CN1801886A - Digital image flashing scene detecting and eliminating method - Google Patents
Digital image flashing scene detecting and eliminating method Download PDFInfo
- Publication number
- CN1801886A CN1801886A CN 200510003654 CN200510003654A CN1801886A CN 1801886 A CN1801886 A CN 1801886A CN 200510003654 CN200510003654 CN 200510003654 CN 200510003654 A CN200510003654 A CN 200510003654A CN 1801886 A CN1801886 A CN 1801886A
- Authority
- CN
- China
- Prior art keywords
- value
- picture
- lightness
- lightness difference
- scene
- 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
- Image Analysis (AREA)
- Studio Devices (AREA)
- Image Processing (AREA)
Abstract
Present invention refers to method for detecting and eliminating digit image flash scene. It uses averaging shot distribution principle in digit image scene as reference for detecting and judging three types of flash scene. The calculation picture brightness difference value is used to substitute practical analyses vision content, to reduce image message complexity operating, accurately detecting flash picture in image signal. Said method includes taking picture from inputted image sequence and calculating relative brightness difference value between and neighbor picture, making a histogram, determine a critical value for detecting flash scene in image sequence and determining three types in it, finally eliminating said flash scene.
Description
Technical field
The present invention relates to the detection and the removing method of flashing scene in a kind of digitized video, particularly relate to a kind ofly by obtaining picture message accurately in the image, and detect the wherein scene of flash of light, and the detection and the removing method of flashing scene in the digitized video of being eliminated.
Background technology
As when viewing and admiring live shows such as prevalent fashion, concert or physical culture report; Shi Changhui produces many flashing scenes (flash scene) that influence picture quality; normally the audience uses photoflash lamp or other optical device to produce in order to promote photograph picture quality or effect; or when rear end making (after-treatment), specific (the special editing) that on picture, adds in order to attract audience's sight.Yet this flashing scene but influences image quality, and can disturb image in the future to divide microscopy to survey the analysis of (shot detection) and digital content.
Flashing scene can result from and belong to same image and divide and manyly under the mirror (shot) do not wish the picture wanted, usually, the accuracy that image divides microscopy to survey (shot detection) is extremely important concerning digital presentation content is analyzed, is urgent and important so the known image of development improvement divides microscopy survey method to remove no picture.
Fig. 1 shows the disclosed scene/image of US2004/0008284A1 United States Patent (USP) and divides mirror to change detection (scene/shot detection) How It Works flow chart, one detector is wherein arranged, it comprises the picture buffer (frame buffer) that can store two pictures, can with the picture conversion of wherein storing color signal, calculate the variation that scene/image divides mirror in the histogram (histogram) that detector then produces by this color signal, its step is as described in following:
During beginning, divide mirror to change, be stored to picture buffer (step S150) respectively by two pictures that capture in the image in order to detect scene/image; Picture conversion with storage is the color message (step S152) of colourity or brightness again; Wherein a histogram detecting unit calculates the histogram (step S154) of relevant its colourity or brightness message; Afterwards, by calculating relatively two histograms, draw interdependent value C (step S156) wherein; Relatively this interdependent value C and default critical value (threshold) again, and judge whether that this value C is less than this critical value? (step S158); If C is less than critical value for this value, then scene changes signal Csc and is output as 1, otherwise scene changes signal Csc is 0 (step S160, S162).
About known scene change-detection shown in Figure 1 be comparison one histogrammic image signal and one the prediction critical value, and wherein image signal from two signals that are stored to the picture conversion of buffer.
See also known technology US2003/0123726A1 United States Patent (USP) shown in Figure 2 and detect the functional block diagram that scene changes, this block schematic diagram includes the first accumulation histogram acquisition unit (accumulated histogram extracting part) 201, first pixel tabulation acquisition unit (first pixel list extracting part) 202, the second accumulation histogram acquisition unit (second accumulated histogram extracting part) 203, second pixel tabulation acquisition unit (second pixel list extracting part) 204, histogram comparing unit (histogram comparing part) 205 and scene conversion determining unit (sceneconversion determining part) 206.
Wherein, the first picture I
1With the second contiguous picture I
2By capturing respectively in the image of cumulative distribution to the first accumulation histogram acquisition unit 201 and the second accumulation histogram acquisition unit 203, afterwards, first pixel tabulation acquisition unit 202 captures pixel column tabular value (pixel list) C by the distribution value that corresponds to the first accumulation histogram acquisition unit 201
1, the acquisition unit 204 of second pixel tabulation simultaneously also captures pixel column tabular value (pixel list) C from the second accumulation histogram acquisition unit 203
2, and draw output valve L respectively
1With L
2, compare L by histogram comparing unit 205 again
1With L
2Draw a picture difference value E, last, relatively with the generation determining scene/image and divide mirror variation (scene/shot change) whether this picture difference value and a default critical value for scene conversion determining unit 206.
