CN1801060A - Method and apparatus for detecting image shift - Google Patents
Method and apparatus for detecting image shift Download PDFInfo
- Publication number
- CN1801060A CN1801060A CNA2004101031766A CN200410103176A CN1801060A CN 1801060 A CN1801060 A CN 1801060A CN A2004101031766 A CNA2004101031766 A CN A2004101031766A CN 200410103176 A CN200410103176 A CN 200410103176A CN 1801060 A CN1801060 A CN 1801060A
- Authority
- CN
- China
- Prior art keywords
- image
- vector
- motion
- sampling
- block
- 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)
Abstract
The invention discloses an image detecting method, which comprises the following steps: reading a reference image; extracting a reference block about the reference image; reading a sampling image; extracting a replacing block instead of the reference block when the reference block doesn't reach the boundary of reference image; finding a sampling block to match the reference block according to the sampling image; counting the moving vector; adapting the sampling image instead of reference image when the sampling block reaches the boundary of sampling image; extracting other reference block about the reference image; rereading the sample image.
Description
Technical field
The invention relates to method and device thereof that a kind of detecting image moves, and the method and the device thereof that particularly move relevant for a kind of detecting image that increases amount of movement and low-angle detecting.
Background technology
Along with the significantly progressive and world-wide web of computing power, the high development of multimedia technology, the processing of image information at present transfers digital transmission to by simulation mostly, and in order to cooperate modern life pattern, computer peripheral device day by day be tending towards designing brief frivolous, action control is accurate and powerful.With the mouse is example, and existing mouse adopts the mechanism of spin as running fix, though the structure assembling is simple, has also to tire easily dust stratification dirt regularly cleaning and the low shortcoming of positioning resolution.The optical mouse of coming out is promptly significantly improved above-mentioned shortcoming recently, and precisely the elements of a fix reduce error.
Tradition is to utilize block relative method (block match method) to judge the amount of movement of image detecting device as the employed detecting image mobile technology of optical mouse.The block relative method can compare the amount of movement of trying to achieve the image detecting device according to sampling block and the reference block that comparison function (for example mean square deviation MSE (mean squared error) or absolute method of average MAD (mean absolute difference)) extracted.
Please refer to Fig. 1, utilize traditional block relative method to judge that the step of amount of movement of image detecting device is as follows: the image detecting device extracts a reference picture (as step S10) when primary importance, and extracts the picture (as step S12) of taking a sample when detector moves to the second place separately.If the above-mentioned reference picture is 6 * 6 picture with the sampling picture, extract 2 * 2 less reference block in reference picture, be divided into 9 nonoverlapping picture blocks, and search with the block relative method in aforementioned sampling picture with each picture block in regular turn, use and obtain the picture identical with the GTG value of picture block.Calculate the related function (as step S13) between several picture blocks and sampling picture, and select reckling calculating displacement and motion-vector (as step S14) between the two.Judge then whether the picture block reaches image boundary (as step S16).If reach image boundary, the picture of then will taking a sample replaces former reference picture and is replaced as new reference picture (as step S18).In fact, the image detecting device itself has many noise sources, except itself manufacture of semiconductor change, power supply noise and signal noise, also is subjected to the temperature of its external environment condition and the influence of brightness.Therefore only can find and the immediate picture of aforementioned picture block GTG value, certainly will have certain detecting error.
In the above-mentioned conventional art,, therefore the restriction of its translational speed is arranged for the image detecting device owing to be subjected to the size of image detecting device and the restriction of picture speed.Suppose that the frame numbers packet size that the image detecting device is extracted is M * M, the picture block is of a size of N * N, and the Pixel Dimensions of image detecting device is PN * PN, and picture speed is FN (frame/sec).The central authorities that are positioned at the picture array that the image detecting device extracted with the picture block are example, the image detecting device in the scope that horizontal direction and vertical direction allow to move be ± (M-N)/2, and the image detecting device is PN * (± (M-N)/2) * FN in the maximum translational speed that horizontal direction and vertical direction allowed.Therefore, the speed that moves when the image detecting device surpasses going up in limited time of above-mentioned hardware, promptly can't in the sampling picture, search the picture identical, the amount of movement that this will cause the misjudgment of block relative method and then correctly pick out the image detecting device with the GTG value of picture block with the block relative method.
