CN102595024B - Method and device for restoring digital video images - Google Patents
Method and device for restoring digital video images Download PDFInfo
- Publication number
- CN102595024B CN102595024B CN201110429508.XA CN201110429508A CN102595024B CN 102595024 B CN102595024 B CN 102595024B CN 201110429508 A CN201110429508 A CN 201110429508A CN 102595024 B CN102595024 B CN 102595024B
- Authority
- CN
- China
- Prior art keywords
- pixel
- unit
- restoring area
- reparation
- default
- 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
Abstract
The invention provides a method for restoring digital video images, which includes the steps of acquiring at least two digital video frames; processing acquired frames with every two frames serving as a processing unit each time, and the processing includes computing a residual error of attribute values of pixels corresponding to the two frames in a preset restoration area; respectively accumulating residual errors of all pixels in the preset restoration area; judging whether the accumulated residual error of each of the pixels is larger than or equal to a preset threshold value or not, and recording the pixel if the accumulated residual error is larger than or equal to the preset threshold value; and performing predictive restoration of the preset restoration area according to the attribute values of the recorded pixels. The invention also provides a device for restoring digital video images. The method and the device for restoring digital video images extract effective pixels from the preset restoration area, perform predictive restoration according to effective pixels, increase the accuracy of prediction and improve the effect of restoration.
Description
Technical field
The present invention relates to image repair technical field, relate in particular to a kind of digital video image restorative procedure and corresponding device.
Background technology
Image repair is that prime information defect area on image is carried out to information removal and prediction filling, and its object is to recover the image that prime information is damaged, makes to repair rear subjective once damaged being repaired of image that cannot or be difficult for perceiving.Image repair generally includes manual intervention repairing method and the full-automatic large class of repairing method two.Though manual intervention repairing method can be realized picture and predict more accurately reparation, take time and effort, can not carry out large batch of quick reparation task.Full-automatic repairing method computer is analyzed, can realize large batch of picture reparation, but the method conventionally carries out whole removing to the data in a larger region and remove, and then prediction recovers, make the mass efficient information dropout comprising in region, reparation precision is poor.For digital video image, conventional is the full-automatic repairing method based on single frames, processes unit, after the information whole removing in predeterminable area is removed when reparation taking single frames as one, adopt the Pixel Information at this predeterminable area edge to predict filling to predeterminable area, repair thereby realize.Owing to conventionally comprising in restoring area as the hollow out region outside word, pattern identification etc., these hollow out regions are active data information, when said method is repaired, these information are not all considered, only adopt the Pixel Information at restoring area edge to repair, and the correlation of edge pixel information and predeterminable area is poor, cause forecasting inaccuracy, when video pictures is play, repair fill area and will occur scintillation.
Summary of the invention
Because the video playback flicker problem that prior art adopts the edge pixel information of restoring area to repair to be brought, the goal of the invention of the embodiment of the present invention is to provide a kind of new digital video image restorative procedure and corresponding device, the method and device parse hollow out data in predeterminable area, utilize hollow out data to carry out the reparation of video image, the problem existing to solve prior art.
The digital video image restorative procedure that the embodiment of the present invention provides comprises:
Obtain at least two frame of digital frame of video;
Taking two frames as a processing unit, the frame obtaining is processed, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area at every turn;
The residual values of each pixel in cumulative default restoring area respectively;
Whether the cumulative residual values that judges each pixel is greater than predetermined threshold value, if so, records this pixel;
Preset the prediction reparation of restoring area according to the property value of the pixel of record.
Preferably, described method also comprises: in cumulative default restoring area, after the residual values of each pixel, the residual error of each pixel is isolated to noise Transformatin.
Preferably, described method also comprises: preset reparation according to recording pixel point before, the pixel of record is carried out to shrink process.
Preferably, in the time predicting reparation, calculate respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point, the pixel of choosing degree of correlation maximum carries out the prediction reparation of complex point to be repaired.
Preferably, described method also comprises: after default restoring area prediction is repaired, default restoring area is carried out to low-pass filtering treatment.
Preferably, described digital video adopts yuv format, and the property value of described pixel comprises luminance component, two color difference components.
Preferably, described default restoring area is rectangular area.
