CN102215321B - Mobile detection method and device - Google Patents
Mobile detection method and device Download PDFInfo
- Publication number
- CN102215321B CN102215321B CN 201010145475 CN201010145475A CN102215321B CN 102215321 B CN102215321 B CN 102215321B CN 201010145475 CN201010145475 CN 201010145475 CN 201010145475 A CN201010145475 A CN 201010145475A CN 102215321 B CN102215321 B CN 102215321B
- Authority
- CN
- China
- Prior art keywords
- value
- pixel
- object pixel
- differences
- space
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to a mobile detection method comprising the following steps of: receiving a previous image and a current image; computing the pixel value difference of a target pixel in the current image and the previous image to judge a time pixel value difference; respectively computing the pixel value difference of the target pixel and surrounding pixels of the target pixel in the previous image and the current image to judge a space pixel value difference; and dividing the space pixel value difference by the time pixel value difference to judge a displacement value of the target pixel.
Description
Technical field
The present invention relates to a kind of movement detection method and device, especially relate to a kind of movement detection method and device that comes the mobile degree of estimating target pixel by time value differences and space value differences.
Background technology
Video may produce noise under various situation such as acquisition, video recording and transmission, and then influences the effect of its subsequent treatment.Therefore, noise removing is a very important task in Video processing.In general, the method for eliminating noise is divided into spatial noise and eliminates two kinds of (Spatial Noise Reduction) and time noise removing (Temporal Noise Reduction).On static picture, the relevance on the service time of the territory is come filtering noise purely, can keep more image detail than the image characteristics on the usage space; On the contrary, on the picture that moves, the relevance on the service time territory is come filtering noise, then can cause the situation of ghost, because image detail can be fuzzy slightly when mobile, the image characteristics on the usage space comes filtering noise, does not have too big sequelae on the contrary.Therefore, in order to want correct in noise filtering, video filter must be worked on spatial domain and time-domain.
Please refer to Fig. 1, Fig. 1 is the schematic diagram of a noise elimination apparatus 10.Noise elimination apparatus 10 is used for a processing system for video and eliminates noise, and it comprises a spatial filter 12, a time filter 14, moves a detecting unit 16 and an output 18.Spatial filter 12 is used for the pixel data of a current picture f (t) is carried out two-dimentional low-pass filtering, to eliminate the noise that has nothing to do with the time among the current picture f (t).Mobile detecting unit 16 is used for judging the degree of dynamism of current picture f (t), and it produces the mobile estimating value of current picture by the pixel data of more current picture f (t) picture f previous with it (t-1).According to the mobile estimating value, termporal filter 14 can carry out suitable filtering to the pixel data of current picture f (t), and f is " (t) as a result to produce a filtering.When picture when being dynamic, the two-dimentional low-pass filtering result that termporal filter 14 usage space filters 12 are exported, f is " (t) as a result as the filtering of current picture; And when picture when being static, then the pixel data to previous picture f (t-1) and current picture f (t) carries out a three-dimensional filtering computing, and f is " (t) as a result with the filtering that produces current picture.
In other words, mobile detecting unit 16 can be judged the degree of dynamism of picture, and termporal filter 14 then carries out suitable filtering according to the degree of dynamism of picture to current picture f (t).Thus, noise elimination apparatus 10 removes and can eliminate noise effectively, improves outside the quality of image, can keep the original definition of tableaux again.
As from the foregoing, prior art is when carrying out noise filtering to vision signal, needing to estimate the mobile degree of current picture, is to want usage space filtering result or time filtering result with the filtering result who determines current picture, or space filtering result and time filtering result's a certain mixed proportion.At present, the industry mobile detection technique that is used for eliminating noise is broadly divided into two kinds.A kind of is value differences by previous picture and current picture, the mobile degree of coming represent pixel.Yet the problem of this kind practice is when pixel is subjected to displacement, though adjacent pictures can produce value differences, have value differences may not represent visible pixel to move.Therefore, utilizing this mode to carry out mobile detection often has wrongheaded situation to take place.
Another kind is to use the mode that moves estimation (motion estimation) and motion compensation (motioncompensated) to carry out the filtering of noise on the time-domain as rule.Yet, motion compensation practice complexity, cost is higher, is only applicable to high-order and uses.
Summary of the invention
Therefore, main purpose of the present invention promptly is to provide a kind of movement detection method and device.
The present invention discloses a kind of movement detection method that is used for a processing system for video.This method includes reception one a previous picture and a current picture; Calculate an object pixel in the value differences of this current picture, to judge a time value differences with this previous picture; In this previous picture and this current picture, calculate the value differences of the surrounding pixel of this object pixel and this object pixel, respectively to judge a space value differences; And should the time value differences divided by this space value differences, to judge a shift value of this object pixel.
