CN105975911B - Energy-aware based on filter moves well-marked target detection method - Google Patents
Energy-aware based on filter moves well-marked target detection method Download PDFInfo
- Publication number
- CN105975911B CN105975911B CN201610272143.7A CN201610272143A CN105975911B CN 105975911 B CN105975911 B CN 105975911B CN 201610272143 A CN201610272143 A CN 201610272143A CN 105975911 B CN105975911 B CN 105975911B
- Authority
- CN
- China
- Prior art keywords
- energy
- target
- image frame
- frame
- video
- 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.)
- Expired - Fee Related
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 11
- 230000004927 fusion Effects 0.000 claims abstract description 8
- 238000013459 approach Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 6
- 238000004220 aggregation Methods 0.000 claims description 3
- 238000005267 amalgamation Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 230000002776 aggregation Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 claims 1
- 238000001914 filtration Methods 0.000 abstract description 12
- 230000000694 effects Effects 0.000 abstract description 3
- 230000003068 static effect Effects 0.000 description 9
- 238000013519 translation Methods 0.000 description 8
- 230000001154 acute effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000004297 night vision Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000004438 eyesight Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000031016 anaphase Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A kind of Energy-aware movement well-marked target detection method based on filter, this method content is: being the independent image frame sequence of YUV420 format by video-information decoding to be observed, resolution ratio is consistent with input video；Frame-skipping number is set as 0, reads in current image frame F；It is detected by motion feature and notable feature, obtains binary image frame F after Fusion Featuresd；To image Fd, carry out horizontally and vertically energy search, energy threshold and divide calculating, then carry out denoising smooth filtering processing respectively, obtain final movement well-marked target region；According to frame-skipping number, next frame picture frame is read as input, and circulation executes step 2, step 3, step 4, and the last frame until having read video terminates circulation, and realization terminates.The present invention can be effectively removed redundancy and false target, and the actual activity region of calibration movement well-marked target avoids artificial initialization marking area selected manually, realizes that motion target tracking provides possibility to be automatic.
Description
Technical field
The invention belongs to move well-marked target detection processing technical field, specifically a kind of energy based on filter
Perception movement well-marked target detection method.
Background technique
Well-marked target detection is moved as a very important research topic in computer vision field, is supervised in video
Control, Industry Control, robot vision and the navigation of autonomous vehicle etc. are all with a wide range of applications.In target detection
In technology, with the continuous variation of application field, application environment, demand also becomes increasingly complex, accuracy, reality to target detection
The requirement of Shi Xing, stability and portability are higher and higher.
Currently, target detection technique mainly positions target by feature detection and Fusion Features, due to application
Environment multiplicity, the factors such as the selection of feature, the complexity of detection method and amalgamation mode all can generate shadow to object detection results
It rings, feature detection and fused testing result usually contain redundancy, and the prior art is unable to satisfy application field to target detection
The performance requirement of technology.There are two types of the approach for usually improving target detection performance: an approach is started with from feature detection, in spy
When sign is extracted, more representative and targeted feature, the diversity of Lai Shiying target, or optimization feature are chosen
The method of detection technique such as extracts multiple local feature substitution global characteristics, expresses that the feature extracted more accurately
Target property；Another approach is started with from Fusion Features mode, usually introduces fuzzy theory or classifier cascade isotype is excellent
Change fusion results, to improve the accuracy of target detection.For the former, due to target multiplicity, and influenced by environmental change,
If Feature Selection is slightly improper, the accuracy of target detection just will affect.For the latter, although the introducing of complicated approach
It can be improved the accuracy of target detection, but method complexity can seriously affect the real-time of object detection method.
Filtering is the common technology of image procossing, is indispensable operation in image preprocessing and anaphase optimization,
Filter result directly influences picture quality and processing and analysis to image.Common filter have nonlinear filter,
Median filter and Morphologic filters etc., these filters are mainly used for the noise reduction of image, smooth or shape recognition, edge inspection
Survey etc..
Energy-aware filter used in the present invention, be by calculate video image in any one frame level with it is vertical
The summation of pixel energy can effectively obtain movement well-marked target profile after handling using signal smoothing filter energy
Region chooses the significant motion target area of vision or delimit moving target initial motion region to substitute original manual type
Process.
