CN102810203A - System and method for detecting unknown article and event based on high-definition video monitoring image - Google Patents
System and method for detecting unknown article and event based on high-definition video monitoring image Download PDFInfo
- Publication number
- CN102810203A CN102810203A CN2012102226427A CN201210222642A CN102810203A CN 102810203 A CN102810203 A CN 102810203A CN 2012102226427 A CN2012102226427 A CN 2012102226427A CN 201210222642 A CN201210222642 A CN 201210222642A CN 102810203 A CN102810203 A CN 102810203A
- Authority
- CN
- China
- Prior art keywords
- image
- clear
- article
- video monitoring
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to a system and a method for detecting an unknown article and event based on a high-definition video monitoring image. The system consists of a high-definition video monitoring image enhancement processing module, an unknown article rapid detection positioning module, a grayscale image binarization processing module, an unknown article image gradient correction module, an unknown article real-time state monitoring and event characteristic extraction module and an unknown article and event detection and credibility evaluation module. Compared with the prior art, the system has the advantages of good effect, wide dynamic range, high unknown article rapid positioning and intelligent analysis (event detection) precision, low false alarm rate and the like aiming at the unknown article rapid positioning and intelligent analysis (event detection) under a complex high-definition video monitoring image at more than 2 million pixels and high-definition video monitoring image enhancement processing under non-uniform illumination conditions; and meanwhile, due to embedded software design, the credibility, reliability and stability of the software can be greatly improved, and the parameter setting difficulty of a front-end embedded system (an intelligent camera) is reduced.
Description
Technical field
The present invention relates to a kind of can be used under public safety, parking management, intelligent transportation, the comprehensive complicated monitoring scene in field such as emergent based on the not clear article of HD video monitoring image location, event detecting method and embedded software fast.
Background technology
In recent years, high-definition camera is applied to public safety, parking management, intelligent transportation more and more, comprehensive field complex background HD video down such as emergent is monitored.Not clear article location and event detecting method based on traditional SD video monitoring image can't be converted into to the HD video monitoring image under the complex scene; Main cause is that the above HD video monitoring image resolution of 2,000,000 pixels is high; Monitoring scene is big, image background is complicated; It is big to set quick location of the not clear article of detection area and event detection difficulty, and rate of false alarm is high.There are a plurality of not clear article need locate and combine field management demand decision event type of detection simultaneously in the HD video monitoring image under the complex scene, and existing not clear article location and the general simple target that only is directed against under the simple background of event detecting method based on the SD video monitoring image.In addition, the HD video monitoring image is set in the detection area not clear article location and event detecting method fast, requires not only that accuracy rate is high, rate of false alarm is low, and requires locating speed fast.Therefore how in HD video monitoring image (setting detection area), locating not clear article quickly and accurately and implement event detection, is generally to face and problem demanding prompt solution in the existing Intellectual Analysis Technology.
Summary of the invention
The present invention is directed to the prior art defective, provide a kind of not clear article based on the HD video monitoring image to locate fast and event detecting method, the image reinforced effects is good, and locating speed is fast, and the event detection accuracy rate is high.
For realizing above-mentioned purpose; The invention provides not clear article event detection system based on the HD video monitoring image, this system by HD video monitoring image enhancement process module, set the not clear article fast detecting locating module of detection area, set detection area gray level image binary conversion treatment module, not clear images of items degree of tilt correction module, not clear article real-time state monitoring and affair character extraction module and not clear article event detection and reliability assessment module and form;
Wherein, Said HD video monitoring image enhancement process module is used for the digital enhancement process of doing to the real-time images acquired of front end high definition video monitoring video camera; Confirm image filter template type and weighting coefficient in real time according to background/scene and content/target image in the HD video monitoring image; Obtain optimum signal-noise ratio and strengthen image, improve follow-up not clear article location and event detection precision;
Said setting detection area fails to understand that article fast detecting locating module is used for realizing setting comparison in real time between detection area and dynamic background image, adopts self-adaption gradient detection and location algorithm to failing to understand that wherein article carry out fast detecting, location;
It is the binary conversion treatment behind the gray level image that said setting detection area gray level image binary conversion treatment module is used to realize setting the detection area image transitions;
Said not clear images of items degree of tilt correction module is used for realizing setting the not clear images of items degree of tilt treatment for correcting of detection area based on the not clear images of items degree of tilt correcting algorithm of the long statistics of vertical runs;
The not clear images of items of binary image that said not clear article real-time status and affair character extraction module are used for being implemented in after the image inclination degree is proofreaied and correct is cut apart and feature extraction;
Said not clear article event detection and reliability assessment module adopt the dynamic sample clustering methodology to realize failing to understand identification of images of items coupling and local optimization, the adjustment of failing to understand feature samples in the images of items sample storehouse.