Based on this, the Taiwan patent promptly proposes a kind of method for detecting of image detecting device amount of movement for No. 529309, in this patent, the image detecting device is in the specific part reference picture of primary importance extraction with reference to image, the image detecting device moves to the second place and extracts a sampling picture separately, the two is compared with reference picture and sampling picture, after waiting to obtain the picture identical with above-mentioned specific part reference picture, judge the moving direction of image detecting device this moment, and determine the position of another new reference picture that the image detecting device extracts in the 3rd position to use smooth judgement image detecting device amount of movement according to the moving direction of detector.This patent is by dynamically adjusting next reference block position with reference to image, in order to increase maximum displacement.But this method only can increase maximum displacement, can't strengthen the ability of displacement angle detecting.
Moreover, other has U.S. Patent number 5729008A to propose a kind of image amount of movement method for detecting, be to utilize quadric surface to mould relevant surfaces model (correlation surfacemodel) in this patent, and be that with the minimum value origination point it estimates the displacement integral part, according to the relation of origination point and surrounding associated numerical value (correlation value), utilize interpolation method to calculate the trickle displacement of fraction part then.Owing to must calculate the difference of each point when calculating connection numerical value, and need there be hardware to cooperate the radix point computing during displacement calculating amount, so operand can be quite big.
Given this, prior art is for being under the prerequisite not increasing hardware structure and reducing operand, and still there is the space of improvement in the amount of movement technology of detecting image shifter.
Summary of the invention
Therefore, purpose of the present invention is exactly in the method that provides a kind of detecting image to move, can be in the same reference image, according to the result of detecting image last time, dynamically extract different reference block, and then increase the ability of folk prescription, in addition to maximum amount of movement detecting with direction in contrast to the image motion-vector, still can increase the trickle mobile ability of detecting all directions, and then the variable quantity of the trickle move angle of detecting image.And when the detecting image amount of movement greater than can detect maximum amount of movement the time, according to the motional inertia principle, compensate the method for motion-vector.
A further object of the present invention provides the device that a kind of detecting image moves, do not increasing hardware resource and reducing under the operand dual requirements, strengthen maximum amount of movement of detecting and trickle mobile ability simultaneously, and then obtain to double the sensitivity of prior art detecting image move angle variable quantity.
In an embodiment of the present invention, the step that moves of detecting image is as described below.Extract first image at first respectively as the reference image, second image is as being the first sampling image, and be reference block with reference image center position, the block of then taking a sample in the sampling image goes out velocity of displacement and its motion-vector according to reference image and sampling area interblock position calculation.When the detecting image amount of movement greater than can detect maximum amount of movement the time, according to the motional inertia principle, the compensation motion-vector.Judge whether that then reference block and sampling block have reached image boundary all, if do not have, then extract a new reference block in the other direction, otherwise then upgrade, repeat above-mentioned steps and reached image boundary all up to reference block and sampling block with reference to image according to aforementioned displacements speed and its motion-vector.To replace this moment with the sampling image of sampling block reach image boundary the time formerly becomes with reference to image that another is new for image, and newly extracts a reference block separately with reference to image thus, repeats above-mentioned steps.
Method of the present invention can be before the sampling block reach image boundary, in same reference image (Reference frame), according to the result of detecting image last time, dynamically extract different reference block (Reference window) with direction in contrast to the image motion-vector, and then in order to increase the maximum amount of movement of detecting.In addition, because method of the present invention is only at reference block and sampling block when having reached image boundary all, just upgrade with reference to image, therefore allowing increases the trickle mobile ability that comprises level and vertical direction all directions, and then calculates the minute angle variable quantity that image moves by the amount of movement of all directions.As being example with image size 8 * 8 and reference block 2 * 2, the minimum move angle that technology of the present invention can be judged is slope 0.5/6 (rough 4.8 degree).
In addition, the present invention also will calculate the variation of acceleration, when acceleration is when stablizing reasonable value, predict next amount of movement simultaneously.When blurring speed greater than can detect amount of movement the time, because the detecting mistake causes the acceleration acute variation, can't reach rational state, the present invention's this moment will be with motion-vector the moving as this of previous prediction, after treating that acceleration is stable, be motion-vector just with the detecting result.