By the running that repeats each unit shown in Figure 2, the difference value of the each computing of memory stores need be arranged, find out the mean difference value of each stored picture at last again, divide mirror to change to judge scene/image.
It is known that scene changes or the technology of flashing scene in order to detect, be all and analyze the intensity that indivedual pictures are found out pixel intensity, but if when being used for the picture of large scale white background, the result of erroneous judgement but takes place easily, and, if only the critical value of a reflection of foundation high-brightness region is judged the flashing scene in the image, also be difficulty.
Because the defective of known technology, the invention provides one can efficient detection and eliminate flashing scene, and uses the method that image accurately divides the image analysing computer and the management of mirror message that obtains.
Summary of the invention
A kind of method that is used to detect and eliminate digital image flashing scene, the principle of utilizing image to divide mirror to distribute, replace actual analysis vision content (content) with the lightness difference value of calculating two continuous pictures (frame), can reduce the known technology complex calculations, more can detect the position of flash of light picture in the image signal exactly.Flashing scene can be divided into three main types, the step of its method includes: capture picture in the image sequence by input, and calculate lightness difference value between picture and adjacent pictures, make a histogram by the lightness difference value, determine a critical value afterwards, use the flashing scene that detects in the image sequence, and determine wherein three types, eliminated this flashing scene at last.
As continuous two histogrammic peak values greater than critical value, promptly be judged as the flashing scene of the first kind, the detection step that it comprises has: compare lightness difference value and its predetermined critical value in the histogram, then calculate continuous two peak value differences greater than the lightness difference value of critical value, again the comparison peak value difference different with a predetermined lightness percentage, percentage was little if peak value difference should become clear, and then was judged as the first kind.
Continue to surpass a picture as flashing scene, and the picture of two or more continuous generation flashes of light can form two peak values, one specific distance is arranged between two peak values, promptly be judged as the flashing scene of second type, the detection step that it comprises has: compare lightness difference value and predetermined critical value in the histogram, then calculate two greater than the number of pictures between two peak values of the lightness difference value of this critical value, peak value difference when calculating two lightness difference value that continue again all greater than critical value, relatively this peak value difference and a predetermined lightness percentage, percentage was little if peak value difference should become clear, and then was second type.
Because the specific picture in the digitized video can cause a plurality of images to divide mirror to change (shotchange), and wherein Yan Xu picture is in the chain-reacting amount of being scheduled to, promptly be judged as the flashing scene of the 3rd type, the detection step that it comprises has: compare lightness difference value and predetermined critical value in the histogram, and judge that image wherein divides mirror to change; Detecting next image divides mirror to change; Calculate two images and divide the number of pictures of mirror between changing; Relatively this number of pictures and the chain-reacting amount that should be scheduled to detect at last to including the 3rd type of specific picture.
Description of drawings
Fig. 1 is the operation workflow figure of known technology scene change-detection;
Fig. 2 is the function block schematic diagram of known technology scene change-detection;
Fig. 3 A and Fig. 3 B are the corresponding relation schematic diagram of each picture and its lightness difference value;
The function block schematic diagram that Fig. 4 detects for flashing scene of the present invention;
Fig. 5 detects and the removal process flow chart for flashing scene of the present invention;
Fig. 6 A is the histogram of the lightness difference value of the first kind of the present invention;
Fig. 6 B is the steps flow chart that the first kind detects;
Fig. 7 A is one of histogram of the lightness difference value of the present invention's second type;
Fig. 7 B is two of the histogram of the lightness difference value of the present invention's second type;
Fig. 7 C is the steps flow chart of second type detection;
Fig. 8 A is the histogram of the lightness difference value of the present invention's the 3rd type;
Fig. 8 B is the steps flow chart of the 3rd type detection.
The reference numeral explanation
The first accumulation histogram acquisition unit 201
First pixel tabulation acquisition unit 202
The second accumulation histogram acquisition unit 203
Second pixel tabulation acquisition unit 204
Scene conversion determining unit 206
Pixel column tabular value C
1
Pixel column tabular value C
2
Pixel tabulation acquisition unit output valve L
1, L
2
Picture difference value E
End value a, b, c, d
Peak value f, g, i, j, V, W, X, Y, Z
Valley h, p, q
Peak value difference e1, e2, e 3
Critical value T
Image sequence data 401
Image member acquisition unit 403
Histogram comparing unit 405
Critical value definition unit 407
Flashing scene analytic unit 409
Flashing scene judging unit 411
Embodiment
The principle that the present invention divides mirror to distribute (averaging shot distribution) according to average image, disclose the method that detects and can eliminate flashing scene in a kind of digitized video, and the histogram analysis that produces by each flashing scene effect and be categorized as three main types, and to calculate the visual intensity of the direct analysis of lightness difference value (luminance difference) replacement between each contiguous picture (frame) in the image sequence (video sequence), not only can reduce the computational complexity when carrying out, more can accurately detect the picture position that flash events takes place.Below disclose preferred embodiment of the present invention and accompanying drawing thereof, embodiment and accompanying drawing be just in order to illustrate, the present invention not with shown in embodiment exceed.