Another object of the present invention provides the not clear article event detection system based on the HD video monitoring image, it is characterized in that, specifically may further comprise the steps:
Step 1; The HD video monitoring image that HD video monitoring image enhancement process module is gathered high-definition camera in real time carries out enhancement process; Detection area image and dynamic background image are set in comparison in real time, carry out fast detecting, location to setting the not clear article of detection area in the HD video monitoring image of real-time collection;
Step 2; Setting the not clear article fast detecting locating module of detection area is mapped to real-time detected setting detection area in the original HD video monitoring image setting detection area dynamic background image through the not clear images of items that step 1 was handled; Detect interfering components in the not clear article according to presetting the eliminating of alarm threshold and field management demand; The not clear article state of monitoring is in real time classified to not clear article affair character;
Step 3, setting detection area gray level image binary conversion treatment module is gray level image with setting the detection area image transitions in the sorted HD video monitoring image, and carries out binary conversion treatment to setting the detection area gray level image in the HD video monitoring image;
Step 4, not clear article real-time status and affair character extraction module will be failed to understand the article real-time status and manage the sample library template and compare in real time, confirm to set the not clear article affair character of detection area and incident/type of detection;
Step 5, real-time learning, the not clear article affair character in optimum management sample storehouse and incident (detection) type, real-time assessment is set not clear article incident (detection) the type confidence level of detection area.
Further, said method step 5 adopts the dynamic sample clustering methodology that not clear article affair character and event detection credible result degree are assessed.
The not clear article of high definition (more than 2,000,000 pixels) video monitoring image that the present invention is directed under the complex background are located and the event detection demand fast; Credible, the reliability of HD video monitoring image enhancement process and software under the embedded software function design realization inhomogeneous illumination condition; Not clear article fast detecting, locating speed are fast; The adaptation dynamic range is big, the event detection precision is high; Improve front end embedded system (intelligent camera) reliability and stability, reduce on-site parameters requirement and difficulty are set.
Description of drawings
Fig. 1 is not clear article, the event detecting method FB(flow block) that the present invention is based on the HD video monitoring image;
Fig. 2 is not clear article, the event detection system logic diagram that the present invention is based on the HD video monitoring image.
Embodiment
Through accompanying drawing and embodiment, technical scheme of the present invention is done further detailed description below.
Fig. 1 is that the not clear article that the present invention is based on the HD video monitoring image are located and the event detecting method process flow diagram fast, and as shown in the figure, the concrete technic relization scheme of the present invention comprises the steps:
Step 505; Real-time learning, the not clear article affair character in optimum management sample storehouse and incident (detection) type (adopting the not clear article real-time status and the affair character sample of self study mode and evaluation algorithm formation according to qualifications), real-time assessment is set not clear article incident (detection) the type confidence level of detection area.
Described step 505 adopts the dynamic sample clustering methodology that not clear article real-time status, affair character and event detection credible result degree are assessed.
Not clear article, the event detection system that the present invention is based on the HD video monitoring image as shown in Figure 2, this system is made up of HD video monitoring image enhancement process module, not clear article fast detecting locating module, gray level image binary conversion treatment module, not clear images of items degree of tilt correction module, not clear article real-time state monitoring and affair character extraction module and not clear article event detection and reliability assessment module;
HD video monitoring image enhancement process module
In the HD video supervisory system; Installation, parameter setting and the environmental factor of on-the-spot HD video rig camera; Cause the HD video monitoring image of real-time collection to degenerate (anamorphose, noise covering etc.), with the not clear article location and the event detection precision that directly influence based on the HD video monitoring image.
" HD video monitoring image enhancement process " module realizes the digital enhancement process to the real-time images acquired of front end high definition video monitoring video camera; Confirm image filter template type (adaptive wiener filter) and weighting coefficient in real time according to background (scene) in the HD video monitoring image and content (target); Obtain optimum signal-noise ratio and strengthen image, improve not clear article location and event detection precision.