The device that the present invention provides a kind of detecting image to move again, can be an optical mouse apparatus in present embodiment, this device includes an optical sensor unit, an auxiliary computational unit, a microprocessing unit, an analog and digital signal converting unit and a filtering signals unit.Aforementioned optics sensing unit is coupled to the analog and digital signal converting unit, and the analog and digital signal converting unit is coupled to the filtering signals unit again.When the optical sensor unit reads image, export digital signal via the analog and digital signal converting unit, this digital signal is then delivered to microprocessing unit after the filtering signals unit removes denoising.Microprocessing unit is coupled to auxiliary computational unit, and microprocessing unit reads a reference picture by the optical sensor unit, and extracts a reference block in reference picture.Microprocessing unit reads a sampling picture in addition, and microprocessing unit, and by the calculating of auxiliary computational unit, is sought a sampling block of match reference block and calculated a motion-vector in this sampling picture according to reference block.Microprocessing unit calculates the variation of acceleration, when acceleration is when stablizing reasonable value, writes down the amount of movement of this moment simultaneously.When blurring speed greater than can detect amount of movement the time, because the detecting mistake causes the acceleration acute variation, can't reach rational state, this moment, microprocessing unit will be with motion-vector the moving as this of previous prediction, after treating that acceleration is stable, be motion-vector just with the detecting result.Then when reference block does not reach the border of reference picture, in reference picture, extract a update blocks separately, to replace reference block, when reference block has reached the border of reference picture and the sampling block also reach the sampling picture the border, then replacing former reference picture with the sampling picture becomes new reference picture, and extracts reference block separately in new reference picture.
In sum, when the user utilized the present invention to come detecting image to move, the present invention can strengthen the maximum amount of movement of detecting simultaneously not increasing hardware resource and reducing under the operand dual requirements.In addition, the user can also strengthen the trickle mobile ability of detecting by this, and then calculates the angle variable quantity that detecting image moves.Person very, when the detecting image amount of movement greater than can detect maximum amount of movement the time, according to the motional inertia principle, promptly compensate motion-vector.
For above and other objects of the present invention, feature and advantage can be become apparent, below in conjunction with accompanying drawing to a preferred embodiment of the present invention will be described in detail.
Description of drawings
Fig. 1 illustrates the process flow diagram of traditional detecting image moving method;
Fig. 2 illustrates a kind of flow chart of steps according to embodiments of the invention;
Fig. 3 A to Fig. 3 F illustrates the step synoptic diagram of the everything of the embodiment of the invention, and wherein left lateral is with reference to image, and right lateral is the sampling image;
Fig. 4 illustrates the functional block diagram of detecting image mobile device of the present invention;
Fig. 5 A illustrates the circular path figure that utilizes traditional detecting image mobile device to be painted;
Fig. 5 B illustrates the circular path figure that utilizes detecting image mobile device of the present invention to be painted.
