CN101489032A - Motion detection method and apparatus in brightness and chroma separation module of video codec - Google Patents
Motion detection method and apparatus in brightness and chroma separation module of video codec Download PDFInfo
- Publication number
- CN101489032A CN101489032A CNA2009100210691A CN200910021069A CN101489032A CN 101489032 A CN101489032 A CN 101489032A CN A2009100210691 A CNA2009100210691 A CN A2009100210691A CN 200910021069 A CN200910021069 A CN 200910021069A CN 101489032 A CN101489032 A CN 101489032A
- Authority
- CN
- China
- Prior art keywords
- pixel
- module
- motion detection
- brightness
- spatial coherence
- 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.)
- Granted
Links
Images
Landscapes
- Processing Of Color Television Signals (AREA)
- Color Television Systems (AREA)
Abstract
The invention discloses a motion diction method in a video decoder brightness and chrominance separation module and an apparatus for implementing the method. The motion detection apparatus combining with a space-time self-adapting luminance and chrominance separation reads a motion detected 4 field data from a frame memory, and performs the motion detection combining a field where the real signal is by using a 5 field signal. A space correlativity detection module detects a vertical correlativity of pixels under test. The apparatus of the invention dynamically selects the motion detection module according to a binary space correlativity detection result, finds out a module weighting average difference value, and judges the motion condition according to the relation of the difference value and the motion detection threshold. A time correlativity detection module detects the time correlativity of the pixels under test, and generates a binary time correlativity detection result. The result combining the space correlativity detection result and pixels forms a new data structure, and improves the video signal handling ability.
Description
Technical field
The invention belongs to digital video image and handle and the Display Technique field, be specifically related to method for testing motion and device in a kind of composite video signal decoder brightness and chroma separation module.
Background technology
The method that the composite video signal YC separates mainly is divided into to be utilized spatial coherence (YC separates in) and utilizes two kinds of separation methods of temporal correlation (YC separates between).The former can lose the high-frequency information of image, and subjective observation can be lost the detail section of image; The latter can not cause losing of image high-frequency information, if but use incorrectly, can cause the hangover and the diplopia of moving image.
Summary of the invention
At the hangover that causes moving image that exists in the prior art and the technical problem of diplopia, the present invention is according to the room and time correlation properties and the modular characteristics of vision signal, method for testing motion in a kind of brightness and chroma separation module of video codec is proposed, obtain motion detection result more accurately, a kind of device of realizing above-mentioned method for testing motion also is provided simultaneously.
The present invention utilizes the time and space correlated characteristic of image: in the space, pixel can be returned according to itself and the correlation of surrounding pixel point and be drawn different modules; On time, the module that correlation is stronger in the space is according to identical regular movement or keep static; For the content of unexpected variation, particularly change frequency video near field frequency, the perhaps video that constantly changes of the characteristics of motion, pixel is also variant to the correlation of front and back field.Utilize the time and the spatial coherence of pixel, obtained motion detection result accurately.Thereby select correct selection separation method for brightness and chroma separation module, obtain more excellent picture quality.
To achieve these goals, the present invention takes following technical solution:
Method for testing motion in a kind of brightness and chroma separation module of video codec, the spatial coherence that carries out pixel detects, generate the spatial coherence testing result of binaryzation, spatial coherence Dynamic Selection motion detection block according to pixel, the weighted average difference and the threshold ratio of module are carried out motion detection, simultaneously, carrying out temporal correlation detects, detect the correlation of pixel and field, front and back corresponding pixel points, generate the temporal correlation testing result of binaryzation, thereby determine the motion detection situation of this pixel, and the spatial coherence testing result of binaryzation and temporal correlation testing result and pixel are constituted new pixel data structure.