Summary of the invention
The present invention overcomes in existing target detection technique, there are redundancy or mistakes with fused testing result for feature detection
The problem of false information, provides a kind of Energy-aware movement well-marked target detection method based on filter, by moving significant mesh
Mark detection post-processing filter, more efficient, more accurately real-time target zone of action detection method.
Low by using complexity, traditional simple method carries out feature and detects and merge, and will test result through the present invention
The Energy-aware movement well-marked target detection method processing based on filter proposed, removal mistake or redundancy object, highlight
Moving region range provides necessary pretreatment for the methods of the calculating of succeeding target central point, color and local shape factor and ensures.
The validity and real-time of target detection technique are greatly strengthened in this way,
In order to solve above-mentioned technical problem, the present invention is achieved by the following technical solutions:
A kind of Energy-aware movement well-marked target detection method based on filter, this method content specifically include following step
It is rapid:
The video-information decoding to be observed of input is independent image frame sequence, by colour space transformation by the first step
At yuv space image, the resolution dimensions of the image frame sequence keep identical as original input video resolution ratio；
Second step, by essential characteristic detection, Fusion Features regular motion well-marked target detection technique, this uses light stream
Method and significant analytic approach extract motion feature and notable feature, using adaptive weighted amalgamation mode, extract in image frame sequence
Target, and obtain the binary image frame information of moving target region, which is determined by moving target region
Point coordinate is constituted；
Third step all carries out energy search meter horizontally and vertically to each frame in image frame sequence respectively
It calculates, remembers that the size of video image frame to be measured is n × m, the energy of note the i-th column of horizontal direction energy line is Note vertical direction jth row energy line energy be
4th step, according to the relative distance between camera lens and tested movement well-marked target, pre-estimation moving target
Energy magnitude range, determine horizontal direction pre-estimation energy threshold PTHX, { PTHX∈Z|PTHX>=0 } and vertical direction pre-estimation
Energy threshold PTHY, { PTHY∈Z|PTHY>=0 }, horizontal direction pre-estimation energy threshold PTHXWith vertical direction pre-estimation energy threshold
PTHYAccording to dollying head mirror head between measured object at a distance from determine, which is obtained by infrared sensor or laser sensor
?；
5th step, to horizontal direction energy PX,And vertical direction
Energy PY,It is smoothed to obtain smooth rear horizontal direction respectively
Energy aggregationWith vertical direction energy aggregation
Wherein, smooth is smoothing processing filter, is passed throughWithIntersection part obtainSo that it is determined that movement
Well-marked target zone boundary vertex, calibration movement well-marked target region.
Due to the adoption of the above technical scheme, a kind of Energy-aware based on filter provided by the invention moves well-marked target
Detection method, have compared with prior art it is such the utility model has the advantages that
Existing target detection technique is more demanding to the degree of stability of video camera mainly for static background, can not be very
It is applied to dynamic background situation well, it is affected by environment larger.Meanwhile existing optimization method generallys use and introduces classifier and learn
Habit mechanism improves accuracy, considerably increases calculation amount in this way, reduces the real-time of detection, can not continue tracking target, and
And it is not easy to apply in the small device of low-power consumption.
The present invention extracts feature using motion feature detection method and notable feature analytic approach, reduces computation complexity, mentions
High autonomous detectability reduces artificial participate in.In the case where camera is static or uniform speed motion state, by by aforementioned two kinds of features from
Weighted Fusion mode is adapted to, target and background information is efficiently separated, redundancy can be effectively removed using Energy-aware filter
Information and false target, the actual activity region of calibration movement well-marked target, obtain more accurate object detection results, avoid
The zone of action range of artificial initial motion well-marked target selected manually, can be used as the pretreatment skill of moving target automatic tracking
Art.