Description of reference numerals:
S10, S20: extract reference picture
S12, S22: extract the sampling picture
S13: calculate the related coefficient between reference picture and sampling picture
S14: select minimum value and displacement calculating vector
S16: judge whether the sampling block reaches image boundary
S18: the picture of will taking a sample is substituted by new reference picture
S24: reference block has reached the border
S25: extract the reference block image according to translational speed
S26: displacement calculating
S261: whether displacement has reached predetermined motion-vector
S262: whether acceleration reaches a pre-determined number in allowed band
S263: predict next displacement
S264: export this displacement
S265: prediction of output displacement
S27: the sampling block has reached image boundary
S28: replace reference picture with the sampling picture
10: reference picture
12: the sampling picture
14: image
15: the image moving direction
16: reference block
17,19: block
18: the sampling block
20: the detecting image mobile device
22: the optical sensor unit
24: the analog and digital signal converting unit
26: the filtering signals unit
28: microprocessing unit
29: auxiliary computational unit
30: compare section
Embodiment
Fig. 2 illustrates the flow chart of steps of embodiments of the invention, and please in the lump with reference to Fig. 3 A to Fig. 3 F, Fig. 3 A to Fig. 3 F is an example with image size 8 * 8, and wherein left lateral is with reference to image, and right lateral is the sampling image, and supposes that the value of full stain is 0, and complete white value is 8.And the present invention can be implemented in a detecting image mobile device with extraction imaging function, for example optical mouse.Event moves up towards left side horizontal direction and with low-angle with mouse and is that example, move angle are about 7 degree (1/8) (this moment, image moved for turning right down).In an embodiment of the present invention, the step and the relevant action that move of detecting image is as described below.As described in step S20, extract first image at first respectively as reference image 10, and to extract with reference to image 10 centers be reference block 16, extract second image as sampling image 12 according to step S22 again.Please pay special attention among Fig. 3 A to Fig. 3 F, block 17 in reference image 10 two block of cells values above the block of cells wherein are 8, following block of cells value is 0, when image 14 moves down direction 15 change positions towards level and low-angle, because move angle is about 7 degree (1/8), 1/8 lattice so sampling image 12 every past lattice that move right, image will move down, and will differ 1 corresponding to the GTG value in the block 19 of block 17.Then in sampling picture 12, seek the size that is same as reference block 16 and with the most close person of GTG value of this reference block 16, extracting becomes sampling area piece 18.Please refer to Fig. 3 A and Fig. 3 B,, promptly go out velocity of displacement and its motion-vector hereat according to reference image and 18 position calculation of sampling block when image 14 moves down direction 15 change positions towards a level and low-angle.Please refer to step S24 among Fig. 3 C and Fig. 2, judge whether that then reference block 16 and sampling block 18 have reached image boundary all, when reference block 16 does not reach image boundary, promptly according to step S25 and S26, according to aforementioned displacements speed and its motion-vector, in same with reference to image on the direction of aforementioned motion-vector (in Fig. 3 B left), extract a update blocks and replace former reference block 16 and become new reference block 16.
Utilize one group of variable to write down each time displacement measurement and movable image is compensated computing.Please refer to step S261 among Fig. 2, when the image translational speed too fast and surpass a predetermined motion-vector, because the acceleration of can't displacement calculating and causing image to move is in unreasonable state for a long time, this displacement (step S265) is used as in prediction of output displacement of the present invention this moment, simultaneously judging whether acceleration drops on the interior count value of allowed band and reset or make zero, in order to correctly judging whether acceleration is in " stablize " state.When treating that but the image translational speed is in reconnaissance range, judge subsequently whether acceleration that image moves drops on and reach a pre-determined number (step S262) in the allowed band.For example: if the continuous acceleration of being detected more than three times is 0 or less than 1pixel/sec
2, judge promptly that hereat the acceleration that image moves is in " stablizing " state, then predict next displacement and this displacement numerical value is exported it (step S263 and step S264), be 0 or if there is not the continuous acceleration of being detected more than three times less than 1pixel/sec
2Then only export this displacement (step S264).In brief, judge whether to surpass maximum translational speed with the variation of acceleration, as reach overspeed condition, just according to the principle of motional inertia with last time predictive displacement numerical value output, otherwise promptly with the measured displacement numerical value output of this time.
Please refer to Fig. 3 D and Fig. 3 E, reference block 16 arrives image boundaries at this moment.Along with image 14 past moving to right, sampling block 18 is gradually near image boundary.If if reference block 16 and sampling block 18 have reached image boundary (illustrating as step S27 among Fig. 2 and Fig. 3 F) all, sampling image when will reach image boundary this moment with sampling block 18 replaces former reference picture becomes another new reference picture (as step S28 among Fig. 2), and new thus reference picture extracts a reference block 16 separately, repeats step S22.