It is the special noun of an image processing that image divides mirror (shot), image unit for smallest addressable (addressable), an image divides mirror can include continuous a plurality of picture (frame), an image divides the length (shot length) of mirror to be defined as the distance that two images divide mirror to change (shot change), and it is according to (pixel-based) that existing image divides microscopy survey (shot detection) technology to include with the pixel, with the histogram is according to (histogram-based), with the image feature is according to (feature-based), be converted into according to detection methods such as (transform-based) according to dicom message statistics (statistic-based) and image.Yet, judge that with the lightness difference value of two continuous pictures it is the method for extensive use that image divides mirror to change, if the absolute value of its difference value judges that promptly having image to divide mirror to change produces greater than a certain threshold.
See also the corresponding relation schematic diagram of Fig. 3 A and each picture of Fig. 3 B and its lightness difference value.Fig. 3 A shows a plurality of continuous pictures (m that a sequence is represented, m+1, m+2 ... m+k, m+k+1, m+k+2 ...), when flash events betides a fragment in the time, m+1 as shown, m+2 picture to the m+k, the flashing scene shown in the m+k+1 picture, the bright intensity of two adjacent pixels is subtracted each other, and is corresponding lightness difference value histogram shown in Fig. 3 B with its absolute value representation.
Wherein | L
M+1-L
m| be expressed as the lightness difference value of m picture with the m+1 picture that flash of light takes place of not flash of light, and on histogram, produce an end value a; Difference value between m+1 picture to the m+2 picture is expressed as | L
M+2-L
M+1|, because m+1 picture and m+2 picture are all the flash of light picture, then its difference is shown as another lower end value b; The lightness difference value of m+k picture to the m+k+1 picture is | L
M+k+1-L
M+k|, produce the 3rd end value c, and | L
M+k+2-L
M+k+1| the lightness difference value of the m+k+2 picture that disappears for m+k+1 picture that flash of light is arranged and flash effect is a higher end value d.
As previously discussed, learn that obviously the generation of end value a represents that the m+1 picture has produced flash effect, m+k+1 picture last picture of flash effect is for this reason then judged in the generation of end value d, and end value b and c have slow fluctuating, be less lightness difference value, variation between display frame is few, being judged as then that flash of light takes place during.
See also the function block schematic diagram that Fig. 4 flashing scene of the present invention detects, during beginning, one image sequence data (video sequence), 401 inputs are arranged, this image sequence is that a series of image frame is formed, receive by image member histogram acquisition unit (component histogramextraction unit) 403 again, then capture wherein many a series of pictures, and according to the message of each picture (as lightness, luminance) produce an image histogram, afterwards, this image histogram then passes through histogram comparing unit (histogram comparing unit) 405, wherein adjacent picture is done one relatively, and calculate lightness difference value between each picture, and be shown in a lightness difference value histogram, simultaneously, 407 histograms that produced by image member acquisition unit 403 of critical value definition unit (predefined thresholdunit) define a reasonable critical value of judging flashing scene.
Then, the critical value that the histogram of lightness difference value is corresponding with it is sent to a flashing scene analytic unit (flash scene analyzing unit) 409, analytic unit thus, wherein lightness difference value and its critical value are compared, and the peak value difference (peak difference) of two adjacent peak values in the further compute histograms, afterwards, judge type, the first kind disclosed (Type 1) and second type (Type2) of this flashing scene as the present invention according to above-mentioned comparative result and peak value difference with flashing scene judging unit (flashscene determining unit) 411.Simultaneously, critical value by lightness difference value between each picture and 407 definition of critical value definition unit compares, and calculate two images that flashing scene analytic unit 409 judged and divide the number of pictures of mirror between changing, judge the 3rd type (Type 3) that the specific picture by image produces with flashing scene judging unit 411 at last.
Fig. 5 then narrates the steps flow chart of flashing scene detection of the present invention and removing method preferred embodiment.When flashing scene detection step begins, import image sequence data (video sequence), its source is the digitized video (step S501) of a sequence, and afterwards, this image sequence captures a wherein sequential images picture (step S503) by image member acquisition unit shown in Figure 4.And calculate many groups lightness difference value (step S505) between adjacent in twos image frame, and for instance, the lightness difference value of m picture and m+1 picture is an absolute value | L
M+1-L
m|, L wherein
mThe bright value of representing the m picture, L
M+1The bright value of representing the m+1 picture.