The embodiment of the present invention also provides a kind of digital video image prosthetic device, and this device comprises: frame acquiring unit, residual computations unit, the cumulative unit of residual error, judging unit, record cell and reparation unit, wherein:
Described acquiring unit, for obtaining at least two frame of digital frame of video;
Described the first computing unit, for the frame obtaining is processed taking two frames as a processing unit at every turn, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area;
Described cumulative unit, for the residual values of each pixel in the default restoring area that adds up respectively;
Whether judging unit, be greater than predetermined threshold value for the cumulative residual values that judges each pixel, if so, calls this pixel of recording unit records;
Described reparation unit, for presetting the prediction reparation of restoring area according to the property value of the pixel of record.
Preferably, described device also comprises noise removal unit, after the residual values for each pixel in cumulative default restoring area, the residual error of each pixel is isolated to noise Transformatin.
Preferably, described device also comprises shrink process unit, for the pixel of record being carried out to shrink process preset reparation according to recording pixel point before.
Preferably, described device also comprises the second computing unit and chooses unit, and described the second computing unit, in the time predict reparation, calculates respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point; The described unit of choosing is for choosing the pixel of degree of correlation maximum: repair unit for carrying out the prediction reparation of complex point to be repaired according to the pixel of the degree of correlation maximum of choosing.
Preferably, described device also comprises: filter unit, after repairing in default restoring area prediction, carries out low-pass filtering treatment to default restoring area.
The embodiment of the present invention is being obtained after digital video frame, the frame obtaining is carried out to residual computations between two, then cumulative residual error, compares cumulative residual error and predetermined threshold value, can be used for the prediction reparation of default restoring area for the pixel that is greater than this predetermined threshold value.Compared with prior art, the embodiment of the present invention is receiving on multiframe data basis, isolate the effective information data in default restoring area by residual error manner of comparison, these information datas are stronger than the correlation of pixel in the edge pixel data of default restoring area and restoring area generally, therefore, utilize these information datas of separating to repair the accuracy of prediction higher, repair image out more level and smooth, can effectively avoid video playback time, occur image flicker phenomenon.And the embodiment of the present invention, by residual computations, is linked to each other multiple frame of video, overcome the true property of the forecasting inaccuracy defect that uses frame of video to produce in prior art isolatedly.In addition, the embodiment of the present invention is in the time predicting reparation, can calculate complex point to be repaired respectively with the degree of correlation of recording pixel point and edge pixel point, then, choose the pixel of degree of correlation maximum according to actual conditions and predict reparation, like this edge pixel point and recording pixel point are combined and predict reparation, further improved prediction accuracy, improved repairing effect.
Brief description of the drawings
Fig. 1 is the flow chart of an embodiment of the method for the present invention;
Fig. 2 is the flow chart of another embodiment of the method for the present invention;
Fig. 3 is the flow chart of an instantiation of the inventive method embodiment;
Fig. 4 is the composition frame chart of device embodiment of the present invention.
Embodiment
Embodiments of the invention provide a kind of new digital video image restorative procedure and have installed accordingly, the method and device are by carrying out residual computations to the digital video frame of obtaining, isolate the valid data in default restoring area, then utilize these valid data to predict reparation, abandon prior art and removed the data in default restoring area completely, only use the edge pixel data of default restoring area to predict the way of repairing, improve the forecasting accuracy to default restoring area, improve prediction repairing effect, while having solved displaying video image in prior art, there is the problem of image flicker.
As previously mentioned, digital video image restorative procedure of the prior art is to utilize the edge pixel data of presetting outside restoring area to preset the reparation of restoring area, default restoring area is larger, the degree of correlation of pixel that departs from edge in edge pixel data and default restoring area is less, the overall repairing effect that utilizes edge pixel data to carry out is poorer, and will there will be obvious image " highland ", with regard to single frames picture, this default restoring area one with region there is strong lofty sense, unsmooth, nature; While forming dynamic menu continuously with regard to multiframe, by predeterminable area one with region there is the point of scintillation of " fluid ", these have had a strong impact on user's video perception.For this reason, the invention provides a kind of new digital video image restorative procedure, efficiently solve the problems referred to above.Fig. 1 shows an embodiment of the method for the present invention, and the digital video image restorative procedure of this embodiment comprises:
Step S101: obtain at least two frame of digital frame of video;
Obtain digital video frame and can disposablely obtain multiframe, also can obtain one by one; Get the step that can calculate immediately residual error after two digital video frame, also the frame first obtaining can be carried out to buffer memory, after obtaining multiple digital frames, carry out again residual computations step.The quantity of the frame obtaining is here as long as can realize above goal of the invention of the present invention at two, but the frame obtaining is more, the residual error reliability of calculating is stronger, the accuracy of repairing prediction is higher, in actual application, can be according to the quantity that the required precision of repairing prediction is determined to the frame of video of obtaining.