Not clear article fast detecting locating module
The HD video monitoring image that on-the-spot HD video rig camera is gathered in real time; Resolution is (1920x1080) more than 2,000,000 pixels; The real-time treatment capacity of entire image is bigger, will directly influence the video image processing speed, thus the system that influences follow-up not clear article location and event detection real-time.
In the real-time images acquired of on-the-spot HD video rig camera; According to system for field monitoring scene and related application regulatory requirement; The effective coverage that can be used for quick location of not clear article and event detection; Be the regional area in the HD video monitoring image, can be provided with, to improve HD video monitoring image (effective coverage) processing speed and not clear article location and event detection real-time and validity according to on-site supervision scene and application management demand." set detection area and fail to understand article fast detecting location " module and realize that image is compared in real time between setting detection area and dynamic background, adopt self-adaption gradient detection and location algorithm failing to understand that wherein article carry out fast detecting, location.
The self-adaption gradient detection and location algorithm that adopts can detect a plurality of not clear article simultaneously; And not clear article incident is mated identification according to not clear article status flag; According to setting not clear article real-time status and characteristics of image in the detection area; The present invention adopts pyramid algorith to decompose (subimage) to setting the not clear article real-time status of detection area; Tracking detects not clear article real-time status and characteristics of image; To fail to understand that at last sample matees in article real-time status and characteristics of image and the local sample storehouse (coupling thresholding and accuracy of detection are set), and not clear article real-time status and characteristics of image will be classified that final definite not clear type of items also (self study) is optimized sample image in the local sample storehouse.The not clear images of items coefficient of dissociation of level, vertical direction that the present invention sets pyramid algorith is respectively γ
x(<1.0) and γ
y(<1.0), the first order is decomposed by not clear images of items I horizontal direction of original detection area and vertical direction difference convergent-divergent γ
xAnd γ
yDoubly, obtain first order pyramid image I
1, again by I
1The not clear images of items of horizontal direction and vertical direction is convergent-divergent γ respectively
xAnd γ
yDoubly, obtain second level pyramid image I
2..., the rest may be inferred can do N (N=I, 2,3 ...) and level decomposes (according to not clear number of articles, accuracy of identification and field management requirements set).Generally get γ
x=0.5, γ
y=0.5 so that improve to set the not clear images of items decomposition rate of detection area, setting detection area destination number to be identified≤20 o'clock, and General N gets 2.
Each grade pyramid diagram is looked like to carry out color space transformation (converting gray level image into); To reduce to set detection area not clear article detection, location and event detection processing operations amount; For the setting detection area image of accomplishing color space transformation; Adopt not clear images of items vertical edge (profile) feature extraction algorithm to fail to understand article fast detecting, location, the speed that is characterized in is fast, rate of false alarm is low.
Gray level image binary conversion treatment module
Binary conversion treatment is the basis of Digital Image Processing, also is the important step of setting the not clear article identification of detection area image.Gray level image carries out binary conversion treatment more easily than coloured image; The present invention carries out binary conversion treatment immediately after will setting the detection area image transitions to be gray level image; The image binaryzation disposal route is a lot; The present invention is directed to the field management demand and set the not clear images of items characteristics of detection area, detection (location) precision is selected; In setting the not clear article location algorithm of detection area,, then adopt histogram method to set the detection area image binaryzation and handle if ambient lighting is even and contrast is stronger; When if ambient lighting is inhomogeneous; Then can't directly adopt histogram method to set the detection area image binaryzation handles; The setting detection area image binaryzation that adopts setting detection area image segmentation and gray level logic level technology GLLT (Gray Logical Level Technique) algorithm effectively to solve under the even low contrast condition of uneven illumination is handled---according to setting the detection area image and failing to understand the images of items characteristics; To set the detection area image division is plurality of sub-regions (according to regulatory requirement and destination number), and in all subregion image, not clear images of items is carried out fast detecting, location and successor and detect processing.The GLLT algorithm flow is following:
1). (x, y) ((x y) is its level and smooth back gray-scale value to g for x, y) some gray-scale value in the detection area image in order to set to establish f.According to setting not clear images of items template W (generally getting W=3) in the detection area, to fail to understand images of items center calculation (2W+l) * (2W+l) template window average gray:
2). (x, y) 8 adjacent pixels points of W pixel are P to set detection area image mid-range objectives pixel
0, P
1... P
7(x is y) than its 4 adjacent pixels P as if g
i, P
(i+4) mod8, P
(i+1) mod8, P
(i+5) mod8(i=0,1,2,3) high T gray level, then (x y) is divided into " white pixel " (value 255); (x is y) than its 4 adjacent pixels P as if g
i, P
(i+4) mod8, P
(i+1) mod8, P
(i+5) mod8T gray level hanged down in (i=0,1,2,3), and then (x y) is divided into " black pixel " (value 128); Otherwise this pixel is labeled as " unfiled pixel " (value 0).Decision rule is:
Wherein, H (P) is true, if
For very, if
Pixel P '
iAnd P '
I+1Be respectively pixel P
iAnd P
I+1(i=0,1,2,3) are over against (180 ° of directions) pixel.