A kind of device of realizing method for testing motion in the brightness and chroma separation module of video codec, this device comprises:
One frame memory, preserve motion detection with between light tone separate the video data of using;
One spatial coherence detection module comprises the high accuracy trapper, and difference is asked for unit and comparator, is used to realize that the spatial coherence of pixel detects, and chooses for motion detection block reference is provided;
Trapper acts on the upper and lower related pixel point of an interior pixel to be detected and same column respectively, choosing of related pixel point is relevant with the standard of composite video signal, for the TSC-system signal, related pixel point selection same column phase two pixels before leaving, and, select respectively be separated by up and down two pixels of delegation of same column for pal signal;
Difference is asked for the luminance difference that pixel to be measured and related pixel point are asked in the unit;
The luminance difference isospace correlation detection threshold value that comparator will be tried to achieve before compares, generate the comparative result of binaryzation, as representing that with 1 difference is less than threshold value, be that correlation is good, represent that with 0 difference is less than threshold value, be that correlation is bad, the flag bit that produces 2 bit wides writes down the correlation information of pixel to be measured and relevant upper and lower pixel, and this result is as motion detection and follow-up important references of going interlacing, convergent-divergent to handle;
An interior adaption brightness chroma separation module, the result who detects according to spatial coherence finishes an interior adaptive YC separation;
One motion detection block choice of dynamical unit, spatial coherence testing result according to binaryzation is selected motion detection block dynamically, in conjunction with the phase relation of carrier chrominance signal in the composite video signal, in continuous several, choose related pixel point dynamically and constitute motion detection block;
One motion detection block, it comprises that difference asks for the unit, comparator and median filter;
Difference is asked for the unit and is asked for the weighted average difference in motion detection block inside, and this difference is that the YC integrated value of corresponding pixel points is poor, and weighted sum obtains;
Comparator compares weighted average difference and motion detection threshold, utilizes comparative result to judge the motion conditions of pixel;
Median filter carries out medium filtering to motion detection result to be handled, according to the modular characteristics of image, after medium filtering is handled motion detection result can reduce the erroneous judgement that noise causes, further improve the motion detection precision;
One time correlation detection module, it comprises that brightness asks for module choice of dynamical unit, and the unit is asked in brightness, and luminance difference is asked for unit and comparator;
Brightness is asked for dynamic module and is chosen the spatial coherence testing result of unit according to binaryzation, select YC separation scheme in the suitable field for the pixel of pixel and time correlation is dynamic, ask for the brightness value of the pixel of trying to achieve pixel and time correlation in the unit in brightness:
For in the space up and down correlation all well pixel select upper and lower reference point and these pixel 3 points of vertical direction, the light tone separation schemes are asked for the brightness value of this pixel in comprehensive 23 row, and are shown in Figure 4, K1=0.5, K2=0.5, K3=0;
2. select last reference point and these pixel 2 points of vertical direction for the pixel that upward correlation is good in the space, utilize 3 a row interior light tone separation scheme to ask for the brightness value of this pixel;
3. select following reference point and these pixel 2 points of vertical direction for the pixel that down correlation is good in the space, utilize light tone separation scheme in 3 row to ask for the brightness value of this pixel;
4. separate for light tone in the row of all bad pixel selection horizontal direction of upper and lower correlation in the space, ask for the brightness value of this pixel;
It is poor with the brightness value of the pixel of pixel and time correlation that luminance difference is asked for the unit; Comparator compares this difference respectively with the temporal correlation detection threshold, generate the testing result of binaryzation, and with testing result with 2 storages, remarked pixel point correlation in time;
The temporal correlation testing result of the spatial coherence testing result of binaryzation and binaryzation is integrated the temporal correlation of remarked pixel point, constitute new data structure with pixel, for the subsequent treatment module provides important references, comprehensively improve the Video processing ability.
Adaption brightness chroma separation module between one, the result who detects according to temporal correlation finishes adaptive YC separation between the field;
Characteristics of the present invention are to have taken into full account the modular characteristics of pixel in the vision signal, and in conjunction with the phase properties of carrier chrominance signal in the composite video signal, the time and space correlation choice of dynamical motion detection block and the temporal correlation detection module of comprehensive utilization pixel carry out respective detection, thereby have obtained the testing result of degree of precision; Can share field memory with 3D adaption brightness chroma separation module in addition on hardware is realized, thereby not need to increase extra hardware cost, the temporal correlation testing result also can be the separation of self adaptation light tone reference is provided simultaneously.
Compared with the prior art, technique effect of the present invention is embodied in:
1. consider the phase properties of carrier chrominance signal in the modular characteristics of pixel and the composite video signal, Dynamic Selection motion detection block and temporal correlation detection module;
2. spatial coherence detects and compares the brightness value that trapper is tried to achieve, and the trapper principle simply is easy to realization, and can obtain very high luminance level resolution, relatively is fit to carry out the spatial coherence detection of vertical direction, guarantees higher precision;
3. motion detection and temporal correlation detect the testing result all make full use of spatial coherence, have realized the pixel modularization and based on the target of temporal correlation dynamic process;
4. motion detection result is handled through median filter, can reduce the erroneous judgement that noise causes, has improved motion detection accuracy;
5. with the weighted average difference of motion detection block and temporal correlation detection module, choose by weights and obtained higher precision with each threshold ratio;
6. generate the spatial coherence and the temporal correlation testing result of binaryzation, and constitute new data structure, for the subsequent treatment of vision signal provides reference with pixel;
By with the light tone separating and combining of three-dimensional adaptive composite video signal decoder, this method does not need to increase extra hardware cost.