Step S102: taking two frames as a processing unit, the frame obtaining is processed, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area at every turn;
If obtaining the mode of frame is to obtain successively, as long as get after two frames, can carries out the residual computations of respective pixel point, and residual result is preserved temporarily; Be the disposable multiframe of obtaining if obtain the mode of frame, can carry out between two residual computations to the frame obtaining, and checkout result is preserved temporarily.In default restoring area, there are multiple pixels of arranging in length and breadth, each pixel has pixel coordinate value and property value, the property value is here relevant with the concrete form that frame of video adopts, such as adopting yuv format, the Y in the property value of pixel represents the luminance component of video information, and U, V represent two color difference components of video information; Also such as adopting rgb format, the R of the property value of pixel, G, B represent respectively Red Green Blue, default restoring area is herein to choose as required, changeable, can rule can be irregular, but repair and improve repairing effect for ease of realizing, this region generally should be as far as possible little, this region should be generally regular domain as far as possible, especially should with the pixel arrangement mode correspondence of image, such as being rectangular area.In the time carrying out residual computations, pixel node-by-node algorithm residual error to two frames in default restoring area, in the default restoring area of two frames, find after respective point respectively according to pixel coordinate figure, read the property value of respective point, then by poor the property value of two pixels, this residual error if zero, before and after illustrating, this corresponding pixel points of two frames does not change, be likely certain constant mark herein, such as the station symbol of TV programme; If residual error is non-vanishing, before and after illustrating there is variation in this corresponding pixel points of two frames, is likely the hollow out point outside constant mark herein.
Step S103: the residual values of each pixel in cumulative default restoring area respectively;
By above-mentioned steps, each pixel in predeterminable area is carried out after residual computations, the residual values that a pixel is repeatedly calculated adds up.Such as, suppose to comprise in a default restoring area 100 pixels, 10 frame of video are obtained altogether for residual computations according to abovementioned steps, for each pixel, if consecutive frame account form before and after taking needs to carry out residual computations 9 times, produce 9 residual values, then these 9 residual values are added up and obtain the accumulation residual values of this pixel, by that analogy, ask for the accumulation residual values of all the other 99 pixels.
Step S104: whether the cumulative residual values that judges each pixel is greater than predetermined threshold value, if so, records this pixel;
Obtain according to above-mentioned steps after the cumulative residual values of each pixel in default restoring area, cumulative residual values and the predetermined threshold value of pixel are compared, if accumulation residual values is greater than predetermined threshold value, illustrate that this pixel is the picture point that in default restoring area, the hollow out point outside certain constant mark or front and back are changing, this pixel can be used for the pixel of predicting that constant mark takies, is considered as effective pixel points; If accumulation residual values is less than predetermined threshold value, illustrate that this pixel is the constant mark in default restoring area or changes little point, this pixel is not useable for the reparation prediction of predeterminable area, is considered as inactive pixels point.
Step S105: preset the prediction reparation of restoring area according to the property value of the pixel of record.
In default restoring area, isolate effective pixel points, can preset according to the property value of these pixels the prediction reparation of restoring area, concrete restorative procedure can adopt mean value method, variance method etc., these methods are on the books in the prior art, herein repeated description not.