3). the corresponding average gray image value G of pixel of value 255 and 128 is calculated in the subregion
1And G
2
4). is that 0 unfiled pixel is classified by following rule to value:
GLLT algorithm adaptability is strong, speed fast, strong robustness, does not need the complicated parameter setting, the more difficult problem identificatioin of threshold value in the time of can effectively solving such as the histogram method binary conversion treatment.
Not clear images of items degree of tilt correction module
Not clear images of items degree of tilt is proofreaied and correct directly influence not clear article location and event detection precision; And degree of tilt correction proportion in whole not clear article location and event detection flow process is bigger; Therefore, not clear images of items degree of tilt correcting algorithm efficient is based on the emphasis and the difficult point of the not clear article incident Detection Algorithm of HD video monitoring image.In not clear images of items, mainly contain two types of horizontal tilt and vertical banks.The present invention is based on the not clear images of items degree of tilt correcting algorithm (horizontal tilt degree correcting algorithm is similar) of the long statistics of vertical runs.
1). find out not clear images of items frame coordinate, x
0, x
1, y
0, y
iAnd calculate its centre coordinate (x
c, y
c);
2). establishing the boundary position off-set value is D
k, then (x y) is displaced to (x to pixel
s, y
s):
y
s=y;
3). to given D
kThe displacement diagram picture, calculate vertical direction black and white run length quadratic sum;
4). establish D
MaxBe boundary position peak excursion value, to [D in the interval
Max,+D
Max] in arbitrary integer off-set value D
k, calculate vertical direction black and white run length quadratic sum, find out maximal value, then its corresponding displacement diagram of institute looks like to be the image after the degree of tilt correction.
Not clear article status monitoring and affair character extraction module
It is in the binary image after not clear images of items degree of tilt is proofreaied and correct that not clear images of items is cut apart, and adopts vertical projection method to mark off not clear article subimage and interrelated relation thereof.This module difficult point is that not clear images of items adhesion (comprising the adhesion between not clear article and the background) and noise image disturb, through failing to understand the images of items prediction and estimating to instruct cutting apart of not clear images of items adhesion and noise image.
Not clear article event detection and reliability assessment module
Adopt the dynamic sample clustering methodology that not clear images of items is mated identification; Through with not clear images of items (characteristic) sample storehouse, this locality in feature samples compare (the training self-adaptation through to these samples obtains weight coefficient); Indexs such as not clear images of items (characteristic) sample size, degree of fitting are assessed; Optimize, adjust not clear images of items (characteristic) sample data; Adopt the mode of learning and the evaluation algorithm of selecting the superior to form not clear article affair character sample, to improve not clear article event detection precision.
The present invention has to the not clear article of high definition under the complex background (more than 2,000,000 pixels) video monitoring image location and inhomogeneous illumination condition hypograph enhancement process is effective, dynamic range big, target localization and event detection precision advantages of higher fast; Embedded Software Design can improve software trust, fiduciary level and stability greatly simultaneously, reduces front end embedded system parameter requirement is set.
The professional also can further recognize; In conjunction with each exemplary module of embodiment disclosed herein and algorithm steps; Can electronic hardware, computer software or the two combination realize; This paper has pressed general each illustrative functions and the realization flow described of correlation function, and these functions realize in which way, depend on related art scheme application and design constraint.The professional and technical personnel can use the different technologies implementation to realize institute's representation function to application-specific, but this realization should not thought and exceeds the scope of the invention.