Since the present invention is in actual operation, mainly with image interframe luminance difference, image time-frequency domain conversation and energy
Based on perception search etc. calculates, without carrying out the complex calculations such as learning training, calculating process complexity is low, convenient for quickly obtaining in time
Take movement well-marked target zone of action, sustainable tracking target.It is of less demanding to the motion state of video camera itself, suitable for taking the photograph
The case where camera is static or uniform translation, uses in the inviolent indoor scene of illumination variation, can rapid detection it is single, multiple
Moving target, meanwhile, the motion state of target is required without too many, target at the uniform velocity, speed change and the case where discontinuous fluid
Under, movement well-marked target zone of action can be accurately detected.The present invention can be effectively removed redundancy and mistake
Target, the actual activity region of calibration movement well-marked target avoid artificial initialization marking area selected manually, realize to be automatic
Motion target tracking provides may.Strong real-time, computation complexity is low, can be efficiently applied to the power consumptions such as embedded, mobile
It is required that stringent small device.
Detailed description of the invention
Fig. 1 is the design principle logic chart of Energy-aware filter；
When Fig. 2 is that video camera is static, the movement well-marked target of single goal movement detects result of implementation schematic diagram；
When Fig. 3 is that video camera is static, the movement well-marked target of two targets movement detects result of implementation schematic diagram；
When Fig. 4 is that video camera is static, the movement well-marked target of single goal variable motion detects result of implementation schematic diagram；
When Fig. 5 is camera translation, the movement well-marked target of single goal movement detects result of implementation schematic diagram；
When Fig. 6 is camera translation, the movement well-marked target of three targets movement detects result of implementation schematic diagram.
Specific embodiment
Present invention is further described in detail with specific embodiment with reference to the accompanying drawing:
A kind of Energy-aware movement well-marked target detection method based on filter, the design of Energy-aware filter are former
Logic chart is managed as shown in Figure 1, the specific implementation steps of the method are as follows:
Step 1: being the independent image frame sequence of YUV420 format by video-information decoding to be observed, and resolution ratio and input regard
Frequency is consistent；
Step 2: frame-skipping number is set as 0, reads in current image frame F；
Step 3: being detected by motion feature and notable feature, obtains binary image frame F after Fusion Featuresd；
Step 4: to image Fd, carry out horizontally and vertically energy search, energy threshold and divide calculating, then distinguish
It is denoised, the disposal of gentle filter, obtains final movement well-marked target region
Step 5: according to frame-skipping number, reading next frame picture frame as input, and circulation executes step 2, step 3, step 4,
Last frame until having read video, terminates circulation, and realization terminates.
Embodiment 1:
Video camera is static, single goal motion conditions
The present embodiment applies the present invention under video camera stationary state, the movement well-marked target of single target movement is examined
It surveys.With this condition, video camera installs and fixes at the top of Mr. Yu robot or tripod, level shooting, in the visual field of camera lens
Interior, a human target enters into camera coverage according to the speed of 0.6m/s from the distant to the near.This video is mainly for indoor scene
Or the no motion of situation of shooting background still includes the indoor common furniture object such as window, desk, chair, chest in background
Product, unrelated with character costume, clothing, during shooting, acute variation does not occur for illuminance, and it is micro- that the present embodiment is not related to night vision etc.
Light particular surroundings.
Embodiment parameter declaration: video format avi, 15 frame of video frame number, video image size 320 × 240.Pre-estimation threshold
Value PTHX=70, PTHY=15.
The present embodiment is by taking arbitrary frame is handled as an example, as a result as shown in Fig. 2, Fig. 2-(1) is decoded input picture frame, warp
Motion feature detection and notable feature detection are crossed, motion feature binary picture and notable feature binary picture are respectively obtained, is such as schemed
Shown in 2- (2) and 2- (3)；The binary image frame information of movement well-marked target region is obtained after adaptively merging,
As shown in Fig. 2-(4)；Horizontally and vertically energy search, energy are carried out to binary image frame using Energy-aware filtering
It measures threshold value and divides calculating, then carry out denoising smooth filtering processing respectively, it is final to obtain movement well-marked target region, such as Fig. 2-(5)
It is shown.
Embodiment 2:
Video camera is static, two target motion conditions
The present embodiment applies the present invention under video camera stationary state, the movement well-marked target of two target movements is examined
It surveys.With this condition, video camera installs and fixes at the top of Mr. Yu robot or tripod, level shooting, in the visual field of camera lens
Interior, two human targets are respectively according to the speed of 0.6m/s and 0.8m/s, from the near to the distant far from camera direction.This video is main
For indoor scene or the no motion of situation of shooting background, still, the interiors such as window, desk, chair, chest are included in background
Common article of furniture, unrelated with character costume, clothing, during shooting, acute variation does not occur for illuminance, and the present embodiment does not relate to
And the low-lights particular surroundings such as night vision.Embodiment parameter declaration: video format avi, 30 frame of video frame number, video image size 320
×240.Pre-estimation threshold value PTHX=70, PTHY=15.
