CN106254864B - Snowflake and noise noise detecting method in monitor video - Google Patents
Snowflake and noise noise detecting method in monitor video Download PDFInfo
- Publication number
- CN106254864B CN106254864B CN201610872016.0A CN201610872016A CN106254864B CN 106254864 B CN106254864 B CN 106254864B CN 201610872016 A CN201610872016 A CN 201610872016A CN 106254864 B CN106254864 B CN 106254864B
- Authority
- CN
- China
- Prior art keywords
- noise
- ratio
- yardstick
- piecemeal
- snowflake
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/02—Diagnosis, testing or measuring for television systems or their details for colour television signals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
- G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
-
- 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
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
Abstract
The present invention relates to the snowflake in a kind of monitor video and noise noise detecting method.The present invention makes the difference frame processing and eliminates fixed background influence first, and carries out maximum variance between clusters and obtain optimal binary-state threshold, optimal threshold binaryzation difference frame figure is reused, to highlight interference characteristic；Secondly it is the statistic in small area statistics zonule to divide equally bianry image；The method that snow noise or noise noise are judged whether finally by the stationarity of statistic.This method can detect the image containing snow noise or approaches uniformity distribution noise noise, well adapting to property.The present invention carries out statistics based on real scene sample data and determines judgment threshold, and snow noise and noise noise measuring rate are high, and real-time is good.
Description
Technical field
The invention mainly relates to video image quality diagnostic field, snow noise in more particularly to a kind of video image and
Noise noise method for detecting abnormality, the noise noise measuring being distributed suitable for snow noise and approaches uniformity.
Background technology
With becoming increasingly popular for video monitoring, the abnormal conditions occurred in video monitoring also rapid growth.Video is approximate
The noise and snow noise of even distribution are exactly two kinds in exception, and the presence of interference can seriously affect the identifiability of image.
Even original information is lost completely.Therefore find that this exception is just particularly important in time.Obviously in substantial amounts of video face
The mode of preceding artificial detection has been unable to meet demand；And the cost of human input also more and more higher, it is not easy to system administration.
The detection method of current snow noise and noise noise has：Zhang Wei, Fu Songlin et al. exist《One kind is based on convolutional Neural
The image noise detection method of network》(the patent No.：201410215084.0) in propose by collecting sample image and according to making an uproar
Vertex type carries out artificial mark classification, and these sample images input convolutional neural networks system is carried out to the instruction of disaggregated model
Practice, and also the sample image block of classification error is collected in assorting process and carries out relearning classification, from there through people
The mode that work and machine coordinate is labeled classification noise, is finally reached testing goal.Luo Tao, height wait people quietly and existed《Based on minimum
The noise detection method that local mean square deviation calculates》(the patent No.：201510688993.0) in propose one kind according to neighborhood of pixel points
Interior local mean square deviation is with removing the local detection of size of the difference progress of the local mean square deviation of itself and judging whether the point is noise
The method of point.Shi Zaifeng, Zhou Jiahui et al. exist《Image denoising method based on the detection of secondary noise》(the patent No.：
201510757953.7) in propose the Adaptive Second noise point detecting method based on directional information.Wan Chen, Yang Bo《It is a kind of
Based on improved four directional operators video noise detection side》(the patent No.：201210428662.X) in propose according to four directions calculate
Son is scanned to image, obtains the minimum value of four directional operator central values, and records preservation minimum value, and then obtains the frame figure
The number of pixel as in smooth region, finally judge that the two field picture whether there is the method for noise.It is He Qing, cold refined etc.
People exists《There are the system and method for snow noise in a kind of monitoring video》(201410636977.2) propose that one kind is based in
The detection method of target image signal to noise ratio.
The content of the invention
The purpose of the present invention is in view of the shortcomings of the prior art, there is provided the present invention proposes the snowflake in a kind of monitor video
With noise noise detecting method.Frame processing is made the difference first and eliminates fixed background influence, and is carried out maximum variance between clusters and obtained most
Good binary-state threshold, optimal threshold binaryzation difference frame figure is reused, to highlight interference characteristic；Secondly it is cell to divide equally bianry image
Statistic in the statistics zonule of domain；Snow noise or noise noise are judged whether finally by the stationarity of statistic
Method.This method can detect the image containing snow noise or approaches uniformity distribution noise noise, well adapting to property.