In conjunction with embodiment method disclosed herein or technic relization scheme, can software, embedded software, hardware or the mode of combining carry out.Software module can place any other stored in form medium known in random access memory (RAM), internal memory, ROM (read-only memory) (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or the technical field.The above is based on the pedestrian's event detecting method and the embedded software technic relization scheme of HD video monitoring image; The object of the invention, technical scheme and beneficial effect have been carried out further explain, and institute it should be understood that the above is merely the concrete technic relization scheme of the present invention; And be not used in qualification protection scope of the present invention; All within spirit of the present invention and principle, any modification of being made, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (3)
1. based on not clear article, the event detection system of HD video monitoring image; It is characterized in that this system is made up of HD video monitoring image enhancement process module, not clear article fast detecting locating module, gray level image binary conversion treatment module, not clear images of items degree of tilt correction module, not clear article real-time state monitoring and affair character extraction module and not clear article event detection and reliability assessment module;
Wherein, Said HD video monitoring image enhancement process module is used for the digital enhancement process of doing to the real-time images acquired of front end high definition video monitoring video camera; Confirm image filter template type and weighting coefficient in real time according to background/scene and content/target image in the HD video monitoring image; Obtain optimum signal-noise ratio and strengthen image, improve follow-up not clear article location and event detection precision;
Said not clear article fast detecting locating module is used for realizing setting comparison in real time between detection area and dynamic background image, adopts self-adaption gradient detection and location algorithm to failing to understand that wherein article carry out fast detecting, location;
It is the binary conversion treatment behind the gray level image that said gray level image binary conversion treatment module is used to realize setting the detection area image transitions;
Said not clear images of items degree of tilt correction module is used for realizing setting the not clear images of items degree of tilt treatment for correcting of detection area based on the not clear images of items degree of tilt correcting algorithm of the long statistics of vertical runs;
The not clear images of items of binary image that said not clear article real-time status and affair character extraction module are used for being implemented in after the image inclination degree is proofreaied and correct is cut apart and feature extraction;
Said not clear article event detection and reliability assessment module adopt the dynamic sample clustering methodology to realize failing to understand identification of images of items coupling and local optimization, the adjustment of failing to understand feature samples in images of items/feature samples storehouse.
2. based on not clear article, the event detecting method of HD video monitoring image, it is characterized in that, specifically may further comprise the steps:
Step 1; The HD video monitoring image that HD video monitoring image enhancement process module is gathered high-definition camera in real time carries out enhancement process; Detection area image and dynamic background image are set in comparison in real time, carry out fast detecting, location to setting the not clear article of detection area in the HD video monitoring image of real-time collection;
Step 2; Setting the not clear article fast detecting locating module of detection area is mapped to real-time detected setting detection area in the original HD video monitoring image setting detection area dynamic background image through the not clear images of items that step 1 was handled; Detect interfering components in the not clear article according to presetting the eliminating of alarm threshold and field management demand; The not clear article state of monitoring is in real time classified to not clear article affair character;
Step 3, setting detection area gray level image binary conversion treatment module is gray level image with setting the detection area image transitions in the sorted HD video monitoring image, and carries out binary conversion treatment to setting the detection area gray level image in the HD video monitoring image;
Step 4, not clear article real-time status and affair character extraction module will be failed to understand the article real-time status and manage the sample library template and compare in real time, confirm to set the not clear article affair character of detection area and incident/type of detection;
Step 5, real-time learning, the not clear article affair character in optimum management sample storehouse and incident (detection) type, real-time assessment is set not clear article incident (detection) the type confidence level of detection area.
