EP1634245A2 - Verfahren zur ermittlung eines verschiebungsvektors in der bildverarbeitung - Google Patents
Verfahren zur ermittlung eines verschiebungsvektors in der bildverarbeitungInfo
- Publication number
- EP1634245A2 EP1634245A2 EP04736747A EP04736747A EP1634245A2 EP 1634245 A2 EP1634245 A2 EP 1634245A2 EP 04736747 A EP04736747 A EP 04736747A EP 04736747 A EP04736747 A EP 04736747A EP 1634245 A2 EP1634245 A2 EP 1634245A2
- Authority
- EP
- European Patent Office
- Prior art keywords
- vector
- vectors
- test
- image
- prediction
- 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.)
- Ceased
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/223—Analysis of motion using block-matching
- G06T7/238—Analysis of motion using block-matching using non-full search, e.g. three-step search
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/97—Determining parameters from multiple pictures
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20228—Disparity calculation for image-based rendering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/144—Movement detection
Definitions
- the present invention relates to a method for determining a displacement vector in image processing.
- Displacement vectors or motion vectors are used in image processing in a well-known manner to:
- the two images can be recorded sequentially in time, and thus illustrate a movement of the object or image area in a displayed image section over time.
- the motion vector is used in such cases, for example for interpolating one or more intermediate frames, the position of the object: to reflect one or more time points between the first and the second image recording, for example, in the •. 50Hz / 100Hz image conversion or when creating so-called "artificial slow-motion".
- the two images can also be taken simultaneously from a first and second camera position, the displacement vector to be determined being used to generate an intermediate image that shows the object from a virtual third camera position that is at a different position from the first and second Camera position.
- FIG. 1 illustrates two images B1, B2, which can be recorded successively in time or from different camera positions, with a first image area 1 at a first position P1 in the first image B1 and a second image area 2 at a second position P2 in the second picture B2.
- the image areas 1, 2 are shown as squares, the center points of which denote the positions P1, P2 of the image areas.
- the image areas can have any shape.
- the second image area 2 in the second image B2 results from a displacement of the first image area 1 of the first image B1 with a displacement vector vec.
- This shift can be due to time when the two images B1, B2 are taken in succession and an object representing the first image area 1 is located in the viewed image section from the position P1 to the recording time of the first image B1 to the position 2 to the recording time of the second image B2 moves.
- This shift of the two image areas 1, 2 in the image detail shown can also be caused by the fact that the two images B1, B2 were taken at the same time, but from different angles.
- test vectors each of which represents a possible displacement vector for the individual image areas.
- image comparisons are carried out by comparing image information, for example luminance values or chrominance values, of the individual pixels of a viewed first image area in the first image with image information values of a second image area in the second image, the position of the second image area in the second image using a selection vector in each case in relation to the position of the first image area in the first image.
- selection vectors those are selected as so-called selection vectors and used for the further image processing, which have the "best" method. provides the same result, in which the best match between the first and second image areas was determined.
- the quality of such a method depends on the selection of the selection vectors, since, of course, an image comparison cannot be carried out for all possible displacement vectors in order to keep the already high computing effort within limits. It is a basic goal in such methods to reduce the number of image comparisons required without accepting a deterioration in the result when determining the at least one selection vector.
- Prediction vectors from which the test vectors are formed or selected are adopted as a test vector for the area of interest and / or the test vectors can be derived from these prediction vectors by changing their length and / or direction a predefined or random scheme can be formed, in which case the prediction vectors can change from search step to search step in the latter case.
- FIG. 2 shows schematically a vector search system which determines displacement vectors for individual image areas and which, to determine a possible displacement vector, includes prediction vectors which contain results from previous search steps for the same image area or results from search steps for adjacent image areas.
- the system comprises a test vector determination unit 10, to which a set of prediction vectors are supplied and which provides a set of test vectors from these prediction vectors.