Technical scheme step is as follows：
Step 1：Obtain target video, extract video image it is wide with it is high.
Step 2：The picture frame in video sequence is extracted, and the storage format of picture frame is converted into single channel by multichannel
Gray level image storage format.
Step 3：The grayscale format image sequence of step 2 is taken turns doing into front and rear frame frame difference processing, obtains error image sequence.
Step 4：The error image sequence of step 3 is carried out into maximum variance between clusters to handle to obtain optimal binary-state threshold,
And binaryzation is carried out to poor frame figure with the threshold value, obtain binaryzation difference frame figure.
Step 5：The binaryzation difference frame figure of step 4 is subjected to piecemeal operation, the yardstick of multiple piecemeal is different.
Binaryzation difference frame figure is divided into multiple zonules by 5-1. under a kind of piecemeal yardstick, and calculates each zonule block
The ratio I of the interior zonule block number without connected region and zonule block total number；
5-2. calculates the number and the ratio of the zonule block area for the pixel that pixel value in each zonule block is 1
II, form the area ratio vector that length is zonule number with ratio II.
5-3. is directed to different piecemeal yardsticks, and ratio I, ratio II and area ratio vector is calculated.
Step 6：For different piecemeal yardsticks, its standard deviation and average are calculated by area ratio vector respectively, so as to
Obtain the ratio III of standard deviation and average；Every kind of piecemeal yardstick has its corresponding ratio III；By the ratio under different piecemeal yardsticks
Value III forms ratio vector；Compared with by the vectorial threshold decision vector formed with different piecemeal yardsticks of ratio, the poor frame is judged
With the presence or absence of snow noise or noise noise.
Described threshold decision vector is obtained by many experiments.
The present invention has the beneficial effect that：
The present invention is directed to snow noise and noise noise problem in monitor video, it is proposed that is removed fix by poor frame first
Background influence, so as to reach prominent snow noise or approaches uniformity distribution noise noise purpose；And pass through maximum variance between clusters
The mode for highlighting interference noise obtains being easy to the bianry image of analysis.Snow noise is drawn again by the mode of statistical sample
With the judgment threshold of noise noise measuring.The method for being finally reached detection snow noise and approaches uniformity distribution noise noise.This
Invention carries out statistics based on real scene sample data and determines judgment threshold, and snow noise and noise noise measuring rate are high, in real time
Property is good.
Brief description of the drawings
Fig. 1：Overall implementation process figure；
Fig. 2：Statistic analysis threshold figure；
Fig. 3：Binaryzation result figure；
Fig. 4：Data result figure；
Fig. 5：Data result figure；
Embodiment
Specific embodiments of the present invention are done below in conjunction with the accompanying drawings and are further described in more detail.
As Figure 1-5, the snowflake in monitor video and noise noise detecting method, it is intended to solve to exist in monitor video
Snow noise and approaches uniformity distribution noise noise abnormality detection problem.It is poor in the presence of snow noise and noise
Frame bianry image has the feature of obvious approaches uniformity distribution, and then using by calculating the standard deviation of data and the ratio of average
It is worth this statistic to judge the uniformity coefficient of data.The differentiation threshold value of uniformity coefficient is drawn finally by the mode of statistical sample
Judge whether snow noise or noise noise.Specific implementation process of the present invention is as shown in figure 1, comprise the following steps：
Step 1：Extraction detection video or video flowing first judge whether video is single channel gray scale frame sequence, if not single-pass
Road gray scale frame sequence, frame of video is converted into single channel gray level image, and extract the width of frame of video and height be designated as respectively W,
H。
Step 2：Front and rear frame frame difference processing is taken turns doing to the sequence of single channel gray level image, obtains error image sequence.It is right
Error image sequence carries out maximum variance between clusters and handles to obtain optimal binary-state threshold, and carries out two to poor frame figure with the threshold value
Value, obtain binaryzation difference frame figure.
Obtain binary image such as accompanying drawing 3.
Step 3：Binaryzation difference frame figure is subjected to piecemeal operation, the yardstick of multiple piecemeal is different, if after i-th kind etc. is divided yardstick
Each zonule block it is wide with it is high and be LiPixel, employs the different decile yardstick of N kinds altogether, the L under every kind of decile yardsticki
Value is i times of M, i.e. Li=M*i.Width, the height of wherein i-th kind etc. the zonule block divided under yardstick are designated as respectively：wi、hi。
Decile rule under every kind of yardstick is：wi=Li* W/ (W+H), hi=Li*H/(W+H)；
Wherein W, H are respectively the width and height of artwork；M is empirical value.N takes 17, M to take 20 in the present embodiment.