3. not clear article, event detecting method based on the HD video monitoring image according to claim 2; It is characterized in that said method step 5 adopts the dynamic sample clustering methodology that not clear article affair character and event detection credible result degree are assessed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012102226427A CN102810203A (en) | 2012-06-29 | 2012-06-29 | System and method for detecting unknown article and event based on high-definition video monitoring image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012102226427A CN102810203A (en) | 2012-06-29 | 2012-06-29 | System and method for detecting unknown article and event based on high-definition video monitoring image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102810203A true CN102810203A (en) | 2012-12-05 |
Family
ID=47233905
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2012102226427A Pending CN102810203A (en) | 2012-06-29 | 2012-06-29 | System and method for detecting unknown article and event based on high-definition video monitoring image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102810203A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105761504A (en) * | 2016-05-17 | 2016-07-13 | 重庆大学 | Vehicle speed real-time measuring method based on inhomogeneous video image frame collection |
CN113409547A (en) * | 2021-06-10 | 2021-09-17 | 维沃移动通信有限公司 | Alarm method, alarm device, electronic equipment and readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101026685A (en) * | 2007-03-23 | 2007-08-29 | 北京中星微电子有限公司 | Static object detecting method and system |
CN101123721A (en) * | 2007-09-30 | 2008-02-13 | 湖北东润科技有限公司 | An intelligent video monitoring system and its monitoring method |
CN101221623A (en) * | 2008-01-30 | 2008-07-16 | 北京中星微电子有限公司 | Object type on-line training and recognizing method and system thereof |
CN101630360A (en) * | 2008-07-14 | 2010-01-20 | 上海分维智能科技有限公司 | Method for identifying license plate in high-definition image |
CN101833859A (en) * | 2010-05-14 | 2010-09-15 | 山东大学 | Self-triggering license plate identification method based on virtual coil |
-
2012
- 2012-06-29 CN CN2012102226427A patent/CN102810203A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101026685A (en) * | 2007-03-23 | 2007-08-29 | 北京中星微电子有限公司 | Static object detecting method and system |
CN101123721A (en) * | 2007-09-30 | 2008-02-13 | 湖北东润科技有限公司 | An intelligent video monitoring system and its monitoring method |
CN101221623A (en) * | 2008-01-30 | 2008-07-16 | 北京中星微电子有限公司 | Object type on-line training and recognizing method and system thereof |
CN101630360A (en) * | 2008-07-14 | 2010-01-20 | 上海分维智能科技有限公司 | Method for identifying license plate in high-definition image |
CN101833859A (en) * | 2010-05-14 | 2010-09-15 | 山东大学 | Self-triggering license plate identification method based on virtual coil |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105761504A (en) * | 2016-05-17 | 2016-07-13 | 重庆大学 | Vehicle speed real-time measuring method based on inhomogeneous video image frame collection |
CN105761504B (en) * | 2016-05-17 | 2018-02-09 | 重庆大学 | Speed method for real-time measurement based on the collection of non-homogeneous video frame image |
CN113409547A (en) * | 2021-06-10 | 2021-09-17 | 维沃移动通信有限公司 | Alarm method, alarm device, electronic equipment and readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102799862A (en) | System and method for pedestrian rapid positioning and event detection based on high definition video monitor image | |
CN102768731A (en) | Method and system for automatic positioning and identifying target based on high definition video images | |
Chen et al. | Ship detection from coastal surveillance videos via an ensemble Canny-Gaussian-morphology framework | |
CN101630360B (en) | Method for identifying license plate in high-definition image | |
US10198657B2 (en) | All-weather thermal-image pedestrian detection method | |
CN102902951A (en) | System and method for vehicle target location and event detection on basis of high-definition video monitoring images | |
CN105574855B (en) | Infrared small target detection method under cloud background based on template convolution and false alarm rejection | |
CN108021891B (en) | Vehicle environment identification method and system based on combination of deep learning and traditional algorithm | |
CN101937508A (en) | License plate localization and identification method based on high-definition image | |
CN106951898B (en) | Vehicle candidate area recommendation method and system and electronic equipment | |
Zheng et al. | Vehicle detection based on morphology from highway aerial images | |
CN113158738A (en) | Port environment target detection method, system, terminal and readable storage medium based on attention mechanism | |
CN104200466A (en) | Early warning method and camera | |
Li et al. | Road vehicle monitoring system based on intelligent visual internet of things | |
CN103530640A (en) | Unlicensed vehicle detection method based on AdaBoost and SVM (support vector machine) | |
CN104766079A (en) | Remote infrared weak object detecting method | |
CN110490150A (en) | A kind of automatic auditing system of picture violating the regulations and method based on vehicle retrieval | |
CN106446929B (en) | Type of vehicle detection method based on edge gradient potential energy | |
CN108269221B (en) | JPEG recompressed image tampering positioning method | |
CN104168462A (en) | Camera scene change detecting method based on image angular point set characteristic | |
CN102810203A (en) | System and method for detecting unknown article and event based on high-definition video monitoring image | |
CN108268866B (en) | Vehicle detection method and system | |
Huang et al. | Ship detection using texture statistics from optical satellite images | |
Chen et al. | An efficient algorithm on vehicle license plate location | |
CN116342644A (en) | Intelligent monitoring method and system suitable for coal yard |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C05 | Deemed withdrawal (patent law before 1993) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20121205 |