The present invention also discloses a kind of motion detection device that is used for a processing system for video, and it includes a first input end, one second input, a time difference computational unit, a space difference computational unit and a Displacement Estimation unit.This first input end and this second input receive a previous picture and a current picture respectively.This time difference computing unit is coupled to this first input end and this second input, is used for calculating an object pixel in the value differences of this current picture with this previous picture, to judge a time value differences.This space difference computational unit is coupled to this first input end and this second input, is used for reaching in this current picture at this previous picture respectively, calculates the value differences of the surrounding pixel of this object pixel and this object pixel, to judge a space value differences.This Displacement Estimation unit is coupled to this time difference computing unit and this space difference computational unit, is used for this time value differences divided by this space value differences, to judge a shift value of this object pixel.
Description of drawings
Fig. 1 is the schematic diagram of a known noise elimination apparatus.
Fig. 2 is the schematic diagram of the present invention's one motion detection device.
Fig. 3 has illustrated mobile detection mode of the present invention.
Fig. 4 is the embodiment schematic diagram of space difference computational unit among Fig. 3.
Fig. 5 and Fig. 6 are respectively the schematic diagram that the present invention one is used for carrying out normalized look-up table.
Fig. 7 moves the schematic diagram of testing process for the embodiment of the invention one.
Wherein, description of reference numerals is as follows:
10 noise elimination apparatus
12 spatial filters
14 termporal filters
16 move detecting unit
18 outputs
The current picture of f (t)
The previous picture of f (t-1)
F " (t) filtering result
20 motion detection devices
21,22 inputs
23 time difference computing units
24 space difference computational unit
25 Displacement Estimation unit
26,27 low pass filters
28 first normalization unit
29 second normalization unit
242 first difference computational unit
244 second difference computational unit
246 minimum values decision unit
TD time value differences
SD space value differences
The m_value shift value
Y (t), y (t-1) pixel value
MSD1, MSD2 max pixel value difference
MLV moves degree
The CLV confidence level
Th1
The th4 threshold value
70 move testing process
700
750 steps
Embodiment
Please refer to Fig. 2, Fig. 2 is the schematic diagram of the present invention's one motion detection device 20.Motion detection device 20 is used for the mobile degree that a processing system for video detects a current picture, and it includes input 21 and 22, a time difference computational unit 23, a space difference computational unit 24 and a Displacement Estimation unit 25. Input 21 and 22 is used for receiving a previous picture f (t-1) and a current picture f (t) respectively.Previous picture f (t-1) preferably was the last picture of current picture f (t), but in other embodiments, and it also can be a certain picture in previous a plurality of pictures of current picture f (t), and is not limited thereto.Time difference computing unit 23 is coupled to input 21 and 22, is used for calculating an object pixel in the value differences of current picture f (t) with previous picture f (t-1), to judge a time value differences TD.Space difference computational unit 24 is coupled to input 21 and 22, be used for calculating formerly among the picture f (t-1) and current picture f (t), the max pixel value difference of object pixel and its surrounding pixel (for example: 8 pixels) on every side is to judge a space value differences SD.Displacement Estimation unit 25 is coupled to time difference computing unit 23 and space difference computational unit 24, is used for time value differences TD divided by space value differences SD, to obtain a shift value m_value of object pixel.
In brief, motion detection device 20 comes moving of estimating target pixel by calculating the time value differences and the space value differences of object pixel.About above-mentioned mobile detection mode, please continue explanation with reference to figure 3.In Fig. 3, y (t) represents the pixel value of object pixel at current picture f (t), y (t-1) expression object pixel is the pixel value of picture f (t-1) formerly, solid line represent object pixel formerly among the picture f (t-1) along the pixel value curve of a specific direction, dotted line then represent object pixel in current picture f (t) along unidirectional pixel value curve.As seen from the figure, during from previous picture f (t-1) to current picture f (t), the pixel value curve has produced one by the displacement of solid line to dotted line, makes the pixel value of object pixel change to y (t) by y (t-1).Because in adjacent pictures, do not have too big change along unidirectional pixel value curve, therefore, if can learn the object pixel pixel curve gradient of picture f (t-1) or current picture f (t) value differences of a pixel (promptly every) formerly, add known time value differences (promptly | y (t)-y (t-1) |), the object pixel formerly displacement between picture f (t-1) and current picture f (t) just can be estimated, shown in the horizontal arrow among the figure.