Binaryzation difference frame figure is divided into multiple zonules by 3-1. under a kind of piecemeal yardstick, and calculates each zonule block
The ratio I of the interior zonule block number without connected region and zonule block total number, is designated as L_counti,
3-2. calculates the number and the ratio of the zonule block area for the pixel that pixel value in each zonule block is 1
II, form the area ratio vector that length is zonule number with ratio II.The number of wherein pixel is designated as S_countikj,
Ratio II is designated as S_countikj/(wi*hi)；
3-3. is directed to different piecemeal yardsticks, and ratio I, ratio II and area ratio vector, specific meter is calculated
It is as follows to calculate formula：
Formula is as follows：
Wherein SikjFor the area accounting of (k, j) individual zonule under decile yardstick in i-th
Wherein miDivide the average of product accounting, V below yardstick for i-th kind etc.iFor the side for dividing product accounting below yardstick such as i-th kind
Difference, TiFor the ratio i.e. uniformity decision content for dividing scale calibration difference and average such as i-th kind.
Snow noise and noise noise corresponding data image such as accompanying drawing 4 are obtained by the step process.
Step 4：By uniformity decision content TiWith L_countiCompared respectively with the corresponding judgment threshold that it has been counted
Compared with due to snow noise and noise noise, approaches uniformity distribution is presented in its noise after step 2 processing.So obtained standard
The region unit number accounting without connected region that is smaller and obtaining of difference and average also can be very small or be approximately zero.Base
Snow noise or noise noise are judged whether in such conclusion.If Tmin<T2<Tmax、T17<TtAnd L_counti<LTWhen
It is determined as snow noise, if 0<T2<=Tmin、T17<TtAnd L_counti<LTWhen be determined as noise, other situations are determined as no snow
Flower and noise noise.
Wherein TminWhen for piecemeal yardstick being 40 or so pixel area than judgment threshold lower bound, TmaxIt is figure for piecemeal yardstick
During 1/8th pixel of image width degree and height sum area than the judgment threshold upper bound, LTFor the judgment threshold upper bound of ratio I；
T2For corresponding TminUniformity decision content under piecemeal yardstick, T17For corresponding TmaxUniformity decision content under piecemeal yardstick.The reality
Apply the T in exampleminTake 0.42, TmaxTake 1, LTTake 0.06.For the judgment threshold in other pixel size image embodiments according to
So applicable premise is that the width and height sum of the piecemeal such as acquirement minimum are in 20 pixels or so.
Accompanying drawing 2 is specifically described：
The sample of snow noise, noise noise and this non-two groups of situations under cut-and-dried multigroup fixed scene to be present
Video sequence, which is used to count, to be calculated.First three illustrated the step of accompanying drawing 1 is repeated several times in sample respectively, draws three groups of statistics
Data.One group is snow noise data be present, and one group is noise noise data be present, and another group is without snowflake and noise situation number
According to.Reliable differentiation threshold value is taken between three, brings sample to be tested inspection into.Part comparing result is shown in accompanying drawing 4 and Fig. 5.