Proposed in a kind of in the technical solution of the present invention and the method that the adaption brightness chrominance separation is handled between the field, this method is distinguished the motion and standstill pixel according to motion detection result, the motion and standstill pixel is adopted different processing methods, thereby reach higher resolution and smooth more picture quality, this method key is based on motion detection accurately, has a strong impact on the decoding quality of composite video signal if judge inaccurate meeting.This motion detection algorithm is not only applicable to the composite video signal of general scene, and can obtain higher precision to the composite video signal of spectrum complex yet.Correlation (being spatial coherence) detects and the record testing result in the field of carrying out at first, inside; Secondly, according to the spatial coherence testing result in Dynamic Selection motion detection block between related pixel point between the field, then with threshold ratio, and then judgement pixel motion conditions with the weighted average difference between related pixel point in the motion detection; Once more, carry out according to relevance detection results in the field that correlation between the field (being the time correlation) detects and writing time correlation testing result.And, be the video scaling in the video reprocessing with room and time relevance detection results and the new data structure of pixel formation, go interlacing (mainly using the spatial coherence result) and field frequency to promote (mainly using the temporal correlation result) reference information is provided.
Description of drawings
Fig. 1 is the phase relation of related pixel point carrier chrominance signal in the Phase Alternation Line system;
Fig. 2 is the phase relation of related pixel point carrier chrominance signal in the TSC-system;
Fig. 3 is the spatial coherence schematic diagram of pixel;
Fig. 4 detects for spatial coherence and an interior adaption brightness chroma separation module schematic diagram;
Fig. 5 is the temporal correlation schematic diagram of pixel;
Fig. 6 is a motion detection dynamic module schematic diagram (1);
Fig. 7 is a motion detection dynamic module schematic diagram (2);
Fig. 8 is a motion detection dynamic module schematic diagram (3);
Fig. 9 is a motion detection dynamic module schematic diagram (4);
Figure 10 is a median filter shape schematic diagram;
Figure 11 is adaption brightness chrominance separation schematic diagram between temporal correlation detection and field;
Figure 12 is a video signal data structure of the present invention;
Figure 13 motion detection block structure chart;
Embodiment
The present invention is described in further detail below in conjunction with accompanying drawing.
Method for testing motion in a kind of brightness and chroma separation module of video codec, the spatial coherence that carries out pixel detects, generate the spatial coherence testing result of binaryzation, according to the spatial coherence Dynamic Selection motion detection block of pixel, the weighted average difference and the threshold ratio of module carried out motion detection.Simultaneously, carrying out temporal correlation detects, detect the correlation of pixel and field, front and back corresponding pixel points, generate the temporal correlation testing result of binaryzation, thereby determine the motion detection situation of this pixel, and the spatial coherence testing result of binaryzation and temporal correlation testing result and pixel constituted new pixel data structure, handling for postorder provides foundation.
A kind of device of realizing motion detection in the brightness and chroma separation module of video codec, this device comprises:
One frame memory be used for preserving motion detection with between light tone separate the video data of using;
One spatial coherence detection module, this module utilizes trapper that the vertical related pixel point with it of pixel is asked for brightness value successively, calculate luminance difference and compare with the spatial coherence detection threshold, obtain the spatial coherence of pixel, this relevance detection results can be separated for self adaptation light tone in the field provides dynamic weights, the motion detection of the binaryzation testing result of Sheng Chenging after being simultaneously, temporal correlation detection etc. basis for estimation is provided;
One motion detection block, this module is utilized spatial coherence testing result Dynamic Selection motion detection block, in module, ask for the weighted average difference, difference and motion detection threshold are compared, draw the motion conditions of pixel, testing result through median filter, is reduced erroneous judgement, improve accuracy of detection;
One time correlation detection module, spatial coherence testing result according to binaryzation, light tone separation scheme in the Dynamic Selection field, try to achieve the brightness value of pixel to be measured and time last front and back corresponding fields pixel successively, calculate luminance difference and relatively can draw the temporal correlation of pixel, and generate the temporal correlation testing result of binaryzation with the temporal correlation detection threshold;
The job step of described spatial coherence detection module is:
At first finishing an interior self adaptation light tone separates, utilize trapper earlier vertical related pixel point to be asked for brightness value respectively then, then that the brightness value of pixel is poor respectively with the brightness value of pairwise correlation pixel up and down, with difference difference isospace correlation detection threshold ratio, according to comparative result, generate the spatial coherence testing result [Pre2_line of binaryzation on the one hand, Next2_line], represented that respectively (Pre2_line represents the correlation with upper point for the spatial coherence of D1-D4 four class pixels, Next2_line represents the correlation with lower point, 1 expression difference is less than threshold value, and correlation is good; 0 expression correlation is bad); Separate each separation scheme for an interior self adaptation light tone on the other hand and generate dynamic weights, weighted average is separated final result as self adaptation light tone in the field.