The present embodiment is by taking arbitrary frame is handled as an example, as a result as shown in figure 3, Fig. 3-(1) is decoded input picture frame, warp
Motion feature detection and notable feature detection are crossed, motion feature binary picture and notable feature binary picture are respectively obtained, is such as schemed
Shown in 3- (2) and 3- (3)；The binary image frame information of movement well-marked target region is obtained after adaptively merging,
As shown in Fig. 3-(4)；Horizontally and vertically energy search, energy are carried out to binary image frame using Energy-aware filtering
It measures threshold value and divides calculating, then carry out denoising smooth filtering processing respectively, it is final to obtain movement well-marked target region, such as Fig. 3-(5)
It is shown.
Embodiment 3:
Video camera is static, single goal variable motion situation
The present embodiment applies the present invention under video camera stationary state, the movement well-marked target of single target variable motion
Detection.With this condition, video camera installs and fixes at the top of Mr. Yu robot or tripod, level shooting, in the view of camera lens
Field, single human target is first moved according to 0.6m/s speed respectively, and gradually speed change, is moved in any direction.This video master
It to be directed to indoor scene or the no motion of situation of shooting background, still, the rooms such as window, desk, chair, chest are included in background
Interior common article of furniture, unrelated with character costume, clothing, during shooting, acute variation does not occur for illuminance, and the present embodiment is not
It is related to the low-lights particular surroundings such as night vision.
Embodiment parameter declaration: video format avi, 100 frame of video frame number, video image size 320 × 240, pre-estimation
Threshold value PTHX=70, PTHY=15.
The present embodiment is by taking arbitrary frame is handled as an example, as a result as shown in figure 4, Fig. 4-(1) is decoded input picture frame, warp
Motion feature detection and notable feature detection are crossed, motion feature binary picture and notable feature binary picture are respectively obtained, is such as schemed
Shown in 4- (2) and 4- (3)；The binary image frame information of movement well-marked target region is obtained after adaptively merging,
As shown in Fig. 4-(4)；Horizontally and vertically energy search, energy are carried out to binary image frame using Energy-aware filtering
It measures threshold value and divides calculating, then carry out denoising smooth filtering processing respectively, it is final to obtain movement well-marked target region, such as Fig. 4-(5)
It is shown.
Embodiment 4:
Camera translation, single goal motion conditions
The present embodiment applies the present invention under video camera moving condition, the movement well-marked target of single target movement is examined
It surveys.With this condition, video camera takes hand-held, is translated with the speed constant level of 0.5m/s, outside the visual field of camera lens, one
Human target enters into camera coverage according to the speed of 0.6m/s from the distant to the near.This video mainly for video camera uniform translation,
Background, there is a situation where moving, still, includes the indoor common furniture such as window, desk, chair, chest relative to camera lens in background
Article, unrelated with character costume, clothing, during shooting, acute variation does not occur for illuminance, and the present embodiment is not related to night vision etc.
Low-light particular surroundings.
Embodiment parameter declaration: video format avi, 30 frame of video frame number, video image size 320 × 240, pre-estimation threshold
Value PTHX=70, PTHY=15.
The present embodiment is by taking arbitrary frame is handled as an example, as a result as shown in figure 5, Fig. 5-(1) is decoded input picture frame, warp
Motion feature detection and notable feature detection are crossed, motion feature binary picture and notable feature binary picture are respectively obtained, is such as schemed
Shown in 5- (2) and 5- (3)；The binary image frame information of movement well-marked target region is obtained after adaptively merging,
As shown in Fig. 5-(4)；Horizontally and vertically energy search, energy are carried out to binary image frame using Energy-aware filtering
It measures threshold value and divides calculating, then carry out denoising smooth filtering processing respectively, it is final to obtain movement well-marked target region, such as Fig. 5-(5)
It is shown.
