordIt is inserted into L2 cache model U
(x) in;
4)Using pixel classifier 2, the real background code word lifting of missing inspection is arrived in master cast M (x), process is as follows:
A)Remove the maximum duration interval of code word in L2 cache model U (x)It is more thanAll code words;By code word in U (x)
Maximum duration intervalIt is more thanThe lifting of all code words arrive in H (x), while when deleting the maximum of code word in H (x)
Between be spaced and be more thanAll code words;
B)Remove the maximum duration interval of code word in level cache model H (x)It is more thanAll code words;By code in H (x)
The maximum duration interval of wordIt is more thanThe lifting of all code words arrive in M (x), while when deleting the maximum of code word in M (x)
Between be spacedIt is more thanAll code words;
5)For new pixelJump to step 2)Handled.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710328438.6A CN107169991A (en) | 2017-05-11 | 2017-05-11 | A kind of moving target detecting method of multilayer background model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710328438.6A CN107169991A (en) | 2017-05-11 | 2017-05-11 | A kind of moving target detecting method of multilayer background model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107169991A true CN107169991A (en) | 2017-09-15 |
Family
ID=59814855
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710328438.6A Withdrawn CN107169991A (en) | 2017-05-11 | 2017-05-11 | A kind of moving target detecting method of multilayer background model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107169991A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110060278A (en) * | 2019-04-22 | 2019-07-26 | 新疆大学 | The detection method and device of moving target based on background subtraction |
-
2017
- 2017-05-11 CN CN201710328438.6A patent/CN107169991A/en not_active Withdrawn
Non-Patent Citations (1)
Title |
---|
曹明伟 等: ""基于多层背景模型的运动目标检测"", 《电子学报》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110060278A (en) * | 2019-04-22 | 2019-07-26 | 新疆大学 | The detection method and device of moving target based on background subtraction |
CN110060278B (en) * | 2019-04-22 | 2023-05-12 | 新疆大学 | Method and device for detecting moving target based on background subtraction |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106845408B (en) | Street garbage identification method under complex environment | |
CN104980622B (en) | Image processing apparatus and image processing method | |
WO2019223254A1 (en) | Construction method for multi-scale lightweight face detection model and face detection method based on model | |
WO2020082258A1 (en) | Multi-objective real-time tracking method and apparatus, and electronic device | |
CN110378235A (en) | A kind of fuzzy facial image recognition method, device and terminal device | |
CN111126412B (en) | Image key point detection method based on characteristic pyramid network | |
WO2022188315A1 (en) | Video detection method and apparatus, electronic device, and storage medium | |
CN103942812A (en) | Moving object detection method based on Gaussian mixture and edge detection | |
CN105809651A (en) | Image saliency detection method based on edge non-similarity comparison | |
CN105354550B (en) | A kind of list method for extracting content based on image local feature point registration | |
WO2014036813A1 (en) | Method and device for extracting image features | |
CN109711375B (en) | Signal lamp identification method and device | |
CN111886600A (en) | Device and method for instance level segmentation of image | |
CN105279771A (en) | Method for detecting moving object on basis of online dynamic background modeling in video | |
CN112417931A (en) | Method for detecting and classifying water surface objects based on visual saliency | |
CN111222514B (en) | Local map optimization method based on visual positioning | |
CN106570888A (en) | Target tracking method based on FAST (Features from Accelerated Segment Test) corner point and pyramid KLT (Kanade-Lucas-Tomasi) | |
CN114638846A (en) | Pickup pose information determination method, pickup pose information determination device, pickup pose information determination equipment and computer readable medium | |
CN109284759A (en) | One kind being based on the magic square color identification method of support vector machines (svm) | |
CN107169991A (en) | A kind of moving target detecting method of multilayer background model | |
Suryawibawa et al. | Herbs recognition based on android using opencv | |
CN111862030B (en) | Face synthetic image detection method and device, electronic equipment and storage medium | |
Singh et al. | Copy move forgery detection on digital images | |
US20200294186A1 (en) | Method of plane tracking | |
Fadl et al. | Copy-rotate-move forgery detection based on spatial domain |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170915 |
|
WW01 | Invention patent application withdrawn after publication |