The selection mode of described motion detection dynamic module is:
Through the detection of spatial coherence, will put mutually and be divided into 4 classes, use [Pre2_line, Next2_line]=00,10,01,11 expression respectively, corresponding to D1-D4 class point, choose corresponding motion detection block for 4 class testing results.
The specific implementation of described motion detection block is:
Carrying out O under the pal mode
N, i, jAdopt following formula to obtain the weighted average difference of dynamic module respectively successively during the point motion detection, represent with d:
When [Pre2_line, Next2_line]=[00],
d＝abs(O
n-2，i，j-O
n+2，i，j)
When [Pre2_line, Next2_line]=[10],
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i-2，j)+0.15·abs(O
n-1，i-2，j-O
n+1，i，j)
When [Pre2_line, Next2_line]=[01],
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i+2，j)+0.15·abs(O
n-1，i+2，j-O
n+1，i，j)
When [Pre2_line, Next2_1ine]=[11],
d＝0.6·abs(O
n-2，i，j-O
n+2，i，j)+0.2·abs(O
n-1，i-1，j-O
n+1，i+1，j)+0.2·abs(O
n-1，i+1，j-O
n+1，i-1，j)
With weighted average difference and motion detection threshold relatively, if greater than threshold value then think this pixel motion, otherwise think that it is static.
Described specific implementation of motion detection result being carried out the filter of medium filtering is:
Testing result is handled through median filter, and median filter selects to have utilized the spatial coherence of pixel, improves accuracy of detection, reduces the probability of erroneous judgement.
The specific implementation of described spatial coherence detection module is:
It comprises that brightness asks for dynamic module and choose the unit, and the unit is asked in brightness, and luminance difference is asked for unit and comparator;
Dynamic module is chosen the spatial coherence testing result of unit according to binaryzation, ask for the unit for the interior brightness in the suitable field of the respective pixel point selection on pixel and the time shaft and form dynamic module, utilize in the frame light tone separation scheme to ask for brightness value in module, the same space coordinate pixel of the time being gone up in two of the relevant front and back according to the standard of TV is selected identical dynamic module simultaneously;
Brightness is asked for the unit according to the dynamic module of selecting, and uses in the suitable field light tone separation scheme to ask for brightness value in module, utilizes identical light tone separation scheme to ask for brightness value to the respective modules of dependent field simultaneously;
Luminance difference is asked for the unit and brightness value that dynamic module obtains and time is asked in the brightness of pixel correspondence to go up the brightness value that the related pixel point obtains poor respectively;
Comparator compares this difference respectively with the spatial coherence detection threshold, generate the testing result of binaryzation, and testing result is stored remarked pixel point correlation in time with 2bit;
The data structure specific implementation of described new pixel is:
With spatial coherence testing result [Pre2_line, Next2_line], temporal correlation testing result [Pre2_Frame, Next2_Frame] and digitized pixel point value (can be forms such as RGB, YUV) combine can constitute new video signal data structure.
Referring to the phase relation that Figure 1 shows that pixel chromatic value in the Phase Alternation Line system 50Hz interlacing composite video signal.Tested point is the pixel A at center among the figure, and trunnion axis is a time shaft, the vertical direction in the vertical axis representation space.The face that their constitute perpendicular to the place on the plane.If the pixel of the capable j row of strange (FIELD_n_0DD) i in n field is expressed as O in the space
N, i, j, defining this chrominance phase is 0 °, can derive O according to colourity negative carrier frequency and line frequency relation
N, i ± 1, jPoint chrominance phase and O
N, i, jChrominance phase differs 90 °, O
N, i ± 2, jPoint chrominance phase and O
N, i, jChrominance phase differs 180 °, O
N, i ± 1, jPoint chrominance phase and O
N, i, jChrominance phase differs 90 °, O
N, i ± 2, jPoint chrominance phase and O
N, i, jChrominance phase differs 180 °, and the carrier chrominance signal of an idol related pixel point also has same phase relation.For TSC-system, its phase relation is more simpler, if O
N, i, jChrominance phase is 0, then O
N, i ± 1, jPoint chrominance phase and O
N, i, jChrominance phase differs 180, as shown in Figure 2, deals with simply relatively, thereby generally all more complicated Phase Alternation Line system is scrutinized, and TSC-system can be easy to derive.