Therefore, the present invention is respectively to previous picture f (t-1) and current picture f (t), and calculating object pixel and its be the max pixel value difference of 8 pixels on every side, to obtain the pixel curve gradient of two dimensional surface.Owing to can calculate a pixel curve gradient from previous picture f (t-1), and also can obtain a pixel curve gradient from current picture f (t), therefore, the embodiment of the invention is got the minimum value of the two, as possible pixel curve gradient (being the space value differences SD among Fig. 1), obtain maximum shift value m_value with expection.
On the other hand, by calculating object pixel, can obtain time value differences TD at the value differences of current picture f (t) and previous picture f (t-1) (promptly | y (t)-y (t-1) |).Therefore, divided by after the value differences SD of space, can estimate the displacement of object pixel to time value differences TD, this object pixel displacement just what pixels.Compared to prior art, though utilize value differences to move detection equally, the present invention not only can improve the accuracy of judgement degree, also can not increase the complexity of system.
Please refer to Fig. 4, Fig. 4 is an embodiment schematic diagram of space difference computational unit 24.Space difference computational unit 24 includes one first difference computational unit 242, one second difference computational unit 244 and minimum value decision unit 246.First difference computational unit 242 is coupled to input 21, is used for calculating the value differences of the middle object pixel of current picture f (t) and its surrounding pixel, to obtain the max pixel value difference MSD1 of current picture f (t).Second difference computational unit 244 is coupled to input 22, is used for calculating the value differences of the middle object pixel of previous picture f (t-1) and its surrounding pixel, to obtain the max pixel value difference MSD2 of previous picture f (t-1).246 of unit of minimum value decision are coupled to first difference computational unit 242 and second difference computational unit 244, be used for obtaining the minimum value among max pixel value difference MSD1 and the MSD2, and this minimum value is exported as space value differences SD.
Preferably, the shift value m_value that motion detection device 20 can further be exported displacement estimation unit 25 carries out normalization (normalize), to judge the mobile degree of object pixel.In addition, as long as space value differences SD or time value differences TD are enough big, can judge that the shift value m_value that Displacement Estimation unit 25 is exported is rational estimated value.Therefore, motion detection device 20 can further carry out normalization to the summation of space value differences SD and time value differences TD, with the foundation as confidence level.
In this case, motion detection device 20 also can include one first normalization unit 28 and one second normalization unit 29, as shown in Figure 2.The first normalization unit 28 can be shown in Figure 5 a look-up tables'implementation; As shift value m_value during greater than a threshold value th2, the mobile degree MLV that judges object pixel is 1, as shift value m_value during less than a threshold value th1, the mobile degree MLV that judges object pixel is 0, and when shift value m_value between threshold value th1 and threshold value th2, judge that then mobile degree MLV is the result of 0 and 1 linear interpolation.That is to say that when th1 of object pixel displacement below the pixel, motion detection device 20 is judged object pixels for mobile, and displacement th2 more than the pixel, it is mobile to judge that then object pixel has in adjacent pictures.
Similarly, the look-up tables'implementation that 29 of the second normalization unit can be shown in Figure 6; As the summation SUM of space value differences SD and time value differences TD during greater than a threshold value th4, judge that the confidence level of shift value m_value is 1, as summation SUM during, judge that the confidence level of shift value m_value is 0 less than a threshold value th3.That is to say, when space value differences SD and time value differences TD the two all little the time, judge that then resulting shift value m_value is insincere, or object pixel as not moving.
Thus, the mobile degree MLV and the confidence level CLV that calculate according to motion detection device 20, video system can be selected usage space filtering or time filtering, or space filtering result and time filtering result's a certain mixed proportion is eliminated the noise of object pixel.
Please refer to Fig. 7.Fig. 7 moves the schematic diagram of testing process 70 for one of the embodiment of the invention.Mobile testing process 70 is an operating process of above-mentioned motion detection device 20, and it includes the following step:
Step 700: beginning.
Step 710: receive previous picture f (t-1) and current picture f (t).
Step 720: calculate object pixel in the value differences of current picture f (t), with judgement time value differences TD with previous picture f (t-1).
Step 730: calculate respectively formerly among the picture f (t-1) and current picture f (t), the max pixel value difference of object pixel and its surrounding pixel is to judge space value differences SD.
Step 740: with time value differences TD divided by space value differences SD, to judge the shift value m_value of object pixel.
Step 750: finish.