Afterwards, this embodiment is promptly according to should many group lightness difference value making a lightness difference histogram (luminance difference histogram) (step S507), and determine a critical value (step S509) according to this lightness difference histogram, use the flashing scene that detects image sequence.Then,, determine the type of this flashing scene, in the embodiment of the invention, be categorized as three types (step S511) according to the relation of lightness difference histogram and critical value.The image of mat in dissimilar divides mirror to change to judge the generation (step S513) of flashing scene again, and eliminates this flashing scene (step S515), and its method of eliminating flashing scene comprises flashing scene by ignoring in the image sequence; Replace it by near the picture mean value the flashing scene of this image sequence; Or the contained picture value of the flashing scene that recomputates this image sequence.
Type with next disclosure three kinds of flashing scenes shown in the present:
The first kind:
Fig. 6 A is the histogram of the lightness difference value of the first kind of the present invention.A flash events that betides a picture can form two continuous and similar peak values at the lightness difference histogram, as shown difference peak value f and g, because these two continuous peak values are all greater than critical value T, then be judged as the flashing scene of the first kind, also can obtain taking place the picture position of flash events easily.And the peak value difference e1 between two peak values need fall within the lightness percentage P, and one embodiment is 10%, and it defines as the formula (1).
H in this formula
nBe the lightness difference value of adjacent n picture and n+1 picture, H
N+1Be the lightness difference value of adjacent n+1 picture and n+2 picture, max () is for drawing H
nWith H
N+1Maximum.Difference value percentage between two different peak values of this lightness percentages show, for instance, lightness difference value H
N+1Equal the absolute value of lightness difference between two pictures | L
N+2-L
N+1|, lightness difference value H
nEqual absolute value | L
N+1-L
n| etc., the detection step of its first kind is shown in Fig. 6 B.
During beginning, need to determine the type of flashing scene, embodiment is as importing image sequence image member acquisition unit, draw many pictures of this image, calculate the lightness difference value of each adjacent pictures again, and make a lightness difference histogram, and determine critical value according to this histogram afterwards, use and detect the first kind (step S601).
Do you judge that whether peak value that the lightness difference value of adjacent n picture and n+1 picture produces is all greater than this critical value? (step S603), if the lightness difference value is less than this critical value, promptly more next lightness difference value, promptly more following two peak values that form by the lightness difference value of picture, if continuous two peak values greater than this critical value, promptly continue to calculate the peak value difference (peakdifference) between peak value.Afterwards, is the histogrammic peak value difference of brighter again degree less than as the formula (1) lightness percentage P? (step S65), if this peak value is greater than lightness percentage P, then follow more following two adjacent peak values greater than this critical value, if peak value difference less than this lightness percentage P, judges that then corresponding picture is flashing scene (step S607).As previously discussed, greater than this critical value, promptly be judged as the flashing scene of the first kind, promptly finish the detection (step S609) of this first kind as the different histogrammic peak value of continuous two bright value differences.
Second type:
Fig. 7 A and Fig. 7 B are the histogram of the lightness difference value of the present invention's second type, continue to surpass a picture as flashing scene, and the picture of two or more continuous generation flashes of light can form two close peak values, and a specific distance is arranged between two peak values, promptly be judged as the flashing scene of second type.According to the principle that known image divides mirror to distribute, two adjacent images in the flashing scene divide mirror to have 10 pictures at least between changing, thus the present invention according to this principle can obtain taking place the picture position of flash events easily.In present embodiment, advise that adjacent image divides the rarest 10 pictures between the mirror variation, but practical application is not as the condition that limits.
Wherein Fig. 7 A shows that flashing scene occurs in two pictures, and in its lightness difference histogram, because the lightness of adjacent pictures is approaching, so back generation one valley h is subtracted each other in system of Himdu logic brightness between two peak value i and j, two peak value differences are peak value difference e2.Fig. 7 B then shows the general type of second type, and flash events betided in a period of time, and a plurality of (k) picture is arranged, but is generally in the given number of pictures, is that k is less than 10 as one embodiment.The lightness difference value of first between the flash of light duration and last picture can form two peak values, and peak value difference e3 is arranged, and the lightness because of picture between two peak values is approaching, so its lightness difference value is slow distribution.The steps flow chart of second type detection sees also Fig. 7 C.
For detecting the flashing scene of second type, before the beginning, the input image sequence is to image member acquisition unit, acquisition many pictures wherein, calculate the lightness difference value of each adjacent pictures again, and make a lightness difference histogram, and determine critical value according to this histogram afterwards, use and begin to detect second type.