Fig. 3 example the spatial coherence of pixel, be divided into D1 according to the difference of spatial coherence, D2, D3, four types of D4, different gray scales have been represented different picture materials among Fig. 3, wherein D1 class point is all bad with vertical direction next-door neighbour's the correlation of pixel up and down; D2 class point is only good with the correlation of last neighbor pixel; D3 class point is only good with the correlation of following neighbor pixel; D4 class point is all fine with the correlation of its neighbouring pixel.This has reflected the modular characteristics of pixel.Motion conditions according to same module pixel point in the vision signal is similar, thereby should take into full account the spatial coherence of pixel, i.e. its modular characteristics when pixel carried out motion detection.
Fig. 4 detects for spatial coherence and an interior adaption brightness chroma separation module schematic diagram.A spatial coherence testing process and an interior adaption brightness chrominance separation are combined closely, and have saved hardware resource so on the one hand, can realize that on the other hand an interior self adaptation light tone is separated efficiently.Correlation detection module in space has mainly been used trapper among the present invention.Because trapper carry out light tone separate the horizontal resolution of the result that obtains very high, and the spatial coherence detection module mainly detects the vertical correlation between pixel, and know the chromatic value phase place difference of neighbor pixel by analysis before, thereby can not directly subtract each other, often judge correlation according to luminance difference.Earlier vertical related pixel point is asked for brightness value respectively so utilize trapper.According to the selection of the different related pixel points of standard also difference to some extent, be example with strange field signal: generally select O for Phase Alternation Line system
N, i, jAnd O
N, i ± 2, j3 points, and the general O that selects of NTSC value
N, i, jAnd O
N, i ± 1, j3 points.Then with O
N, i, jThe brightness value and the brightness value of upper and lower pairwise correlation pixel poor respectively, with difference difference isospace correlation detection threshold ratio, according to comparative result, generate the spatial coherence testing result [Pre2_line of binaryzation, Next2_line], represented pixel spatial coherence (Pre2_line represents the correlation with upper point, and Next2_line represents the correlation with lower point, 1 the expression correlation good; 0 expression correlation is bad); So [Pre2_line, the Next2_line] of D1 class point shown in Figure 3 value is [0,0]; [Pre2_line, Next2_line] value that D2 is ordered is [1,0]; [Pre2_line, Next2_line] value of D3 class point is [0,1]; [Pre2_line, Next2_line] value of D4 class point is [1,1].The dynamic weights K1 of self adaptation light tone separation scheme in producing on the other hand, K2, K3 (K1+K2+K3=1) are with the weighted average as a result of each the light tone separation module output result as self adaptation light tone separation module in the field: Y_line, C_line.
With the analysis classes of Fig. 3 seemingly, pixel also has temporal correlation.Fig. 5 has reflected this correlation.Horizontal direction express time axle among the figure, vertical axis are the vertical direction of place on the plane.The plane that their constitute perpendicular to the place on the plane.Because the motion conditions difference of pixel, correlation in time also can be divided into D5, D6, D7, D8 4 classes.The corresponding two pixel correlations with the field, front and back of D5 class point expression are all bad; The expression of D6 class point is only good with front court corresponding pixel points correlation; The expression of D7 class point is only good with back court corresponding pixel points correlation; A pixel correlation is all fine before and after the D8 class point.
The selection of dynamic module when Fig. 6-Fig. 9 has represented motion detection.Detection through spatial coherence is divided into 4 classes with pixel, uses [Pre2_line respectively, Next2_line]=00,10,01,11 expressions, with the Phase Alternation Line system is example, has selected module shown in Figure 7 respectively for 4 class pixels, and TSC-system can be selected similar module.When carrying out motion detection, adopt following formula to obtain the weighted average difference of dynamic module respectively successively, represent with d:
When [Pre2_line, Next2_line]=[00], when belonging to D1 class point:
d＝abs(O
n-2，i，j-O
n+2，i，j)
When [Pre2_line, Next2_line]=[10], when belonging to D2 class point:
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i-2，j)+0.15·abs(O
n-1，i-2，j-O
n+1，i，j)
When [Pre2_line, Next2_line]=[01], when belonging to D3 class point:
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i+2，j)+0.15·abs(O
n-1，i+2，j-O
n+1，i，j)
When [Pre2_line, Next2_line]=[11], when belonging to D4 class point:
d＝0.6·abs(O
n-2，i，j-O
n+2，i，j)+0.2·abs(O
n-1，i-1，j-O
n+1，i+1，j)+0.2·abs(O
n-1，i+1，j-O
n+1，i-1，j)
With weighted average difference and motion detection threshold relatively, if greater than threshold value then think that this pixel is the motor point, otherwise think that it is static.