Claims (5)
1. snowflake and noise noise detecting method in monitor video, it is characterised in that comprise the following steps：
Step 1：Obtain target video, extract video image it is wide with it is high；
Step 2：The picture frame in video sequence is extracted, and the storage format of picture frame is converted into single pass ash by multichannel
Spend image storage format；
Step 3：The grayscale format image sequence of step 2 is taken turns doing into front and rear frame frame difference processing, obtains error image sequence；
Step 4：The error image sequence of step 3 is carried out into maximum variance between clusters to handle to obtain optimal binary-state threshold, is used in combination
The threshold value carries out binaryzation to poor frame figure, obtains binaryzation difference frame figure；
Step 5：The binaryzation difference frame figure of step 4 is subjected to piecemeal operation, the yardstick of multiple piecemeal is different；If i-th kind of Equant ruler
After degree each zonule block it is wide with it is high and be LiPixel, employs the different decile yardstick of N kinds altogether, under every kind of decile yardstick
LiValue is i times of M, i.e. Li=M*i；M is empirical value；The width of wherein i-th kind etc. the zonule block divided under yardstick, high difference
It is designated as：wi、hi；
Step 6：For different piecemeal yardsticks, its standard deviation and average are calculated by area ratio vector respectively, so as to obtain
The ratio III of standard deviation and average；Every kind of piecemeal yardstick has its corresponding ratio III；By the ratio III under different piecemeal yardsticks
Form ratio vector；Compared with by the vectorial threshold decision vector formed with different piecemeal yardsticks of ratio, whether the poor frame is judged
Snow noise or noise noise be present；
Step 5 is specific as follows：
Binaryzation difference frame figure is divided into multiple zonules by 5-1. under a kind of piecemeal yardstick, and is calculated in each zonule block not
The ratio I of zonule block number and zonule block total number containing connected region, is designated as L_counti,
5-2. calculates the number and the ratio II of the zonule block area for the pixel that pixel value in each zonule block is 1, uses
Ratio II forms area ratio vector of the length for zonule number；
5-3. is directed to different piecemeal yardsticks, and ratio I, ratio II and area ratio vector is calculated；It is specific to calculate public affairs
Formula is as follows：
Formula is as follows：
Wherein SikjFor the area accounting of (k, j) individual zonule under decile yardstick in i-th, S_countikjFor of pixel
Number；
Wherein miDivide the average of product accounting, V below yardstick for i-th kind etc.iDivide the variance of product accounting, T below yardstick for i-th kind etc.i
For the ratio i.e. uniformity decision content for dividing scale calibration difference and average such as i-th kind.
2. snowflake and noise noise detecting method in monitor video according to claim 1, it is characterised in that every kind of chi
Degree under decile rule be：wi=Li* W/ (W+H), hi=Li*H/(W+H)；
Wherein W, H are respectively the width and height of artwork；M is empirical value.
3. snowflake and noise noise detecting method in monitor video according to claim 2, it is characterised in that step 5-2
The number of described pixel is designated as S_countikj, then ratio II be designated as S_countikj/(wi*hi)。
4. snowflake and noise noise detecting method in monitor video according to claim 3, it is characterised in that step 6
Specific judgement is as follows：
By uniformity decision content TiWith L_countiRespectively compared with its corresponding judgment threshold counted, if Tmin<T2<
Tmax、T17<TtAnd L_counti<LTWhen be determined as snow noise, if 0<T2<=Tmin、T17<TtAnd L_counti<LTWhen be determined as
Noise, other situations are determined as no snowflake and noise noise；
Wherein TminWhen for piecemeal yardstick being 40 or so pixel area than judgment threshold lower bound, TmaxIt is that image is wide for piecemeal yardstick
Degree and area during 1/8th pixel of height sum than the judgment threshold upper bound, LTFor the judgment threshold upper bound of ratio I；T2For
Corresponding TminUniformity decision content under piecemeal yardstick, T17For corresponding TmaxUniformity decision content under piecemeal yardstick.
5. snowflake and noise noise detecting method in monitor video according to claim 4, it is characterised in that when N takes
17th, when M takes 20, TminTake 0.42, TmaxTake 1, LT0.06 is taken, when the width and height sum for obtaining the piecemeals such as minimum are in 20 pixels
During left and right, TminAnd TmaxValue the judgment threshold of other pixel size images is still applicable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610872016.0A CN106254864B (en) | 2016-09-30 | 2016-09-30 | Snowflake and noise noise detecting method in monitor video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610872016.0A CN106254864B (en) | 2016-09-30 | 2016-09-30 | Snowflake and noise noise detecting method in monitor video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106254864A CN106254864A (en) | 2016-12-21 |
| CN106254864B true CN106254864B (en) | 2017-12-15 |
Family
ID=57612075