The present embodiment is obtaining after digital video frame, and the frame obtaining is carried out to residual computations between two, and then cumulative residual error, compares cumulative residual error and predetermined threshold value, can be used for the prediction reparation of default restoring area for the pixel that is greater than this predetermined threshold value.Compared with prior art, the present embodiment has been obtained at least following technique effect: (1) the present embodiment is receiving on multiframe data basis, isolate the effective information data in default restoring area by residual error manner of comparison, these information datas are stronger than the correlation of pixel in the edge pixel data of default restoring area and restoring area generally, therefore, utilize these information datas of separating repair prediction accuracy higher, reparation image is out more level and smooth, occurs image flicker phenomenon can effectively avoid video playback time; (2) the present embodiment, by residual computations, is linked to each other multiple frame of video, has overcome the true property of the forecasting inaccuracy defect that uses frame of video to produce in prior art isolatedly.
Above-described embodiment directly adds up after out to the residual computations of each pixel in default restoring area, in fact, for ease of computer processing, in actual application, the present embodiment preferably carries out " digitlization " to residual error to be processed, that is: first set a critical value, if the pixel residual error calculating is greater than this critical value, this residual error is considered as to numeral " 1 ", if the pixel residual error calculating is less than this critical value, this residual error is considered as to digital " 0 ", correspondingly, also be set as an integer to follow-up for the predetermined threshold value that cumulative residual values is compared.After such processing, the step such as relatively of asking for cumulative, the cumulative residual values of residual error and predetermined threshold value only relates to the computing between integer, thereby has accelerated processing speed, has improved on the whole the remediation efficiency of digital video.
Above-described embodiment carries out after residual values accumulation calculating each pixel in default restoring area, these cumulative residual values should exist certain law to distribute, if there is " abnormal height " or " abnormal low " in some residual values, the cumulative residual values entanglement of even adjacent or close pixel is unordered, probably there is problem in these pixels under noise effect, can not truly reflect the variation between frame and frame, for this reason, the present invention preferably after the residual values of each pixel, isolates noise Transformatin to the residual error of each pixel in cumulative default restoring area.The concrete grammar of isolated noise Transformatin has multiple, and the preferred median filtering method of the present invention is realized.So-called median filtering method refers to take out from pending pixel periphery (up and down) value of respective pixel point, and the value that is communicated with oneself sorts, and gets the value in centre position as the new value of this pending pixel.This treatment step can " purify " the cumulative residual values of the pixel in default restoring area, gets rid of some and obviously can not be used for predicting the pixel of restoring area, thereby be conducive to improve the accuracy that prediction is repaired, and further improves repairing effect.
In above-described embodiment, obtain can putting corresponding property value according to recording pixel after recording pixel point and predict reparation, but may there is the erroneous judgement of pixel in the boundary member of the connected region forming at recording pixel point, for this reason, the present invention preferably carried out shrink process to the pixel of record preset reparation according to recording pixel point before.So-called shrink process refers to whether eight pixels of surrounding that judge pending pixel are recording pixel point, if not, this recording pixel point is got rid of.Be conducive to further improve by recording pixel point being carried out to shrink process reference points validity, reduce predictive pictures repair after the lofty sense of subjectivity.
In above-described embodiment, predict after reparation according to the pixel of record, realize goal of the invention of the present invention, but, due to the impact of various factors, in the picture that completes reparation, still likely there is lofty sense in various degree, for this reason, the present invention preferably, after default restoring area prediction is repaired, carries out low-pass filtering treatment to default restoring area.After low-pass filtering treatment, more level and smooth transition between other parts of default restoring area itself and default restoring area and video image picture, has avoided the lofty sense of video pictures, has further strengthened people's vision perception.
Above-described embodiment is predicted to repair according to the property value of these pixels and has been realized goal of the invention of the present invention after the pixel that obtains record.But, although only predeterminable area is predicted to the effect of repairing effective than with reference to the outer peripheral pixel of predeterminable area on the whole with reference to these pixels, but, with regard to the diverse location of predeterminable area, all adopting recording pixel point to replace edge pixel point may not be optimal selection, in fact, pixel for those in predeterminable area edge annex, may predict that repairing effect will be better with reference to the pixel outside predeterminable area edge, reason is that the degree of correlation of pixel in pixel outside edge and edge may be higher.For this reason, the invention provides another embodiment, this embodiment is from previous embodiment except following step is different, and other steps are all identical.Shown in Figure 2, the step that the present embodiment increases is: in the time predicting reparation, calculate respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point, the pixel of choosing degree of correlation maximum carries out the prediction reparation of complex point to be repaired.Like this according to actual conditions, what choose all the time is the prediction that the property value of the pixel that the degree of correlation is the highest carries out, and the accuracy of prediction is higher, and repairing effect will be better.Here the degree of correlation between calculating pixel point can adopt Furthest Neighbor, i.e. two nearer points of distance, and the degree of association is higher, otherwise lower; Also can adopt according to the real image in predeterminable area and give different weights to pixel, then determine the degree of correlation between points according to weight size, can also adopt additive method, the openly various technology of relatedness computation, here no longer repetition in prior art.