Then testing result is handled through median filter, filter be typically chosen in cross or the X type, shape is as shown in figure 10.The selection of this median filter according to the spatial coherence of pixel, can further improve accuracy of detection equally, reduces probability of miscarriage of justice.
Figure 11 is adaption brightness chrominance separation schematic diagram between temporal correlation detection and field.It separates with 3D self adaptation light tone and is closely connected.Save like this and be used for the hardware cost of depositing the field and realize that 3D self adaptation light tone is separated efficiently.The spatial coherence that temporal correlation detects still with pixel is closely connected.According to value (corresponding D1-D4 class point) the choice of dynamical module of [Pre2_line, Next2_line] and utilize in the different fields light tone separation scheme to ask for brightness value.Still ask for O with Phase Alternation Line system
N, i, jBrightness value be example: if belong to D1 class point, use trapper to ask for O successively
N, i, jAnd O
N, i ± 2, jBrightness value; If belong to D2 class point, use the interior light tone separation scheme in field of top 3 row to ask for O successively
N, i, jAnd O
N, i ± 2, j3 brightness value; If the interior light tone separation scheme in the field that belongs to D3 class point then use bottom 3 to go; If belong to D4 class point then used light tone separation schemes in the field of 5 row.After the time of having asked for goes up relevant 3 brightness value, with the O that tries to achieve
N, i, jBrightness value and O
N ± 2, i, jBrightness value poor, and with difference and temporal correlation threshold ratio, the isospace correlation is checked similar, can generate the temporal correlation testing result [Pre2_Frame, Next2_Frame] of binaryzation.The temporal correlation (Pre2_Frame represents the correlation with the front court corresponding pixel points, and Next2_Frame represents the correlation with the back court corresponding pixel points, and 1 expression correlation is good, and 0 expression correlation is bad) of having represented pixel; So [Pre2_Frame, the Next2_Frame] of D5 class point shown in Figure 5 value is [0,0]; [Pre2_Frame, Next2_Frame] value that D6 is ordered is [1,0]; [Pre2_Frame, Next2_Frame] value of D7 class point is [0,1]; [Pre2_Frame, Next2_Frame] value of D8 class point is [1,1].The dynamic weights K4 of self adaptation light tone separation scheme between producing on the other hand, K5, K6 (K4+K5+K6=1) are with the weighted average as a result of each the light tone separation module final output result as self adaptation light tone separation module in comprehensive and between the field: Y_Frame, C_Frame.
As shown in figure 12, with spatial coherence testing result [Pre2_line, Next2_line], temporal correlation testing result [Pre2_Frame, Next2_Frame] and pixel (can be forms such as RGB, YUV) constitute new video signal data structure.This structure is united vision signal and its corresponding space time information, [Pre2_line, Next2_line] and [Pre2_Frame, Next2_Frame] value can be used as the important references of Video processing, avoid rechecking, comprehensively promoted video signal handling ability.
Figure 13 has represented the structure and the workflow of whole motion detection block.
Claims (8)
1. method for testing motion in the brightness and chroma separation module of video codec, it is characterized in that, the spatial coherence that carries out pixel detects, generate the spatial coherence testing result of binaryzation, spatial coherence Dynamic Selection motion detection block according to pixel, the weighted average difference and the threshold ratio of module are carried out motion detection, simultaneously, carrying out temporal correlation detects, detect the correlation of pixel and field, front and back corresponding pixel points, generate the temporal correlation testing result of binaryzation, thereby determine the motion detection situation of this pixel, and the spatial coherence testing result of binaryzation and temporal correlation testing result and pixel are constituted new pixel data structure.
2. a device of realizing motion detection in the brightness and chroma separation module of video codec is characterized in that, this device comprises:
One frame memory be used for preserving motion detection with between light tone separate the video data of using;
One spatial coherence detection module, this module utilizes trapper that the vertical related pixel point with it of pixel is asked for brightness value successively, calculate luminance difference and compare with the spatial coherence detection threshold, obtain the spatial coherence of pixel, this relevance detection results can be separated for self adaptation light tone in the field provides dynamic weights, the motion detection of the binaryzation testing result of Sheng Chenging after being simultaneously, temporal correlation detection etc. basis for estimation is provided;
One motion detection block, this module is utilized spatial coherence testing result Dynamic Selection motion detection block, in module, ask for the weighted average difference, difference and motion detection threshold are compared, draw the motion conditions of pixel, testing result through median filter, is reduced erroneous judgement, improve accuracy of detection;
One time correlation detection module, spatial coherence testing result according to binaryzation, light tone separation scheme in the Dynamic Selection field, try to achieve the brightness value of pixel to be measured and time last front and back corresponding fields pixel successively, calculate luminance difference and relatively can draw the temporal correlation of pixel, and generate the temporal correlation testing result of binaryzation with the temporal correlation detection threshold.