Embodiment 5:
Camera translation, three target motion conditions
The present embodiment applies the present invention under video camera moving condition, the movement well-marked target of three target movements is examined
It surveys.With this condition, video camera takes hand-held, with the speed uniform translation of 0.5m/s, outside the visual field of camera lens, and three personages
Target keeps a direction movement according to the speed of 0.6m/s, 0.8m/s and 1.2m/s.This video mainly for video camera at the uniform velocity
Translation, background relative to camera lens occur movement and target be it is multiple, still, in background include window, desk, chair,
The indoor common article of furniture such as chest, unrelated with character costume, clothing, during shooting, acute variation does not occur for illuminance, this
Embodiment is not related to the low-lights particular surroundings such as night vision.
Embodiment parameter declaration: video format avi, 50 frame of video frame number, video image size 320 × 240, pre-estimation threshold
Value PTHX=70, PTHY=15.
The present embodiment is by taking arbitrary frame is handled as an example, as a result as shown in fig. 6, Fig. 6-(1) is decoded input picture frame, warp
Motion feature detection and notable feature detection are crossed, motion feature binary picture and notable feature binary picture are respectively obtained, is such as schemed
Shown in 6- (2) and 6- (3)；The binary image frame information of movement well-marked target region is obtained after adaptively merging,
As shown in Fig. 6-(4)；Horizontally and vertically energy search, energy are carried out to binary image frame using Energy-aware filtering
It measures threshold value and divides calculating, then carry out denoising smooth filtering processing respectively, it is final to obtain movement well-marked target region, such as Fig. 6-(5)
It is shown.
Claims (1)
1. a kind of Energy-aware based on filter moves well-marked target detection method, this method content specifically includes following step
It is rapid:
The video-information decoding to be observed of input is independent image frame sequence, by colour space transformation at YUV by the first step
Spatial image, the resolution dimensions of the image frame sequence keep identical as original input video resolution ratio；
Second step, by essential characteristic detection, Fusion Features regular motion well-marked target detection technique, this using optical flow method with
Significant analytic approach extracts motion feature and notable feature, using adaptive weighted amalgamation mode, extracts the mesh in image frame sequence
Mark, and the binary image frame information of moving target region is obtained, which is sat by the fixed point of moving target region
Mark is constituted；
Third step, the energy search all carried out respectively to each frame in image frame sequence horizontally and vertically calculate,
The size for remembering video image frame to be measured is n × m, and the energy of note the i-th column of horizontal direction energy line is Note vertical direction jth row energy line energy be
4th step, according to the relative distance between camera lens and tested movement well-marked target, the energy of pre-estimation moving target
Magnitude range is measured, determines horizontal direction pre-estimation energy threshold PTHX, { PTHX∈Z|PTHX>=0 } and vertical direction pre-estimation energy
Threshold value PTHY, { PTHY∈Z|PTHY>=0 }, horizontal direction pre-estimation energy threshold PTHXWith vertical direction pre-estimation energy threshold PTHY
According to dollying head mirror head between measured object at a distance from determine, which is obtained by infrared sensor or laser sensor；
5th step, to horizontal direction energy PX,With vertical direction energy
PY,It is smoothed to obtain smooth rear horizontal direction energy respectively
SetWith vertical direction energy aggregation
Wherein, smooth is smoothing processing filter, is passed throughWithIntersection part obtainSo that it is determined that movement is significant
Target area boundaries vertex, calibration movement well-marked target region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610272143.7A CN105975911B (en) | 2016-04-28 | 2016-04-28 | Energy-aware based on filter moves well-marked target detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610272143.7A CN105975911B (en) | 2016-04-28 | 2016-04-28 | Energy-aware based on filter moves well-marked target detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105975911A CN105975911A (en) | 2016-09-28 |
| CN105975911B true CN105975911B (en) | 2019-04-19 |
Family