- the test vectors can be made available in that some or all of the prediction vectors are passed on unchanged as test vectors. It is also possible to change some or all of the prediction vectors according to a fixed or random scheme, in order to thereby form the test vectors.
- test vectors are used in an image comparison unit 20 arranged downstream of the test vector determination unit 10 to carry out image comparisons for the individual image areas, with reference to the example according to FIG. 1 always comparing an image area 1 from the first image B1 with an image area 2 from the second image B2, whereby For each comparison, such a second image area is selected from the second image B2, the geometry of which corresponds to the geometry of the first image area 1 and which is shifted relative to the first image area 1 of the first image B1 by means of one of the selection vectors.
- the results determined for the individual comparisons are fed to a selection unit 30, which uses the comparison results to select at least one selection vector from the group of test vectors for which the best comparison result was determined.
- This at least one vector selected on the basis of the comparison result represents the actual displacement vector or at least the most probable displacement vector which transfers the first image area 1 from the first image B1 to the second image area 2 in the second image B2.
- the selection vectors are always a function of the prediction vectors.
- a sensible use of these prediction vectors presupposes, however, that the shift of neighboring image areas or the shift of the image area of interest is not subject to strong fluctuations over time, since it must be ensured that the selection vectors determined from the prediction vectors also contain the actual displacement vector. Otherwise, a significant increase in the number of selection vectors is required.
- the aim of the present invention is to provide an improved method for determining at least one selection vector in image processing, which is in particular able to take rapid changes of displacement vectors into account.
- the method according to the invention for determining a selection vector which indicates a shift of an image area from a first position in a first image to a second position in a second image, provides for a set of prediction vectors and a set of test vectors to be made available.
- the prediction vectors can be conventional methods for generating prediction vectors can be obtained, for example by using information from the search of displacement vectors to neighboring image areas or from the previous search of displacement vectors to the image area of interest.
- An image comparison is carried out using selected test vectors from the set of test vectors in order to provide an image comparison result for each of these selected test vectors.
- the selected test vectors and at least one of the prediction vectors are compared to provide at least one vector comparison result for each of the selected test vectors.
- the image comparison result determined for a selected test vector and the at least one vector comparison result determined for a selected test vector are then linked in order to provide at least one quality value for each of the selected test vectors.
- a ranking order is determined for these quality values, at least one test vector being selected as the displacement vector on the basis of the ranking order determined.
- test vectors are fixed and cover different possible shifts in the image area of interest within a given image grid.
- the information contained in the prediction vectors is taken into account in the method according to the invention in that when one of the test vectors is selected as the displacement vector, the distance between the test vector and the prediction vectors is also determined in addition to the image comparison result obtained on the basis of the test vector.
- the image comparison result for the selected test vector can be determined using conventional methods, in that image information of the pixels of a first image area in the first image with image information values of the pixels of a second image area in the second image are compared, the position of the second image area being shifted relative to the position of the first image area by the test vector under consideration. If these two image areas are superimposed, each pixel of the first image area, which is located at a given position in this image area, is assigned a pixel of the second image area, which is located at the corresponding position in the second image area, assuming geometrically identical image areas.
- the determination of the comparison result can include the comparison of the image information values of all pixels in the two image areas or the comparison of the image information values of selected, representative pixels, the individual pixel comparison results being added up in order to obtain the image comparison result.
- a pixel comparison result for a pixel of the first image area and an associated pixel of the second image area can be obtained, for example, by
- the image comparison is carried out in such a way that for each comparison of two image areas a numerical value is supplied which, for example, the smaller the greater the correspondence between the compared image areas.
- the amount or the square of a difference vector of these two vectors is determined, for example.
- a numerical value is obtained for each comparison of a test vector with a prediction vector, the smaller the less the test vector and the prediction vector differ from one another.
- the prediction vector can be a displacement vector that was determined for adjacent image areas of the image area of interest in the same image, or the prediction vector can be a displacement vector that was determined for the image area of interest or adjacent image areas in a previous image.