3. according to the device of motion detection in the described brightness and chroma separation module of video codec of claim 2, it is characterized in that the job step of described spatial coherence detection module is:
At first finishing an interior self adaptation light tone separates, utilize trapper earlier vertical related pixel point to be asked for brightness value respectively then, then that the brightness value of pixel is poor respectively with the brightness value of pairwise correlation pixel up and down, with difference difference isospace correlation detection threshold ratio, according to comparative result, generate the spatial coherence testing result [Pre2_line of binaryzation on the one hand, Next2_line], the spatial coherence of having represented D1-D4 four class pixels respectively, Pre2_line represents the correlation with upper point, Next2_line represents the correlation with lower point, and 1 expression difference is less than threshold value, and correlation is good; 0 expression correlation is bad; Separate each separation scheme for an interior self adaptation light tone on the other hand and generate dynamic weights, weighted average is separated final result as self adaptation light tone in the field.
4. the device of motion detection is characterized in that in the brightness and chroma separation module of video codec according to claim 2, and the selection mode of described motion detection dynamic module is:
Through the detection of spatial coherence, will put mutually and be divided into 4 classes, use [Pre2_line, Next2_line]=00,10,01,11 expression respectively, corresponding to D1-D4 class point, choose corresponding motion detection block for 4 class testing results.
5. the device of motion detection is characterized in that in the brightness and chroma separation module of video codec according to claim 2, and the specific implementation of described motion detection block is:
Carrying out O under the pal mode
N, i, jAdopt following formula to obtain the weighted average difference of dynamic module respectively successively during the point motion detection, represent with d:
When [Pre2_line, Next2_line]=[00],
d＝abs(O
n-2，i，j-O
n+2，i，j)
When [Pre2_line, Next2_line]=[10],
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i-2，j)+0.15·abs(O
n-1，i-2，j-O
n+1，i，j)
When [Pre2_line, Next2_line]=[01],
d＝0.7·abs(O
n-2，i，j-O
n+2，i，j)+0.15·abs(O
n-1，i，j-O
n+1，i+2，j)+0.15·abs(O
n-1，i+2，j-O
n+1，i，j)
When [Pre2-line, Next2-line]=[11],
d＝0.6·abs(O
n-2，i，j-O
n+2，i，j)+0.2·abs(O
n-1，i-1，j-O
n+1，i+1，j)+0.2·abs(O
n-1，i+1，j-O
n+1，i-1，j)
With weighted average difference and motion detection threshold relatively, if greater than threshold value then think this pixel motion, otherwise think that it is static.
6. the device of motion detection is characterized in that in the brightness and chroma separation module of video codec according to claim 2, and described specific implementation of motion detection result being carried out the filter of medium filtering is:
Testing result is handled through median filter, and median filter selects to have utilized the spatial coherence of pixel, improves accuracy of detection, reduces the probability of erroneous judgement.
7. the device of motion detection is characterized in that in the brightness and chroma separation module of video codec according to claim 2, and the specific implementation of described spatial coherence detection module is:
It comprises that brightness asks for dynamic module and choose the unit, and the unit is asked in brightness, and luminance difference is asked for unit and comparator;
Dynamic module is chosen the spatial coherence testing result of unit according to binaryzation, ask for the unit for the interior brightness in the suitable field of the respective pixel point selection on pixel and the time shaft and form dynamic module, utilize in the frame light tone separation scheme to ask for brightness value in module, the same space coordinate pixel of the time being gone up in two of the relevant front and back according to the standard of TV is selected identical dynamic module simultaneously;
Brightness is asked for the unit according to the dynamic module of selecting, and uses in the suitable field light tone separation scheme to ask for brightness value in module, utilizes identical light tone separation scheme to ask for brightness value to the respective modules of dependent field simultaneously;
Luminance difference is asked for the unit and brightness value that dynamic module obtains and time is asked in the brightness of pixel correspondence to go up the brightness value that the related pixel point obtains poor respectively;
Comparator compares this difference respectively with the spatial coherence detection threshold, generate the testing result of binaryzation, and testing result is stored remarked pixel point correlation in time with 2bit.