For more clearly describing the present invention, describe with a concrete example below.In this example, video image format adopts yuv format, and the record of pixel is realized by array.Referring to Fig. 3, the performing step of this example comprises:
Step S301: array is set, each element N (x of this array, y) with the interior each pixel P (X of default restoring area, Y) correspondence, the initial value of N (x, y) is predisposed to 0, N (x, y) after initialization before video image repair process end-of-job, all do not empty;
Step S302: obtain digital video frame Fn-1, Fn, default restoring area is carried out to pointwise surface sweeping, extract the property value of each pixel, i.e. YUV component;
Step S303: frame Fn-1, the each pixel of Fn frame in default restoring area are asked for to property value residual error one by one according to the following formula:
DYn(x，y)＝abs(fYn(x，y)-fYn-1(x，y))
DUn(x，y)＝abs(fUn(x，y)-fUn-1(x，y))
DVn(x，y)＝abs(fVn(x，y)-fVn-1(x，y))
Step S304: residual values and residual error critical value to each pixel in above-mentioned default restoring area compare according to the following formula:
DYn(x，y)＞＝TY
DUn(x，y)＞＝TU
Dvn(x，y)＞＝TV
If the residual values of three components is all more than or equal to residual error critical value, before and after showing, between frame, there is larger difference at this pixel, Nn (x, y) corresponding points are put to 1; If all or part of residual error critical value that is less than of the residual values of three components, sets to 0 Nn (x, y) corresponding points.
Step S305: Nn (x, y) is isolated to noise and remove, it should be noted that after this step also can be placed on step S306 and carry out.
Step S306: N (x, y) corresponding to each pixel in default restoring area added up, obtain cumulative residual values Nn|
iteration(x, y), i.e. Nn|
iteration(x, y)=Nn-1|
iteration(x, y)+Nn (x, y).
Step S307: by the Nn| of each pixel in default restoring area
iteration(x, y) and predetermined threshold value θ n compare according to the following formula:
Nn|
iteration(x, y) >=θ n
If Nn|
iteration(x, y) is more than or equal to predetermined threshold value θ n, shows that this pixel can be used as the effective information data that prediction is repaired, and records this pixel E (x, y); If Nn|
iteration(x, y) is less than predetermined threshold value θ n, illustrates that this pixel is removable invalid information data.
Step S308: the pixel to record carries out shrink process.Shrink process flow process is: the recording pixel point of judging according to above-mentioned steps is labeled as to 1, i.e. E (x, y)=1, other pixel assignment are 0, get adjacent 8 pixels, the i.e. (x-1 of this pixel, y-1), (x, y-1), (x+1, y-1), (x-1, y), (x+1, y), (x-1, y+1), (x, y+1), (x+1, y+1), if these adjacent 8 pixel E (x, y) are 1, E|
upgrade(x, y)=1, otherwise, E|
upgrade(x, y)=0, processes whole recording pixel points in this manner, and the E (x, y) after upgrading is as new recording pixel point.
Step S309: calculate respectively the degree of correlation of the interior each pixel of default restoring area and above-mentioned new record pixel, default restoring area edge pixel point, choose the pixel of degree of correlation maximum.The edge pixel point here can be after default restoring area is determined, the pixel in the edge pixel district delimiting at this default restoring area outer edge, these edge pixel points are larger near the pixel referential default restoring area of prediction inside edge, and accuracy is higher.
Step S310: default restoring area is predicted to reparation according to the property value of the pixel of the degree of correlation maximum of choosing.
Step S311: default restoring area is carried out to low-pass filtering, to avoid the lofty sense in part after image prediction is filled.