- the vector comparison result there is, for example, the possibility of "punishing" a prediction vector which was determined on the basis of a previous image, that is to say output a worse vector comparison result for this vector than for an identical vector which determines on the basis of an adjacent image area in the same image has been.
- the punishment can take place, for example, by multiplying the determined comparison result by a multiplication factor, this multiplication factor being adjustable from the outside and, for example, determining the type of generation of the prediction vector.
- a value dependent on the desired level of punishment can also be added to the comparison result determined.
- the punishment relates to all comparison results of this prediction vector with the test vectors.
- other criteria for punishment can also be used. For example, it is possible to use the absolute length of a test vector as a punishment criterion and to punish test vectors with a large absolute length if, for example, it can be assumed that the movement sequences occurring in the images are slow.
- the structure of the image in the determination of a suitable selection vector. If, for example, self-similar structures, i.e. monotonous surfaces, are present, it may be useful to obtain selection vectors that differ as little as possible from the prediction vectors previously determined. If the result of the vector comparison is a numerical value, the greater the more a test vector deviates from a prediction vector, this goal can be achieved in that each vector comparison result is weighted with a value greater than one, resulting in test vectors which In absolute terms, "deviating" vector comparison results are considered when deviating strongly from the prediction vectors. Self-similar structures can be determined by an image comparison - usually a block-wise image comparison - within the image.
- the quality value of a test vector in relation to a prediction vector this is determined for the test vector.
- Half the image comparison result and the vector comparison result are linked to one another, the linkage being such that, with the same image comparison result, the quality value is the better the less the test vector and the selected prediction vector differ.
- the origin of a prediction vector can be taken into account in the image comparison, for example, by providing the comparison result with an offset, the value of which depends on the origin of the respective prediction vector.
- the quality value can be determined by a weighted addition of the numerical value obtained from the image comparison and the value obtained from the vector comparison result, it being possible to determine several quality values for a test vector, the number of which corresponds to the number of prediction vectors with which the selected test vector is compared becomes.
- vector comparisons are carried out between the selected test vectors and all prediction vectors in order to obtain a number of vector comparison results corresponding to the number of prediction vectors for each of the selected test vectors.
- a ranking of the vector comparison results is determined and only the "best" vector comparison result is used to form the quality value, so that exactly one quality value is determined for each test vector.
- each prediction vector determines the best test vector for this prediction vector, wherein, for example, the test vector (s) is / are selected as the selection vector which is most frequently the best test vector for the selection vectors were determined. It is also possible to select a separate set of test vectors for each prediction vector.
- All test vectors are preferably selected, that is to say image comparisons are carried out for all test vectors, wherein vector comparisons with all prediction vectors can additionally be carried out for all test vectors.
- the selection of the at least one test vector as a selection vector is preferably carried out iteratively by first providing a first set of fixed selection vectors, from which the "best" test vector (s) are selected on the basis of the ranking determined by means of the image comparison results and the vector comparison results. From these best test vectors, a second group of test vectors is subsequently determined for the next iteration step, the test vectors of this second group being formed from the previously selected best test vectors of the first set.
- test vectors of the second group are formed, for example, by adding predetermined modification vectors to the previously selected test vectors of the first set.
- One or more test vectors of the second group can be generated for one of the selected test vectors.
- the modification vectors are preferably chosen such that the difference between such test vectors, which result from adding the modification vectors to one of the previously selected test vectors, is smaller than the deviations of the individual test vectors of the first group.
- a further ranking is formed for the test vectors of the second group, based on which at least one test vector used to form a selection vector is selected or on the basis of which one or more test vectors are selected, from which a further group of test vectors is formed by means of modification vectors for a further iteration step.
- prediction vectors are divided into groups with regard to their origin, for example prediction vectors that were determined on the basis of a previous image for adjacent image regions to the image region of interest, prediction vectors that were determined on the basis of a previous image for the image region of interest and prediction vectors, which were determined on the basis of the current image for adjacent image areas to the image area of interest.