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610872016.0A Active CN106254864B (en) | 2016-09-30 | 2016-09-30 | Snowflake and noise noise detecting method in monitor video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106254864B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107453944B (en) * | 2017-07-07 | 2021-04-02 | 台州市吉吉知识产权运营有限公司 | Method and system for determining optimal test connection number of network throughput test |
| CN109547777B (en) * | 2018-11-06 | 2021-04-02 | 中国铁路上海局集团有限公司科学技术研究所 | Method for rapidly detecting video noise of complex scene |
| CN109859124B (en) * | 2019-01-11 | 2020-12-18 | 深圳奥比中光科技有限公司 | Depth image noise reduction method and device |
| CN112270317A (en) * | 2020-10-16 | 2021-01-26 | 西安工程大学 | Traditional digital water meter reading identification method based on deep learning and frame difference method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6741277B1 (en) * | 2000-01-13 | 2004-05-25 | Koninklijke Philips Electronics N.V. | System and method for automated testing of digital television receivers |
| CN101123681A (en) * | 2007-09-20 | 2008-02-13 | 宝利微电子系统控股公司 | A digital image noise reduction method and device |
| CN103095967A (en) * | 2011-10-28 | 2013-05-08 | 浙江大华技术股份有限公司 | Video noise quantization calculation method and video noise quantization calculation system |
| CN103377472A (en) * | 2012-04-13 | 2013-10-30 | 富士通株式会社 | Method for removing adhering noise and system |
| CN104469345A (en) * | 2014-12-10 | 2015-03-25 | 北京理工大学 | Video fault diagnosis method based on image processing |
-
2016
- 2016-09-30 CN CN201610872016.0A patent/CN106254864B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6741277B1 (en) * | 2000-01-13 | 2004-05-25 | Koninklijke Philips Electronics N.V. | System and method for automated testing of digital television receivers |
| CN101123681A (en) * | 2007-09-20 | 2008-02-13 | 宝利微电子系统控股公司 | A digital image noise reduction method and device |
| CN103095967A (en) * | 2011-10-28 | 2013-05-08 | 浙江大华技术股份有限公司 | Video noise quantization calculation method and video noise quantization calculation system |
| CN103377472A (en) * | 2012-04-13 | 2013-10-30 | 富士通株式会社 | Method for removing adhering noise and system |
| CN104469345A (en) * | 2014-12-10 | 2015-03-25 | 北京理工大学 | Video fault diagnosis method based on image processing |
Non-Patent Citations (1)
| Title |
|---|
| 《基于FPGA的视频图像雪花检测研究》;来晟;《信息通信》;20160115;第27-29页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106254864A (en) | 2016-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106254864B (en) | Snowflake and noise noise detecting method in monitor video | |
| CN104101600B (en) | Cross Section of CC Billet testing of small cracks method | |
| CN102842203B (en) | Method for monitoring bridge fault on basis of video image | |
| CN102855617B (en) | Method and system for processing adaptive images | |
| CN104268505A (en) | Automatic cloth defect point detection and recognition device and method based on machine vision | |
| CN109472261B (en) | Computer vision-based automatic monitoring method for grain storage quantity change of granary | |
| CN106934800A (en) | A kind of metal plate and belt detection method of surface flaw and device based on YOLO9000 networks | |
| CN112149543B (en) | Building dust recognition system and method based on computer vision | |
| CN107742307A (en) | Based on the transmission line galloping feature extraction and parameters analysis method for improving frame difference method | |
| CN105718964B (en) | A kind of visible detection method of power transmission line damper | |
| CN105427275A (en) | Filed environment wheat head counting method and device | |
| CN113706566B (en) | Edge detection-based perfuming and spraying performance detection method | |
| CN102621154B (en) | Method and device for automatically detecting cloth defects on line based on improved differential box multi-fractal algorithm | |
| CN113313677A (en) | Quality detection method for X-ray image of wound lithium battery | |
| CN117250208B (en) | Machine vision-based nano-imprint wafer defect accurate detection system and method | |
| CN108596196B (en) | Pollution state evaluation method based on insulator image feature dictionary | |
| CN114332081A (en) | Textile surface abnormity determination method based on image processing | |
| CN206951597U (en) | A kind of Blueberry hierarchical detection system based on machine vision | |
| CN109490776A (en) | A kind of good substandard products detection method of mobile phone vibrating motor based on machine learning | |
| CN109472779A (en) | A kind of yarn appearance characteristic parameter extraction and analysis method based on morphosis | |
| CN111882549A (en) | Automatic detection and identification method and system for grayish green small foreign fibers | |
| CN111914743A (en) | Method and device for detecting safety helmet of transformer substation worker | |
| CN110610136A (en) | Transformer substation equipment identification module and identification method based on deep learning | |
| CN113484867B (en) | Method for detecting density of fish shoal in closed space based on imaging sonar | |
| CN108982544A (en) | A kind of printed circuit board flaw component detection method |
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 |