The detection of second type start from this critical value determine after (step S701), lightness difference value between adjacent pixels and critical value are relatively in twos, in second type, flashing scene can continue in a plurality of pictures, as k picture, judge whether that then the lightness difference value of n picture and n+k+1 picture (is H
nWith H
N+k+1) greater than this critical value? and judge that difference value between all the other pictures is less than this critical value, and the continuity number of pictures is k, that the k value is less than a given number? the principle of dividing the mirror distribution according to image, preferred embodiment can 10 is example, but be not limited to this numeral (step S703), wherein if the lightness difference value less than critical value, if then more next lightness difference value is two lightness difference value (H that a plurality of pictures are apart arranged
nWith H
N+k+1) greater than critical value, then continue the step that next calculates peak value difference.And, greater than this given number, then continue the step of more next lightness difference value and this critical value as if the number of pictures between two peak values, less than this given number, then continue next step as if the number of pictures between two peak values, promptly calculate peak value difference.
Afterwards, are the two peak value differences (peak difference) of brighter degree difference histogram less than the lightness percentage P shown in the formula (1)? (step S705), if peak value difference is greater than this lightness percentage, then more following again two peak values greater than this critical value, if this peak value difference is less than this lightness percentage, then relative a plurality of pictures (i.e. n picture to the n+k+1 picture) are judged as the flashing scene (step S707) of the present invention's second type between these two peak values.Promptly finish the detection step (step S709) of second type.
The 3rd type:
Specific picture in the digitized video (special editing) can cause a plurality of images to divide mirror to change the false judgment of (shot change), the embodiment that the present invention is directed to the judgement of the 3rd type is the principle of dividing mirror to distribute according to image, if wherein the situation of flash effect is not can classify in the first kind or second type, can be categorized as the 3rd type.Wherein the picture of the 3rd type flashing scene continuity is in the chain-reacting amount of being scheduled to, lightness difference value histogram shown in Fig. 8 A, the specific picture of image can be judged as a series of flashing scene, if producing the image of some in certain hour, flashing scene divide mirror to change, the number of pictures that contains during this time also within a chain-reacting amount, is judged as the flashing scene of the 3rd type.In the lightness difference histogram shown in Fig. 8 A, in certain chain-reacting amount, the lightness difference value is greater than having produced a plurality of peak value V more than the critical value T, W, X, Y and Z, a plurality of valleies are more arranged during this time, and as p and q, the present invention promptly is judged as the group flashing light scene that this image has a specific picture to produce.
Fig. 8 B has disclosed the steps flow chart of the 3rd type detection.
The image sequence input of numeral by wherein capturing a plurality of pictures, calculates the lightness difference value between each adjacent pictures, and makes a histogram, and histogram can be determined a critical value thus.
Critical value determines that back the 3rd type detection begins (step S801), each lightness difference value and this critical value compare, relatively is the lightness difference value of n picture greater than this critical value? (step S803) if not, then need continue to detect lightness difference value greater than critical value, if, being shown in the n picture has produced image and divides mirror to change, promptly continue to detect the next n+k picture that has image to divide mirror to change? represent that specific picture has continued k picture (step S805), afterwards this a number of pictures k and a chain-reacting amount of being scheduled to are compared, this example is an example with 10, but be not limited to this, do you judge that promptly number of pictures is greater than this chain-reacting amount? (step S807), if less than this chain-reacting amount, still having the image that specific picture causes in backward the picture of expression divides mirror to change, then continuing to detect next image divides mirror to change, if greater than this chain-reacting amount, then detecting has image to divide mirror to change, and carry out two images in the compute histograms and divide the number of pictures j (step S809) of mirror between changing, picture shown in this number of pictures is judged as this image and divides the specific picture of mirror between changing, promptly n+1 picture to the n+j+1 picture is specific picture (step S811), promptly finish the detection (step S813) of specific picture this time, can eliminate at the flashing scene that detects specific picture generation afterwards.
The present invention eliminates this detected flashing scene method and comprises flashing scene by ignoring in this image sequence; The picture that flashing scene is arranged by near the replacement of the picture mean value the flashing scene of image sequence; Or recomputate the flashing scene of this image sequence.
In sum, lightness difference value in the mat comparative figures image between picture is also made a lightness difference histogram, by wherein determining a critical value, and calculate the wherein difference between peak value, more, give a detection else and the method for eliminating flashing scene by three main flashing scene types of histogrammic tagsort.
The above only is a preferable possible embodiments of the present invention, and is non-so promptly limit claim of the present invention, and the equivalent structure that all utilizations specification of the present invention and diagramatic content are done changes, and all is contained in the scope of the present invention.