Describe embodiment of the method for the present invention above in detail, correspondingly, the device embodiment that the present invention also provides a kind of video image to repair.Referring to Fig. 4, this device 400 comprises: acquiring unit 401, the first computing unit 402, cumulative unit 403, judging unit 404, record cell 405 and reparation unit 406, wherein:
Acquiring unit 401, for obtaining at least two frame of digital frame of video;
The first computing unit 402, for the frame obtaining is processed taking two frames as a processing unit at every turn, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area;
Cumulative unit 403, for the residual values of each pixel in the default restoring area that adds up respectively;
Whether judging unit 404, be greater than predetermined threshold value for the cumulative residual values that judges each pixel, if so, calls record cell 405 and record this pixel;
Repair unit 406, for preset the prediction reparation of restoring area according to the property value of the pixel of record.
The course of work of this device embodiment is: after acquiring unit 401 obtains at least two frame of digital frame of video, taking two frames as a processing unit, the frame obtaining is processed by the first computing unit 402, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area at every turn; , then by the cumulative unit 403 cumulative residual values of presetting each pixel in restoring area respectively; Judging unit 404 judges whether the cumulative residual values of each pixel is greater than predetermined threshold value, if so, calls record cell 405 and records this pixel; Then, repair by repairing the prediction of presetting restoring area according to the property value of the pixel of record in unit 406.
The present embodiment is obtaining after digital video frame, and the frame obtaining is carried out to residual computations between two, and then cumulative residual error, compares cumulative residual error and predetermined threshold value, can be used for the prediction reparation of default restoring area for the pixel that is greater than this predetermined threshold value.Compared with prior art, the present embodiment has been obtained at least following technique effect: (1) the present embodiment is receiving on multiframe data basis, isolate the effective information data in default restoring area by residual error manner of comparison, these information datas are stronger than the correlation of pixel in the edge pixel data of default restoring area and restoring area generally, therefore, utilize these information datas of separating repair prediction accuracy higher, reparation image is out more level and smooth, occurs image flicker phenomenon can effectively avoid video playback time; (2) the present embodiment, by residual computations, is linked to each other multiple frame of video, has overcome the true property of the forecasting inaccuracy defect that uses frame of video to produce in prior art isolatedly.
Digital video described in said apparatus embodiment can adopt different forms according to actual conditions, if adopt yuv format, the property value of described pixel comprises luminance component, two color difference components; If employing rgb format, the property value of described pixel comprises red, green, blue three primary colors numerical value.Default restoring area is herein to choose as required, changeable, can rule can be irregular, but repair and improve repairing effect for ease of realizing, this region generally should be as far as possible little, and this region should be generally regular domain as far as possible, especially should with the pixel arrangement mode correspondence of image, such as being rectangular area.
Said apparatus embodiment can also comprise noise removal unit 407, after the residual values for each pixel in cumulative default restoring area, the residual error of each pixel is isolated to noise Transformatin.This unit " purification " the cumulative residual values of pixel in default restoring area, got rid of some and obviously can not be used for predicting the pixel of restoring area, thereby improved the accuracy that prediction is repaired, repairing effect further improves.
Said apparatus embodiment can also comprise shrink process unit 408, for the pixel of record being carried out to shrink process preset reparation according to recording pixel point before.
Said apparatus embodiment can also comprise: the second computing unit 409 and choose unit 410, described the second computing unit, in the time predicting reparation, calculates respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point; The described unit of choosing is for choosing the pixel of degree of correlation maximum: repair unit for carrying out the prediction reparation of complex point to be repaired according to the pixel of the degree of correlation maximum of choosing.Like this according to actual conditions, what choose all the time is the prediction that the property value of the pixel that the degree of correlation is the highest carries out, and the accuracy of prediction is higher, and repairing effect will be better.
Above-described embodiment can also comprise filter unit 411, after repairing in default restoring area prediction, default restoring area is carried out to low-pass filtering treatment.After low-pass filtering treatment, more level and smooth transition between other parts of default restoring area itself and default restoring area and video image picture, has avoided the lofty sense of video pictures, has further strengthened people's vision perception.
The foregoing is only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any amendment of doing, be equal to replacement, improvement etc., within all should being included in the protection range of invention.