- the best test vector (s) can be determined for each of these groups in each iteration step and used for the subsequent iteration step.
- the prediction vector determined for a group in addition to the determination of a selection vector based on the ranking of the best test vectors, there is the possibility of storing the best test vector determined for a group as a new prediction vector.
- the modification vectors preferably comprise the zero vector, so that the test vectors selected at the end of an iteration step, from which the modification vectors are used the set of test vectors for the next iteration step are formed, are contained in this set of test vectors for the next iteration step.
- the direction of the individual modification vectors preferably coincide in the individual iteration steps, but their amount becomes smaller from iteration step to iteration step, so that test vectors are generated with each iteration step, the difference of which is smaller.
- test vectors are selected depending on the quality measure determined previously. If, for example, the best test vector is determined for each selection vector, one or more of which, however, provides a poor quality measure, then in this embodiment these test vectors are subjected to major modifications, for example, since it can be assumed that minor modifications should not bring about any significant improvement , Test vectors, on the other hand, which give a good result, are only slightly or not modified during the subsequent step.
- such an iterative method is based on a set of test vectors, which is formed from the original set of test vectors by all possible combinations with the modification vectors, whereby test vectors are repeatedly rejected during the individual iteration steps in order to keep the computational effort within limits.
- the method according to the invention for determining a selection vector can be used in a method for determining a set of selection vectors, in which selection vectors are determined exclusively on the basis of the method explained, with as many test vectors as selection being selected on the basis of the determined ranking of the quality values - vectors are required.
- the method can also serve as an extension of a known method, in which one or more prediction vectors are used as selection vectors, and which is improved by using at least one test vector selected on the basis of a ranking of quality values as the selection vector.
- the selection vectors can serve as prediction vectors for the next vector search step of the image area under consideration, it being possible for the selection vectors to be adopted unchanged as prediction vectors or to be changed in accordance with a predetermined or random scheme for the formation of the new prediction vectors.
- the selection vectors of the previous step which deliver the worst image comparison result, are preferably no longer used as new prediction vectors but replaced by the test vectors determined on the basis of the ranking.
- FIG. 1 schematically shows a first and a second image, each with an image area which are related to one another via a displacement vector
- FIG. 2 shows a system for providing a set of selection vectors using prediction vectors according to the prior art
- FIG. 3 shows a flow chart for a method according to the invention for providing at least one selection vector using a set of test vectors and a set of prediction vectors
- FIG. 4 tables which comprise image comparison results and vector comparison results which are linked to quality values for selected test vectors
- FIG. 5 shows a quality value matrix
- FIG. 6 shows a flowchart for a further embodiment of a method according to the invention, in which the at least one selection vector is iteratively selected from a sentence
- FIG. 7 schematically shows a possible procedure for an iterative method for providing a set of test vectors on the basis of a selected test vector and on the basis of modification vectors according to FIG. 6.
- FIG. 3 uses a flow chart to illustrate an exemplary embodiment of a method according to the invention for determining at least one selection vector which, with reference to the example according to FIG. 1, shows a possible displacement vector for a displacement of the first image area 1 from the first position P1 in the first image B1 represents second position P2 in the second image B2.
- selection vectors serve to make image comparisons between image areas in the first
- Image B1 and the second image B2 are carried out, the compared image areas being shifted relative to one another by one of the selection vectors.
- the selection vector which delivers the "best" image comparison result is assumed to be the actual or most probable displacement vector.
- the method according to the invention for providing at least one selection vector provides for a set of prediction vectors 101 to be provided, with four such prediction vectors V1, V2, V3, V4 being shown by way of example in FIG.
- These prediction vectors V1-V4 can be provided by means of conventionally known methods for determining prediction vectors. Such methods include, for example, the provision of prediction vectors for a given image area on the basis of displacement vectors already determined for neighboring image areas of the given image area.
- the method according to the invention further provides one
- test vectors 102 To provide a set of test vectors 102, three such test vectors T1, T2, T3 being shown by way of example in FIG.