Claims (20)
1. a digital image flashing scene detects and removing method, and this method step includes:
Import image sequence data;
Capture a sequential images picture;
Calculate the many groups lightness difference value between this adjacent in twos image frame;
Make a lightness difference histogram according to these many group lightness difference value;
Determine a critical value according to this lightness difference histogram, use the flashing scene that detects this image sequence;
According to the related type of determining this flashing scene of this lightness difference histogram with this critical value; And
Eliminate this flashing scene.
2. digital image flashing scene as claimed in claim 1 detects and removing method, wherein should determine the step of flashing scene type, greater than this critical value, promptly be judged as the flashing scene of the first kind as continuous two these histogrammic peak values, the detection step that it comprises has:
(a) relatively the lightness difference value in this histogram with should predetermined critical value;
(b) calculate continuous two peak value differences greater than the lightness difference value of this critical value;
(c) relatively this peak value difference and a predetermined lightness percentage; And
(d) detecting is the first kind.
3. digital image flashing scene as claimed in claim 2 detects and removing method, wherein the lightness difference value that is calculated by continuous two pictures forms a peak value, as described in step (a), if this lightness difference value is less than this critical value, promptly more next lightness difference value, if continuous two peak values greater than this critical value, promptly calculate this peak value difference.
4. digital image flashing scene as claimed in claim 2 detects and removing method, wherein as described in the step (c), if this peak value is greater than this lightness percentage, then follow more following two adjacent peak values greater than this critical value, if this peak value difference is less than this lightness percentage, then corresponding picture is a flashing scene.
5. digital image flashing scene as claimed in claim 2 detects and removing method, and wherein this lightness percentage P is
H in this formula
nBe the lightness difference value of adjacent n picture and n+1 picture, H
N+1Be the lightness difference value of adjacent n+1 picture and n+2 picture, max () is for drawing H
nWith H
N+1Maximum.
6. digital image flashing scene as claimed in claim 1 detects and removing method, wherein should determine the step of flashing scene type, continue to surpass a picture as flashing scene, and the picture of two or more continuous generation flashes of light can form two peak values, one specific distance is arranged between two peak values, promptly be judged as the flashing scene of second type, the detection step that it comprises has:
(a) relatively the lightness difference value in this histogram with should predetermined critical value;
(b) calculate two greater than the number of pictures between two peak values of the lightness difference value of this critical value, and compare with a given number;
Peak value difference when (c) calculating two lightness difference value that continue all greater than this critical value;
(d) relatively this peak value difference and a predetermined lightness percentage; And
(e) detecting is second type.
7. digital image flashing scene as claimed in claim 6 detects and removing method, wherein the lightness difference value that is calculated by continuous two pictures forms a peak value, as step (a) and (b), if this lightness difference value is less than this critical value, then more next lightness difference value, then, if two lightness difference value that a plurality of pictures are apart arranged greater than this critical value, then continue to calculate peak value difference.
8. digital image flashing scene as claimed in claim 6 detects and removing method, wherein as described in the step (d), if this peak value difference is greater than this lightness percentage, then continue more following two peak values greater than this critical value, if this peak value difference is less than this lightness percentage, then relative a plurality of pictures are flashing scene between these two peak values.
9. digital image flashing scene as claimed in claim 6 detects and removing method, wherein as described this given number of step (b), if the number of pictures between two peak values is greater than this given number, then continue the step of the more next lightness difference value of step (a) and this critical value, less than this given number, then continue the step that step (c) is calculated peak value difference as if the number of pictures between two peak values.
10. digital image flashing scene as claimed in claim 6 detects and removing method, and wherein this lightness percentage P is
H in this formula
nBe the lightness difference value of adjacent n picture and n+1 picture, H
N+1Be the lightness difference value of adjacent n+1 picture and n+2 picture, max () is for drawing H
nWith H
N+1Maximum.
11. digital image flashing scene as claimed in claim 1 detects and removing method, wherein because the specific picture in the digitized video can cause a plurality of images to divide mirror to change, and wherein Yan Xu picture is in the chain-reacting amount of being scheduled to, promptly be judged as the flashing scene of the 3rd type, the detection step that it comprises has:
(a) relatively the lightness difference value in this histogram with should predetermined critical value, and judge that image wherein divides mirror to change;
(b) detecting next image divides mirror to change;
(c) two images of calculating divide the number of pictures between the mirror variation;
(d) relatively this number of pictures and chain-reacting amount that should be predetermined; And
(e) detect to including the 3rd type of specific picture.
12. digital image flashing scene as claimed in claim 11 detects and removing method, wherein the lightness difference value that is calculated by continuous two pictures forms a peak value, as described in step (a), if this lightness difference value is less than this critical value, then more next lightness difference value, if greater than this critical value, then detecting, the difference value that should become clear have image to divide mirror to change.