8. the device of motion detection is characterized in that in the brightness and chroma separation module of video codec according to claim 2, and the data structure specific implementation of described new pixel is:
With spatial coherence testing result [Pre2_line, Next2_line], temporal correlation testing result [Pre2_Frame, Next2_Frame] and digitized pixel point value, or forms such as RGB, YUV, combine and can constitute new video signal data structure.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100210691A CN101489032B (en) | 2009-02-06 | 2009-02-06 | Motion detection method and apparatus in brightness and chroma separation module of video codec |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100210691A CN101489032B (en) | 2009-02-06 | 2009-02-06 | Motion detection method and apparatus in brightness and chroma separation module of video codec |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101489032A true CN101489032A (en) | 2009-07-22 |
| CN101489032B CN101489032B (en) | 2011-08-31 |
Family
ID=40891693
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009100210691A Expired - Fee Related CN101489032B (en) | 2009-02-06 | 2009-02-06 | Motion detection method and apparatus in brightness and chroma separation module of video codec |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101489032B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103686171A (en) * | 2013-12-19 | 2014-03-26 | 中山大学深圳研究院 | Adaptive filtering separation video decoding device |
| CN106797465A (en) * | 2014-10-07 | 2017-05-31 | 高通股份有限公司 | Derive and offset for the quantization parameter of adaptive color conversion in video coding |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20050090005A (en) * | 2003-01-14 | 2005-09-09 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | Method and device for separating a chrominance signal from a composite video base-band signal |
| KR100674940B1 (en) * | 2005-01-13 | 2007-01-26 | 삼성전자주식회사 | Digital video signal processing apparatus and method for adaptive and temporal and spatial Y/C separation in several directions |
| US7345710B2 (en) * | 2005-02-07 | 2008-03-18 | Mediatek Inc. | Systems and methods for processing composite television signals |
-
2009
- 2009-02-06 CN CN2009100210691A patent/CN101489032B/en not_active Expired - Fee Related
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103686171A (en) * | 2013-12-19 | 2014-03-26 | 中山大学深圳研究院 | Adaptive filtering separation video decoding device |
| CN106797465A (en) * | 2014-10-07 | 2017-05-31 | 高通股份有限公司 | Derive and offset for the quantization parameter of adaptive color conversion in video coding |
| CN106797465B (en) * | 2014-10-07 | 2019-07-09 | 高通股份有限公司 | A kind of method and device of decoding video data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101489032B (en) | 2011-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7336316B2 (en) | De-interlacing of video data | |
| US7701508B2 (en) | Accurate motion detection for the combination of motion adaptive and motion compensation de-interlacing applications | |
| US9444977B2 (en) | Method and device for adaptive noise measurement of a video signal | |
| US7321400B1 (en) | Method and apparatus for adaptive image data interpolation | |
| US20040001621A1 (en) | Image processing apparatus and method | |
| US8401318B2 (en) | Motion vector detecting apparatus, motion vector detecting method, and program | |
| JP2000078533A (en) | Method for detecting still area in video image sequence | |
| CN101510985B (en) | Self-adapting de-interleave method for movement compensation accessory movement | |
| CN101489032B (en) | Motion detection method and apparatus in brightness and chroma separation module of video codec | |
| US7460734B2 (en) | Method and apparatus of adaptive interpolation based on edge detection | |
| JP4385077B1 (en) | Motion vector detection device and image processing device | |
| CN101141655A (en) | Video signal picture element point chromatic value regulation means | |
| US7616693B2 (en) | Method and system for detecting motion between video field of same and opposite parity from an interlaced video source | |
| JPH0973540A (en) | Motion vector calculation device | |
| CN1819662A (en) | Systems and methods for processing composite television signals | |
| US7957593B2 (en) | Method and apparatus for processing image, recording medium, and computer program | |
| US20050243934A1 (en) | Processing auxiliary data of video sequences | |
| EP0654941A2 (en) | Motion detection circuit and method using spatial information | |
| CN102263965B (en) | Method and device for eliminating noises from analogue television signal | |
| US8553145B2 (en) | Method and related apparatus for image de-interlacing | |
| US7791673B2 (en) | Method and system for polarity change count | |
| CN108711164A (en) | A kind of method for testing motion based on LBP and Color features | |
| US8902360B2 (en) | Method for detecting image sequences having linewise repeated data | |
| KR100252943B1 (en) | scan converter | |
| US9275295B1 (en) | Noise estimation based on a characteristic of digital video |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110831 Termination date: 20140206 |