- This set of test vectors 102 becomes at least one
- Test vector selected preferably all test vectors T1-T3 being selected, and using these test vectors, image comparisons are carried out in a step 103 in order to provide an image comparison result for each of the selected test vectors T1-T3, which is denoted by B (Tn) in FIG. 3 where Tn stands for any selected test vector T1-T3.
- the image comparison comprises the comparison of a first image area 1 in the first image B1 with a second image area 2 in the second image B2, a position P2 of the second image area B2 relative to the position P1 of the first image area 1 by that of the selected one Test vectors for which the image comparison result is determined is shifted.
- the displacement vector vec is selected by a different one for each image comparison
- Test vectors T1-T3 replaced.
- image areas built up from a number of image points are compared in that image information values, for example luminance values or chrominance values, of two image points in each case, which are located in the first image area 1 and in the second image area 2 at corresponding image point positions, are compared, the individual comparison results be added up.
- image information values for example luminance values or chrominance values
- the squares of all differences between the image information values of the pixels of the first image area 1 and the pixels of the second image area 2 are added, whereby an image comparison result is obtained, the numerical value of which is smaller, the greater the correspondence between the two viewed image areas are 1, 2 and in extreme cases this is zero if the two viewed image areas are identical, that is, if the two image areas match pixel by pixel.
- the number of image comparison results B (Tn) provided by the image comparison step 103 corresponds to the number of selected test vectors T1-T3, all of which are preferably selected.
- the selected test vectors T1-T3 are compared with at least one of the prediction vectors V1-V4 in order to provide vector comparison results V (Tn, Vm).
- V (Tn, Vm) denotes a comparison result which is determined by comparing the prediction vector Vm with the test vector Tn, Tn standing for any of the selected test vectors T1-T3 and Vm for any of the selected prediction vectors V1-V4.
- All prediction vectors Vl-V4 are preferably used for a comparison with all test vectors T1-T3, which results in a number of vector comparison results V (Tn, Vm) which corresponds to the product of the number of test vectors and the number of prediction vectors.
- the comparison of a test vector Tn with a prediction vector Vm includes, for example, determining the square of the amount of a difference vector between the test vector Tn and the prediction vector Vm.
- the vector comparison results V (Tn, Vm) and the image comparison results B (Tn) are linked in a method step 105 to quality values G (Tn, Vm).
- G (Tn, Vm) in FIG. 3 denotes a quality value, which is obtained by linking the image comparison result B (Tn) with the vector comparison result
- V (Tn, Vm) was obtained.
- the link operator is represented in FIG. 6 by the symbol "o".
- the linkage of the vector comparison result V (.) And B (.) Can comprise different linkage types depending on the way in which the image comparison result B (.) And V (.) Are obtained.
- the image comparison result B (.) Represents a numerical value, the smaller the larger the correspondences between the viewed image areas are, and that the vector comparison result V (.) Represents a numerical value, the value of which is the smaller, each less the difference between the considered test vector and the considered prediction vector
- the link between the image comparison result B (.) and the vector comparison result V (.) is, for example, a weighted addition, so that:
- FIG. 4 illustrates a table 113 generated in method step 103 with image comparison results B (T1) to B (T3) and a table determined in method step 104 with vector comparison results V (T1, V1) to V (T3, V4), which are shown in FIG the method step 105 to form a table 115 with the quality values G (Tl, Vl) to G (T3, V4).
- a ranking is determined by sorting these quality values G (.,.) And at least the test vector for which the best quality value was determined is selected as the selection vector.
- the selected selection vector (s) are designated by Ti in FIG. 3.
- FIG. 5 illustrates this procedure using a matrix in which the test vectors T1-T3 and the selection vectors VI-V4 are plotted, the individual matrix elements each representing, for example, a quality value associated with a test vector and a prediction vector.