13. digital image flashing scene as claimed in claim 11 detects and removing method, wherein as described in the step (d), if this number of pictures is greater than this chain-reacting amount, then the picture shown in this number of pictures is judged as this image and divides specific picture between the mirror variation.
14. digital image flashing scene as claimed in claim 11 detects and removing method, wherein as described in the step (d), divides the mirror variation if this number of pictures, then detects next image less than this chain-reacting amount.
15. digital image flashing scene as claimed in claim 1 detects and removing method, the step of wherein eliminating flashing scene comprises this flashing scene by ignoring in this image sequence.
16. digital image flashing scene as claimed in claim 1 detects and removing method, the step of wherein eliminating flashing scene comprises and being replaced by near the picture mean value the flashing scene of this image sequence.
17. digital image flashing scene as claimed in claim 1 detects and removing method, the step of wherein eliminating flashing scene comprises the flashing scene that recomputates this image sequence.
18. a digital image flashing scene detects and removing method, this method step includes:
Import image sequence data;
Capture a sequential images picture;
Calculate the many groups lightness difference value between this adjacent in twos image frame;
Make a lightness difference histogram according to these many group lightness difference value;
Determine a critical value according to this lightness difference histogram, use the flashing scene that detects this image sequence;
Relatively lightness difference value in this lightness difference histogram and critical value that should be predetermined;
Calculate continuous two peak value differences greater than the lightness difference value of this critical value;
Relatively this peak value difference and a predetermined lightness percentage;
Detect and to be the first kind, when the peak value of continuous two these lightness difference histograms greater than this critical value, promptly be judged as the flashing scene of the first kind; And
Eliminate this flashing scene.
19. a digital image flashing scene detects and removing method, this method step includes:
Import image sequence data;
Capture a sequential images picture;
Calculate the many groups lightness difference value between this adjacent in twos image frame;
Make a lightness difference histogram according to these many group lightness difference value;
Determine a critical value according to this lightness difference histogram, use the flashing scene that detects this image sequence;
Relatively lightness difference value in this lightness difference histogram and critical value that should be predetermined;
Calculate two greater than the number of pictures between two peak values of the lightness difference value of this critical value;
Peak value difference when calculating two lightness difference value that continue all greater than this critical value;
Relatively this peak value difference and a predetermined lightness percentage;
Detection is second type, continues to surpass a picture as flashing scene, and two peak values of picture formation of two or more continuous generation flashes of light, between this two peak value one specific distance is arranged, promptly be judged as the flashing scene of second type; And
Eliminate this flashing scene.
20. a digital image flashing scene detects and removing method, this method step includes:
Import image sequence data;
Capture a sequential images picture;
Calculate the many groups lightness difference value between this adjacent in twos image frame;
Make a lightness difference histogram according to these many group lightness difference value;
Determine a critical value according to this lightness difference histogram, use the flashing scene that detects this image sequence;
Relatively lightness difference value in this lightness difference histogram and critical value that should be predetermined, and judge that image wherein divides mirror to change;
Detecting next image divides mirror to change;
Calculate two images and divide the number of pictures of mirror between changing;
Relatively this number of pictures and chain-reacting amount that should be predetermined;
Detection is the 3rd type, and the specific picture in the factor word image causes a plurality of images to divide mirror to change, and the picture of wherein continuity is in the predetermined chain-reacting amount, promptly is judged as the flashing scene of the 3rd type; And
Eliminate this flashing scene.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100036540A CN100391232C (en) | 2005-01-07 | 2005-01-07 | Digital image flashing scene detecting and eliminating method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100036540A CN100391232C (en) | 2005-01-07 | 2005-01-07 | Digital image flashing scene detecting and eliminating method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1801886A true CN1801886A (en) | 2006-07-12 |
| CN100391232C CN100391232C (en) | 2008-05-28 |
Family
ID=36811619
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100036540A Expired - Fee Related CN100391232C (en) | 2005-01-07 | 2005-01-07 | Digital image flashing scene detecting and eliminating method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100391232C (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101394558B (en) * | 2007-09-20 | 2011-05-11 | 奇扬网科股份有限公司 | Method for detecting and playing imagery zone of computer picture |