Claims (10)
1. a digital video image restorative procedure, described image repair is for prime information defect area on image being carried out to information removal and predicting and fill, and it is characterized in that, described method comprises:
Obtain at least two frame of digital frame of video;
Taking two frames as a processing unit, the frame obtaining is processed, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area at every turn;
The residual values of each pixel in cumulative default restoring area respectively;
Whether the cumulative residual values that judges each pixel is more than or equal to predetermined threshold value, if so, records this pixel;
Preset the prediction reparation of restoring area according to the property value of the pixel of record;
In the time predicting reparation, calculate respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point, the pixel of choosing degree of correlation maximum carries out the prediction reparation of complex point to be repaired.
2. method according to claim 1, is characterized in that, described method also comprises: in cumulative default restoring area, after the residual values of each pixel, the residual error of each pixel is isolated to noise Transformatin.
3. method according to claim 1, is characterized in that, described method also comprises: preset reparation according to recording pixel point before, the pixel of record is carried out to shrink process.
4. method according to claim 1, is characterized in that, described method also comprises: after default restoring area prediction is repaired, default restoring area is carried out to low-pass filtering treatment.
5. according to the method described in any one in claim 1 to 4, it is characterized in that, described digital video adopts yuv format, and the property value of described pixel comprises luminance component, two color difference components.
6. according to the method described in any one in claim 1 to 4, it is characterized in that, described default restoring area is rectangular area.
7. a digital video image prosthetic device, described image repair is for carrying out information removal and prediction filling to prime information defect area on image, it is characterized in that, this device comprises: acquiring unit, the first computing unit, cumulative unit, judging unit, record cell and reparation unit, wherein:
Described acquiring unit, for obtaining at least two frame of digital frame of video;
Described the first computing unit, for the frame obtaining is processed taking two frames as a processing unit at every turn, described processing comprises the residual error of calculating the property value of two frame corresponding pixel points in default restoring area;
Described cumulative unit, for the residual values of each pixel in the default restoring area that adds up respectively;
Whether described judging unit, be greater than predetermined threshold value for the cumulative residual values that judges each pixel, if so, calls this pixel of recording unit records;
Described reparation unit, for presetting the prediction reparation of restoring area according to the property value of the pixel of record;
Described device also comprises: the second computing unit and choose unit, and described the second computing unit, in the time predict reparation, calculates respectively the degree of correlation of the edge pixel point of complex point to be repaired and default restoring area, complex point to be repaired and recording pixel point; The described unit of choosing is for choosing the pixel of degree of correlation maximum: repair unit for carrying out the prediction reparation of complex point to be repaired according to the pixel of the degree of correlation maximum of choosing.
8. device according to claim 7, is characterized in that, described device also comprises: noise removal unit, after the residual values for each pixel in cumulative default restoring area, isolates noise Transformatin to the residual error of each pixel.
9. device according to claim 7, is characterized in that, described device also comprises: shrink process unit, and for the pixel of record being carried out to shrink process preset reparation according to recording pixel point before.
10. device according to claim 7, is characterized in that, described device also comprises: filter unit, after repairing in default restoring area prediction, carries out low-pass filtering treatment to default restoring area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110429508.XA CN102595024B (en) | 2011-12-16 | 2011-12-16 | Method and device for restoring digital video images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110429508.XA CN102595024B (en) | 2011-12-16 | 2011-12-16 | Method and device for restoring digital video images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102595024A CN102595024A (en) | 2012-07-18 |
| CN102595024B true CN102595024B (en) | 2014-10-22 |