According to flow process 70, motion detection device 20 at first calculates object pixel at the value differences of current picture f (t) and previous picture f (t-1) (promptly | y (t)-y (t-1) |), to obtain time value differences TD.Then, difference is picture f (t-1) and current picture f (t) formerly, and calculating object pixel and its be the max pixel value difference of 8 pixels on every side, to obtain the pixel curve gradient of two dimensional surface.At last, divided by space value differences SD, can estimate the shift value of object pixel to time value differences TD, this object pixel displacement just what pixels.The detailed operation mode of motion detection device 20 does not repeat them here in above-mentioned explanation.
In sum, the present invention comes moving of estimating target pixel mainly by the time value differences of adjacent pictures and the pixel slope of curve of space value differences representative.Compared to prior art, though utilize value differences to move detection equally, the present invention can improve the accuracy of judgement degree, and can significantly not increase the complexity of system.
The above only is the preferred embodiments of the present invention, and all equalizations of doing according to claim of the present invention change and modify, and all should belong to covering scope of the present invention.
Claims (10)
1. a movement detection method that is used for a processing system for video is characterized in that, includes:
Receive a previous picture and a current picture;
Calculate an object pixel in the value differences of this current picture, to judge a time value differences with this previous picture;
In this previous picture and this current picture, calculate the value differences of this object pixel and this object pixel surrounding pixel, respectively to judge a space value differences; And
Should the time value differences divided by this space value differences, to judge a shift value of this object pixel, wherein this shift value is a displacement pixel number.
2. movement detection method as claimed in claim 1 is characterized in that, obtains the step of this space value differences of this object pixel, further includes:
According to the value differences of this object pixel and this object pixel surrounding pixel in this previous picture, calculate a max pixel value difference of this previous picture;
According to the value differences of this object pixel and this object pixel surrounding pixel in this current picture, calculate a max pixel value difference of this current picture; And
This max pixel value difference of this max pixel value difference of this previous picture and this current picture relatively, obtaining a minimum value, and with this minimum value as this space value differences.
3. movement detection method as claimed in claim 1 is characterized in that, this method further includes:
According to a look-up table, this shift value is carried out normalization, and judge that one of this object pixel moves degree.
4. movement detection method as claimed in claim 3 is characterized in that this look-up table during less than a first threshold, judges that this mobile degree of this object pixel is 0 at this shift value; And during greater than one second threshold value, judge that this mobile degree of this object pixel is 1 at this shift value.
5. movement detection method as claimed in claim 1 is characterized in that, this method further includes:
According to a look-up table, a summation of this time value differences and this space value differences is carried out normalization, and judge a confidence level of this shift value, wherein during less than one the 3rd threshold value, this confidence level of judging this shift value is 0 to this look-up table in this summation; And in this summation during greater than one the 4th threshold value, this confidence level of judging this shift value is 1.
6. a motion detection device that is used for a processing system for video is characterized in that, includes:
One first input end is used for receiving a previous picture;
One second input is used for receiving a current picture;
One time difference computational unit is coupled to this first input end and this second input, is used for calculating an object pixel in the value differences of this current picture with this previous picture, to judge a time value differences;
One space difference computational unit is coupled to this first input end and this second input, is used for reaching in this current picture at this previous picture respectively, calculates the value differences of this object pixel and this object pixel surrounding pixel, to judge a space value differences; And
One Displacement Estimation unit, be coupled to this time difference computing unit and this space difference computational unit, be used for should the time value differences divided by this space value differences, to judge a shift value of this object pixel, wherein this shift value is a displacement pixel number.
7. motion detection device as claimed in claim 6 is characterized in that, this space difference computational unit further includes:
One first difference computational unit is coupled to this first input end, is used for calculating the value differences of this object pixel and this object pixel surrounding pixel in this previous picture, to obtain a max pixel value difference of this previous picture;
One second difference computational unit is coupled to this second input, is used for calculating the value differences of this object pixel and this object pixel surrounding pixel in this current picture, to obtain a max pixel value difference of this current picture; And
One minimum value decision unit, be coupled to this first difference computational unit and this second difference computational unit, with the minimum value in this max pixel value difference of this max pixel value difference that decides this previous picture and this current picture, and this minimum value exported as this space value differences.
8. motion detection device as claimed in claim 6 is characterized in that, this Displacement Estimation unit pack contains:
One first normalization unit is coupled to this time difference computing unit and this space difference computational unit, is used for this shift value being carried out normalization, and judging that one of this object pixel moves degree according to a look-up table.
9. motion detection device as claimed in claim 8 is characterized in that, this look-up table during less than a first threshold, judges that this mobile degree of this object pixel is 0 at this shift value; And during greater than one second threshold value, judge that this mobile degree of this object pixel is 1 at this shift value.