- the quality value is determined such that a test vector is considered to be the better
- the lower its quality value in the example the test vector Tl as the best vector would become the prediction vector Vl
- the test vector Tl as the best vector for the prediction vector V2
- the test vector T3 as the best vector for the prediction vector V3
- the test vector T2 as the best vector for the prediction vector V4.
- test vector or several test vectors can then be selected from these best test vectors, with this selection taking into account the quality measure determined for the respective test vector and / or the frequency with which a test vector was determined as the best test vector for the prediction vectors.
- the test vector determined for a selection vector can be stored as a new prediction vector, it being possible to classify the selection vectors according to their origin and to store the test vectors obtained accordingly in an associated group.
- test vectors 102 are provided, from which, after image comparisons have been carried out and prediction vectors used, at least one test vector Ti is selected, which can be used as a selection vector.
- At least one selection vector is determined by means of an iterative method.
- this method also provides for a set of test vectors T1-T3 to be provided and at least one test vector Ti from this set of test vectors T1-T3 by means of a method already explained with reference to FIGS. 3 and 4 select.
- this test vector Ti is not output as a selection vector after the first run of the method, but this at least one test vector Ti serves in a method step 107 as the basis for generating a further set of test vectors, which are shown in FIG 6 are designated by way of example with TiI, Ti2, Ti3, Ti4, the method already explained with this new set of test vectors being explained in FIG.
- switches S1, S2 are shown in FIG. 6, the switch S1 being in the position shown on the left at the beginning of the method, to change the original set 102 of test vectors T1, T2, T3 in to process step 103 and wherein switch S2 is in the right one at the beginning of the process position is shown in order to feed the at least one selected test vector Ti to method step 107 for forming a new set 112 of test vectors TiI-Ti4.
- the first switch S1 is in the position shown on the right in order to feed the test vectors of the set of test vectors formed from the selected test vectors Ti to method step 103.
- the second switch is moved to the position shown on the left in order to output the at least one selected test vector Ti as a selection vector.
- FIG. 7 illustrates a possible embodiment of method step 107, in which the new set of test vectors 112 is made available from at least one selected test vector Ti.
- FIG. 7 shows a selected test vector Ti in a Cartesian coordinate system, a set of four further test vectors TiI, Ti2, Ti3, Ti4 being generated from this test vector Ti by vectorial addition of four modification vectors M1-M4.
- Modification vectors a zero vector, so that the selected test vector Ti is also adopted in the set of new test vectors 112.
- the modification vectors M1-M4 shown can be represented in Cartesian coordinates as ( ⁇ a, 0) and (0, ⁇ a), where a represents the amount of the modification vectors M1-M4.
- more than four modification vectors can also be used in order to thereby more than a selected test vector Ti . four, or when using the zero vector as a modification vector to form more than five test vectors from a selected test vector Ti.
- Additional modification vectors are, for example ( ⁇ 0.5 • a • V2, ⁇ 0.5 • a • V2).
- the amount a of the modification vectors preferably decreases with each iteration step, as a result of which one set is selected from a selected test vector Ti with each iteration step is generated by test vectors that differ less and less from the selected test vector Ti.
- the best test vector determined for each prediction vector can be stored as a new prediction vector.
- test vector selected for the next iteration step is preferably modified as a function of the quality value determined during the previous iteration step. For example, test vectors with a good quality value are modified less strongly than test vectors with poor quality values. This can be achieved, for example, by setting the amount of the modification vector as a function of the previously determined quality value.
- the method according to the invention for determining a selection vector on the basis of a set of test vectors and a set of prediction vectors is completely suitable for determining selection vectors, for example providing a set of test vectors which contains more test vectors than are required for selection vectors and which of the determined method, the best test vectors are selected as selection vectors.
- the method according to the invention is also suitable in connection with a conventional method in which selection vectors are determined on the basis of prediction vectors. There is the possibility, in addition to the prediction vectors, of at least one test vector as
- selection vector which was determined using the inventive method.