Family
ID=46483200
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110429508.XA Active CN102595024B (en) | 2011-12-16 | 2011-12-16 | Method and device for restoring digital video images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102595024B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104581386B (en) * | 2014-12-23 | 2017-11-07 | 深圳市九洲电器有限公司 | A kind of television program playing method and system |
| CN110335389B (en) * | 2019-07-01 | 2021-10-12 | 上海商汤临港智能科技有限公司 | Vehicle door unlocking method, vehicle door unlocking device, vehicle door unlocking system, electronic equipment and storage medium |
| CN110490904B (en) * | 2019-08-12 | 2022-11-11 | 中国科学院光电技术研究所 | Weak and small target detection and tracking method |
| CN111292255B (en) * | 2020-01-10 | 2023-01-17 | 电子科技大学 | Filling and correcting technology based on RGB image |
| CN111614996B (en) * | 2020-04-07 | 2022-03-18 | 上海推乐信息技术服务有限公司 | Video repair method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1234691A (en) * | 1998-04-14 | 1999-11-10 | 汤姆森多媒体公司 | Method for detecting static area in a series of video pictures |
| CN1423487A (en) * | 2001-12-03 | 2003-06-11 | 微软公司 | Automatic detection and tracing for mutiple people using various clues |
| CN1841049A (en) * | 2005-03-28 | 2006-10-04 | 大日本网目版制造株式会社 | Welding disk pattern inspection method and inspection device |
| CN101355708A (en) * | 2007-07-25 | 2009-01-28 | 中兴通讯股份有限公司 | Self-adapting method for shielding error code |
| CN102025919A (en) * | 2010-11-17 | 2011-04-20 | 无锡中星微电子有限公司 | Method and device for detecting image flicker and camera applying device |
-
2011
- 2011-12-16 CN CN201110429508.XA patent/CN102595024B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1234691A (en) * | 1998-04-14 | 1999-11-10 | 汤姆森多媒体公司 | Method for detecting static area in a series of video pictures |
| CN1423487A (en) * | 2001-12-03 | 2003-06-11 | 微软公司 | Automatic detection and tracing for mutiple people using various clues |
| CN1841049A (en) * | 2005-03-28 | 2006-10-04 | 大日本网目版制造株式会社 | Welding disk pattern inspection method and inspection device |
| CN101355708A (en) * | 2007-07-25 | 2009-01-28 | 中兴通讯股份有限公司 | Self-adapting method for shielding error code |
| CN102025919A (en) * | 2010-11-17 | 2011-04-20 | 无锡中星微电子有限公司 | Method and device for detecting image flicker and camera applying device |
Non-Patent Citations (2)
| Title |
|---|
| 一种累计帧差视频对象分割算法;许悦雷等;《光电工程》;20040731;第31卷(第7期);第69-72页 * |
| 许悦雷等.一种累计帧差视频对象分割算法.《光电工程》.2004,第31卷(第7期),69-72页. |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102595024A (en) | 2012-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102595024B (en) | Method and device for restoring digital video images | |
| KR101001086B1 (en) | Method and apparatus for modeling film grain patterns in the frequency domain | |
| CN103093419B (en) | A kind of method of detected image sharpness and device | |
| US9424630B2 (en) | Method and apparatus for detecting and removing false contour, method and apparatus for verifying whether pixel is included in contour, and method and apparatus for calculating simplicity | |
| CN102883175B (en) | Methods for extracting depth map, judging video scene change and optimizing edge of depth map | |
| US8437567B2 (en) | Image inpainting apparatus and method using restricted search region | |
| US20100183236A1 (en) | Method, medium, and apparatus of filtering depth noise using depth information | |
| CN101873509B (en) | Method for eliminating background and edge shake of depth map sequence | |
| CN100548028C (en) | Partial image filtering method based on Noise Estimation | |
| CN101867832B (en) | Apparatus and method for determining threshold value | |
| CN107645621A (en) | A kind of method and apparatus of Video processing | |
| CN102025959A (en) | System and method for producing high definition video from low definition video | |
| CN106412720A (en) | Method and device of removing video watermarks | |
| CN107659806B (en) | The appraisal procedure and device of video quality | |
| JP2005198274A (en) | Ringing eliminating filter | |
| CN107958481A (en) | A kind of three-dimensional rebuilding method and device | |
| CN112233055B (en) | Video mark removing method and video mark removing device | |
| CN110211069A (en) | A kind of image denoising model and training method, system, equipment, computer media | |
| CN103067671A (en) | Method and device of image display | |
| CN107360435B (en) | Blockiness detection methods, block noise filtering method and device | |
| TWI524782B (en) | Image processing apparatus and image processing method | |
| CN108701353A (en) | It is a kind of to inhibit the pseudo- color method and device of image | |
| CN1331097C (en) | Spot eliminating method for digital image | |
| CN109348288A (en) | A kind of processing method of video, device, storage medium and terminal | |
| CN100481881C (en) | Method and device for reducing intraframe noise |
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 |