ID=56994115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610272143.7A Expired - Fee Related CN105975911B (en) | 2016-04-28 | 2016-04-28 | Energy-aware based on filter moves well-marked target detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105975911B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107992875B (en) * | 2017-12-25 | 2018-10-26 | 北京航空航天大学 | A kind of well-marked target detection method based on image bandpass filtering |
| CN108961313B (en) * | 2018-06-29 | 2021-06-29 | 大连民族大学 | Overlooking pedestrian risk quantification method of two-dimensional world coordinate system |
| CN112101075B (en) * | 2019-06-18 | 2022-03-25 | 腾讯科技（深圳）有限公司 | Information implantation area identification method and device, storage medium and electronic equipment |
| CN112699832B (en) * | 2021-01-12 | 2023-07-04 | 腾讯科技（深圳）有限公司 | Target detection method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714256A (en) * | 2009-11-13 | 2010-05-26 | 河北工业大学 | Omnibearing vision based method for identifying and positioning dynamic target |
| CN103747240A (en) * | 2013-12-25 | 2014-04-23 | 浙江大学 | Fusion color and motion information vision saliency filtering method |
| CN105427292A (en) * | 2015-11-11 | 2016-03-23 | 南京邮电大学 | Salient object detection method based on video |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8965048B2 (en) * | 2011-09-02 | 2015-02-24 | Audience Entertainment, Llc | Heuristic motion detection methods and systems for interactive applications |
-
2016
- 2016-04-28 CN CN201610272143.7A patent/CN105975911B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714256A (en) * | 2009-11-13 | 2010-05-26 | 河北工业大学 | Omnibearing vision based method for identifying and positioning dynamic target |
| CN103747240A (en) * | 2013-12-25 | 2014-04-23 | 浙江大学 | Fusion color and motion information vision saliency filtering method |
| CN105427292A (en) * | 2015-11-11 | 2016-03-23 | 南京邮电大学 | Salient object detection method based on video |
Non-Patent Citations (2)
| Title |
|---|
| Video Saliency Detection via Dynamic Consistent Spatio-Temporal Attention Modelling;Sheng-hua Zhong et al.;《Twenty-seventh Aaai Conference on Artificial Intelligence》;20130718;第1063-1069页 |
| 基于光流特征的运动目标检测与跟踪算法研究;秦岳;《中国优秀硕士学位论文全文数据库 信息科技辑》;20140315;第2014年卷(第03期);第I138-610页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105975911A (en) | 2016-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105718878B (en) | The aerial hand-written and aerial exchange method in the first visual angle based on concatenated convolutional neural network | |
| CN105975911B (en) | Energy-aware based on filter moves well-marked target detection method | |
| CN109858415A (en) | The nuclear phase followed suitable for mobile robot pedestrian closes filtered target tracking | |
| CN104821010A (en) | Binocular-vision-based real-time extraction method and system for three-dimensional hand information | |
| CN106204658A (en) | Moving image tracking and device | |
| CN104574357B (en) | The system and method for datum mark of the positioning with known figure | |
| Lian et al. | A novel method on moving-objects detection based on background subtraction and three frames differencing | |
| CN103105924A (en) | Man-machine interaction method and device | |
| Xiao et al. | A new tracking approach for visible and infrared sequences based on tracking-before-fusion | |
| Tu et al. | A biologically inspired vision-based approach for detecting multiple moving objects in complex outdoor scenes | |
| Li et al. | Integrated registration and occlusion handling based on deep learning for augmented-reality-assisted assembly instruction | |
| Fang et al. | 3d human pose estimation using rgbd camera | |
| CN110349176A (en) | Method for tracking target and system based on triple convolutional networks and perception interference in learning | |
| CN106780505B (en) | Super-pixel well-marked target detection method based on region energy | |
| Zhang et al. | A robust visual odometry based on RGB-D camera in dynamic indoor environments | |
| CN110490165B (en) | Dynamic gesture tracking method based on convolutional neural network | |
| Dai et al. | Federated filter algorithm with positioning technique based on 3D sensor | |
| Lei et al. | Improved motion information-based infrared dim target tracking algorithms | |
| Kalutskiy et al. | Method of moving object detection from mobile vision system | |
| Karthik et al. | Localized Harris-FAST interest point detector | |
| Hoang | Object size measurement and camera distance evaluation for electronic components using Fixed-Position camera | |
| Wibowo et al. | Multi-scale color features based on correlation filter for visual tracking | |
| Wang et al. | Camera Handoff for Multicamera Multiobject Tracking. | |
| Zhao et al. | High-altitude parabolic identification technology for urban buildings | |
| CN105184811B (en) | Multi-scale target tracking method based on structure support vector machine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190419 Termination date: 20200428 |