10. motion detection device as claimed in claim 6 is characterized in that, this Displacement Estimation unit pack contains:
One second normalization unit, be coupled to this time difference computing unit and this space difference computational unit, be used for according to a look-up table, a summation of this time value differences and this space value differences carried out normalization, and judge a confidence level of this shift value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010145475 CN102215321B (en) | 2010-04-08 | 2010-04-08 | Mobile detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010145475 CN102215321B (en) | 2010-04-08 | 2010-04-08 | Mobile detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102215321A CN102215321A (en) | 2011-10-12 |
| CN102215321B true CN102215321B (en) | 2013-07-24 |
Family
ID=44746435
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010145475 Expired - Fee Related CN102215321B (en) | 2010-04-08 | 2010-04-08 | Mobile detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102215321B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108647222B (en) * | 2018-03-22 | 2021-01-08 | 中国互联网络信息中心 | Line three-dimensional roaming hotspot icon positioning method and system |
| CN108764078B (en) * | 2018-05-15 | 2019-08-02 | 上海芯仑光电科技有限公司 | A kind of processing method and calculating equipment of event data stream |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5276512A (en) * | 1991-03-07 | 1994-01-04 | Matsushita Electric Industrial Co., Ltd. | Video signal motion detecting method and noise reducer utilizing the motion |
| CN1622613A (en) * | 2004-12-16 | 2005-06-01 | 上海交通大学 | Texture information based video image motion detecting method |
| CN1926881A (en) * | 2004-03-01 | 2007-03-07 | 索尼株式会社 | Motion vector detecting apparatus, motion vector detection method and computer program |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0757326B2 (en) * | 1989-06-19 | 1995-06-21 | 富士ゼロックス株式会社 | Pulverizer |
| JP3892059B2 (en) * | 1995-03-07 | 2007-03-14 | 松下電器産業株式会社 | Moving body tracking device |
| JPH09121356A (en) * | 1995-10-25 | 1997-05-06 | Oki Electric Ind Co Ltd | Motion vector detector |
-
2010
- 2010-04-08 CN CN 201010145475 patent/CN102215321B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5276512A (en) * | 1991-03-07 | 1994-01-04 | Matsushita Electric Industrial Co., Ltd. | Video signal motion detecting method and noise reducer utilizing the motion |
| CN1926881A (en) * | 2004-03-01 | 2007-03-07 | 索尼株式会社 | Motion vector detecting apparatus, motion vector detection method and computer program |
| CN1622613A (en) * | 2004-12-16 | 2005-06-01 | 上海交通大学 | Texture information based video image motion detecting method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102215321A (en) | 2011-10-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200005468A1 (en) | Method and system of event-driven object segmentation for image processing | |
| KR102605747B1 (en) | Video noise removal methods, devices, and computer-readable storage media | |
| US20120019728A1 (en) | Dynamic Illumination Compensation For Background Subtraction | |
| US8295607B1 (en) | Adaptive edge map threshold | |
| CN101262559A (en) | A method and device for eliminating sequential image noise | |
| US20110299597A1 (en) | Image processing method using motion estimation and image processing apparatus | |
| KR20060076176A (en) | A method of temporal noise reduction in video sequence ans system therefore | |
| EP3149940B1 (en) | Block-based static region detection for video processing | |
| US10607321B2 (en) | Adaptive sharpness enhancement control | |
| CN102752483A (en) | Filtering noise reduction system and filtering noise reduction method based on FPGA (field programmable gate array) platform | |
| KR102445762B1 (en) | Method and device for processing images | |
| EP4072142A3 (en) | Video motion processing including static scene determination, occlusion detection, frame rate conversion, and adjusting compression ratio | |
| KR101537559B1 (en) | Device for detecting object, device for detecting object for vehicle and method thereof | |
| CN107993254A (en) | Moving target detecting method based on disassociation frame calculus of finite differences | |
| TWI413023B (en) | Method and apparatus for motion detection | |
| CN102215321B (en) | Mobile detection method and device | |
| US9818178B1 (en) | Method and system for detection of ghosting artifact in a video | |
| CN101237523B (en) | Main edge detection method and noise reduction method and device based on this method | |
| US8189874B2 (en) | Method for motion detection of horizontal line | |
| US20130064295A1 (en) | Motion detection method and associated apparatus | |
| JP2009116686A (en) | Imaging target detection apparatus and method | |
| US9466094B1 (en) | Method to improve video quality under low light conditions | |
| JP4289170B2 (en) | Noise amount measuring apparatus and video receiver | |
| JP5091880B2 (en) | Moving image noise removing apparatus and moving image noise removing program | |
| GB2514557A (en) | Image processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| 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: 20130724 Termination date: 20160408 |