- the selection vectors determined for a given image area can be used for subsequent search steps for the determination of the prediction vectors.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE10327577A DE10327577A1 (de) | 2003-06-18 | 2003-06-18 | Verfahren zur Ermittlung eines Verschiebungsvektors in der Bildverarbeitung |
| PCT/EP2004/006398 WO2004111939A2 (de) | 2003-06-18 | 2004-06-14 | Verfahren zur ermittlung eines verschiebungsvektors in der bildverarbeitung |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP1634245A2 true EP1634245A2 (de) | 2006-03-15 |
Family
ID=33520721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP04736747A Ceased EP1634245A2 (de) | 2003-06-18 | 2004-06-14 | Verfahren zur ermittlung eines verschiebungsvektors in der bildverarbeitung |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US7720296B2 (de) |
| EP (1) | EP1634245A2 (de) |
| DE (1) | DE10327577A1 (de) |
| WO (1) | WO2004111939A2 (de) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8588513B2 (en) * | 2005-07-18 | 2013-11-19 | Broadcom Corporation | Method and system for motion compensation |
| EP1841231A1 (de) * | 2006-03-29 | 2007-10-03 | Sony Deutschland Gmbh | Verfahren zur Bewegungsschätzung |
| DE102007051175B4 (de) | 2007-10-25 | 2012-01-26 | Trident Microsystems (Far East) Ltd. | Verfahren zur Bewegungsschätzung in der Bildverarbeitung |
| DE102007051174B4 (de) * | 2007-10-25 | 2011-12-08 | Trident Microsystems (Far East) Ltd. | Verfahren zur Bewegungsschätzung in der Bildverarbeitung |
| CN101971209A (zh) * | 2008-01-11 | 2011-02-09 | 卓然(法国) | 超高分辨率视频处理的稀疏几何结构 |
| KR101538654B1 (ko) * | 2008-07-07 | 2015-07-22 | 삼성전자주식회사 | 손 떨림 보정 동작을 제어하는 디지털 촬영장치 및 이의제어방법 |
| JP5075757B2 (ja) * | 2008-08-05 | 2012-11-21 | オリンパス株式会社 | 画像処理装置、画像処理プログラム、画像処理方法、および電子機器 |
| AT505913B8 (de) * | 2008-08-22 | 2009-06-15 | Arc Austrian Res Centers Gmbh | Verfahren zur ermittlung der lage von bildpunkten |
| US8659600B2 (en) * | 2011-07-13 | 2014-02-25 | Pixar | Generating vector displacement maps using parameterized sculpted meshes |
| US9227563B2 (en) * | 2012-09-14 | 2016-01-05 | Bendix Commercial Vehicle Systems Llc | Backward movement indicator apparatus for a vehicle |
| DE102014210739A1 (de) * | 2014-06-05 | 2015-12-17 | Robert Bosch Gmbh | Verfahren zur Kalibrierung eines Drehratensensors und elektrisches Gerät |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FI89121C (fi) * | 1991-11-11 | 1993-08-10 | Nokia Oy Ab | Foerfarande foer soekning av roerelsevektorer foer en videosignal |
| KR950009699B1 (ko) * | 1992-06-09 | 1995-08-26 | 대우전자주식회사 | 움직임벡터 검출방법 및 장치 |
| WO1999016251A1 (en) * | 1997-09-23 | 1999-04-01 | Koninklijke Philips Electronics N.V. | Motion estimation and motion-compensated interpolation |
-
2003
- 2003-06-18 DE DE10327577A patent/DE10327577A1/de not_active Ceased
-
2004
- 2004-06-14 EP EP04736747A patent/EP1634245A2/de not_active Ceased
- 2004-06-14 WO PCT/EP2004/006398 patent/WO2004111939A2/de not_active Ceased
- 2004-06-14 US US10/561,658 patent/US7720296B2/en not_active Expired - Fee Related
Non-Patent Citations (1)
| Title |
|---|
| See references of WO2004111939A2 * |
Also Published As
| Publication number | Publication date |
|---|---|
| US7720296B2 (en) | 2010-05-18 |
| WO2004111939A3 (de) | 2005-09-09 |
| DE10327577A1 (de) | 2005-01-13 |