| CN101242547B (en) * | 2007-02-05 | 2011-07-20 | 凌阳科技股份有限公司 | System for detecting bad editing of video signals |
| CN102473282A (en) * | 2010-06-17 | 2012-05-23 | 松下电器产业株式会社 | External light glare assessment device, line of sight detection device and external light glare assessment method |
| CN106548763A (en) * | 2015-09-22 | 2017-03-29 | 中兴通讯股份有限公司 | A kind of method for displaying image and device and terminal |
| CN117156210A (en) * | 2023-02-07 | 2023-12-01 | 荣耀终端有限公司 | Method and device for detecting splash screen |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2915476B2 (en) * | 1990-03-29 | 1999-07-05 | キヤノン株式会社 | Image processing method |
| US5751378A (en) * | 1996-09-27 | 1998-05-12 | General Instrument Corporation | Scene change detector for digital video |
| KR100213094B1 (en) * | 1997-02-21 | 1999-08-02 | 윤종용 | Histogram and cumulative distribution function(cdf) extracting method and circuit for image enhancing apparatus |
| AU5793199A (en) * | 1998-09-29 | 2000-04-17 | General Instrument Corporation | Method and apparatus for detecting scene changes and adjusting picture coding type in a high definition television encoder |
| US6512850B2 (en) * | 1998-12-09 | 2003-01-28 | International Business Machines Corporation | Method of and apparatus for identifying subsets of interrelated image objects from a set of image objects |
| KR20010081894A (en) * | 2000-02-18 | 2001-08-29 | 구자홍 | Multimedia Query System And Histogram Converting System Based On Contents |
| US7176976B2 (en) * | 2001-08-30 | 2007-02-13 | Lightsurf Technologies, Inc. | Autoexposure methodology in a digital camera |
| KR100850935B1 (en) * | 2001-12-27 | 2008-08-08 | 주식회사 엘지이아이 | Apparatus for detecting scene conversion |
| KR20040006065A (en) * | 2002-07-09 | 2004-01-24 | 삼성전자주식회사 | Scene change detector and method thereof |
-
2005
- 2005-01-07 CN CNB2005100036540A patent/CN100391232C/en not_active Expired - Fee Related
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101242547B (en) * | 2007-02-05 | 2011-07-20 | 凌阳科技股份有限公司 | System for detecting bad editing of video signals |
| CN101394558B (en) * | 2007-09-20 | 2011-05-11 | 奇扬网科股份有限公司 | Method for detecting and playing imagery zone of computer picture |
| CN102473282A (en) * | 2010-06-17 | 2012-05-23 | 松下电器产业株式会社 | External light glare assessment device, line of sight detection device and external light glare assessment method |
| CN102473282B (en) * | 2010-06-17 | 2015-01-14 | 松下电器（美国）知识产权公司 | External light glare assessment device, line of sight detection device and external light glare assessment method |
| CN106548763A (en) * | 2015-09-22 | 2017-03-29 | 中兴通讯股份有限公司 | A kind of method for displaying image and device and terminal |
| CN117156210A (en) * | 2023-02-07 | 2023-12-01 | 荣耀终端有限公司 | Method and device for detecting splash screen |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100391232C (en) | 2008-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102685398B (en) | News video scene generating method | |
| CN1207924C (en) | Method for testing face by image | |
| US9224211B2 (en) | Method and system for motion detection in an image | |
| CN101042735A (en) | Image binarization method and device | |
| CN1272747C (en) | Method and device for tracking moving object in image | |
| US9747507B2 (en) | Ground plane detection | |
| CN112104869B (en) | Video big data storage and transcoding optimization system | |
| CN1801886A (en) | Digital image flashing scene detecting and eliminating method | |
| CN113177438B (en) | Image processing method, device and storage medium | |
| CN106934374A (en) | The recognition methods of traffic signboard and system in a kind of haze scene | |
| US11948373B2 (en) | Automatic license plate recognition | |
| CN108040244B (en) | Snapshot method and device based on light field video stream and storage medium | |
| CN116249015A (en) | Camera shielding detection method and device, camera equipment and storage medium | |
| CN116012764A (en) | Road disease identification and positioning method and device, electronic equipment and storage medium | |
| CN110378934A (en) | Subject detection method, apparatus, electronic equipment and computer readable storage medium | |
| CN112884805A (en) | Cross-scale self-adaptive mapping light field imaging method | |
| CN1294532C (en) | Text discrimination method and related apparatus | |
| JP5820284B2 (en) | Image processing apparatus, method, and program | |
| CN1299242C (en) | Imaging processing equipment for distinguishing mobile target profile and method thereof | |
| CN112581489A (en) | Video compression method, device and storage medium | |
| CN116664829A (en) | RGB-T semantic segmentation method, system, device and storage medium | |
| CN1875379A (en) | Object tracking within video images | |
| CN116188826A (en) | Template matching method and device under complex illumination condition | |
| CN115761580A (en) | Detection method of water surface floater, edge calculation terminal and storage medium | |
| CN1811794A (en) | Greyscale image partition method based on light distribution character of reflection or penetrance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080528 Termination date: 20200107 |