| US20070098279A1 (en) | 2007-05-03 |
| WO2004111939A2 (de) | 2004-12-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE69031055T2 (de) | Bewegungsabhängige Videosignalverarbeitung | |
| DE69812882T2 (de) | Verfahren zur bewegungskompensierten Mehrhalbbildverbesserung von Videostandbildern | |
| DE69030896T2 (de) | Bewegungsabhängige Videosignalverarbeitung | |
| DE69021431T2 (de) | Bewegungsabhängige Videosignalverarbeitung. | |
| DE69126930T2 (de) | Video-Bewegungsvektorabschätzung mit asymmetrischem Aktualisierungsgebiet | |
| DE69015419T2 (de) | Umwandlung eines Videosignals auf einen photographischen Film. | |
| DE69421139T2 (de) | Bewegungsvektordetektion | |
| DE69423166T2 (de) | Verfahren und Vorrichtung zur Bewegungsauswertung mit Blockübereinstimmung | |
| DE69028792T2 (de) | Bewegungsabhängige Videosignalverarbeitung | |
| DE102004017145B4 (de) | Verfahren und Vorrichtung zur Ermittlung von Bewegungvektoren, die Bildbereichen eines Bildes zugeordnet sind | |
| EP1634245A2 (de) | Verfahren zur ermittlung eines verschiebungsvektors in der bildverarbeitung | |
| DE69031144T2 (de) | Bewegungsabhängige Videosignalverarbeitung | |
| DE69031322T2 (de) | Bewegungsabhängige Videosignalverarbeitung | |
| EP1770635B1 (de) | Iteratives Verfahren zur Interpolation von Bildinformationswerten | |
| DE4400645A1 (de) | Verfahren und Schaltung zur Interpolation digitaler Bilddaten | |
| EP0897247A2 (de) | Verfahren zur Berechnung von Bewegungsvektoren | |
| DE102009026983A1 (de) | Bildverarbeitungsverfahren mit einer Bewegungsschätzung und Bildverarbeitungsanordnung | |
| DE3000918C2 (de) | Korrelationsvorrichtung | |
| EP2105883A1 (de) | Verfahren zum Testen eines Bewegungsvektors | |
| DE19648017A1 (de) | Verfahren zur Bestimmung von Maßstabsfaktoren für fraktale Bildkodierung | |
| DE602005000425T2 (de) | Verfahren und Anordnung zur Erzeugung von Kandidatenvektoren für Bildinterpolierungssysteme, die Bewegungsabschätzung und -kompensation verwenden | |
| EP3465608B1 (de) | Verfahren und vorrichtung zum bestimmen eines übergangs zwischen zwei anzeigebildern, und fahrzeug | |
| DE69132133T2 (de) | Verfahren und Gerät zur Bewegungserfassung | |
| EP1976288B1 (de) | Iteratives Verfahren zur Interpolation von Bildinformationswerten | |
| EP1605405A1 (de) | Verfahren zur Ermittlung von Bewegungsvektoren, die Bildbereichen eines Bildes zugeordnet sind |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| 17P | Request for examination filed |
Effective date: 20051124 |
|
| AK | Designated contracting states |
Kind code of ref document: A2 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LI LU MC NL PL PT RO SE SI SK TR |
|
| AX | Request for extension of the european patent |
Extension state: AL HR LT LV MK |
|
| DAX | Request for extension of the european patent (deleted) | ||
| RBV | Designated contracting states (corrected) |
Designated state(s): DE FR GB IT NL |
|
| RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: TUSCHEN, CHRISTIAN Inventor name: HAHN, MARKO |
|
| 17Q | First examination report despatched |
Effective date: 20080415 |
|
| RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: TRIDENT MICROSYSTEMS (FAR EAST) LTD. |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION HAS BEEN REFUSED |
|
| 18R | Application refused |
Effective date: 20110129 |