Can detect low-angle characteristics in order to highlight the present invention, be example with absolute method of average MAD as the detecting moving method at this, and the computing example of Fig. 3 D to Fig. 3 F decision sampling block is described.On behalf of vertical direction, top two fritters in the block 19 and the MAD calculated value of block 17 do not have the MAD value of mobile (dy=0), and on behalf of vertical direction, the MAD calculated value of following two fritters in the block 19 and block 17 move down the MAD value of little lattice (dy=-1).Among Fig. 3 D, the MAD value of dy=0 is 4, and the MAD value of dy=-1 is 8, because the MAD value of dy=0 is less than the MAD value of dy=-1, so select vertical direction not have mobile.Among Fig. 3 E, the MAD value of dy=0 is 5, and the MAD value of dy=-1 is 6, because the MAD value of dy=0 still is less than the MAD value of dy=-1, so it is mobile still to select vertical direction not have.Among Fig. 3 F, the MAD value of dy=0 is 6, and the MAD value of dy=-1 is 4, because the MAD value of dy=0 is greater than the MAD value of dy=-1, so select vertical direction to move down little lattice, therefore, when proceeding to Fig. 3 F, system can detect and move down.Relatively, for existing method,, move down so can't detect because can upgrade automatically with reference to image (similar reset action).
Present embodiment only moves down direction 15 with image 14 towards a level and low-angle, and to change positions be example, illustrate that the present invention can go out velocity of displacement and its motion-vector according to reference image and 18 position calculation of sampling block, and judge whether that reference block 16 and sampling block 18 have reached image boundary all, former reference picture becomes another new reference picture and new thus reference picture extracts a reference block 16 separately and the image that determines whether to take a sample replaces.But present embodiment is not in order to limit the present invention, any those of ordinary skills, without departing from the spirit and scope of the present invention, will be appreciated that the present invention is at reference block and sampling block when having reached image boundary all, just upgrade with reference to image, therefore allow when image when vertical direction or other direction move, increase and detect the trickle mobile ability of all directions, and then calculate the minute angle variable quantity that image moves by the amount of movement of all directions.
Please be simultaneously with reference to Fig. 3 A to Fig. 3 F and Fig. 4, the present invention proposes the device 20 that a kind of detecting image moves in addition, and step and relevant action that its device 20 inner each unit connection relations, detecting image move are as described below.This device 20 includes an optical sensor unit 22, an auxiliary computational unit 29, a microprocessing unit 28, an analog and digital signal converting unit 24 and a filtering signals unit 26.Aforementioned optics sensing unit 22 is coupled to analog and digital signal converting unit 24, and analog and digital signal converting unit 24 is coupled to filtering signals unit 26 again.When optical sensor unit 22 reads image, export digital signals via analog and digital signal converting unit 24, this digital signal is then delivered to microprocessing unit 28 after filtering signals unit 26 removes denoising.
Detecting image mobile device 20 of the present invention, its detecting image moves step and relevant action is as described below.At first microprocessing unit 28 by control optical sensor unit 22 extract respectively first image as reference image 10, the second images as sampling image 12, and to extract with reference to image 10 centers be reference block 16.Then microprocessing unit 28 is by auxiliary computational unit 29 computings, in sampling picture 12, seek the size that is same as reference block 16 and with the most close person of GTG value of this reference block 16, extracting becomes sampling area piece 18.Please refer to Fig. 3 A and Fig. 3 B, change positions (in Fig. 3 A to the right), promptly go out velocity of displacement and its motion-vector hereat according to reference image and 18 position calculation of sampling block when image 14 moves horizontally direction and low-angle downward 15 towards one.Next described as previous embodiment, compensate computing by 28 pairs of movable images of microprocessing unit.Variation with acceleration judges whether to surpass maximum translational speed, as reaches overspeed condition, just export with predictive displacement numerical value according to the principle of motional inertia, otherwise promptly with the measured displacement numerical value output of this time.Please refer to Fig. 3 C, then microprocessing unit 28 judges whether that reference block 16 and sampling block 18 have reached image boundary all, when reference block 16 does not reach image boundary, promptly according to aforementioned displacements speed and its motion-vector, in same with reference to image on the direction of aforementioned motion-vector (in Fig. 3 B left), extract a update blocks and replace former reference block 16 and become new reference block 16.Please refer to Fig. 3 D and Fig. 3 E, reference block 16 arrives image boundaries at this moment.Along with image 14 past moving to right, sampling block 18 is gradually near image boundary.If if reference block 16 and sampling block 18 have reached image boundary (illustrating as Fig. 3 F) all, sampling image when microprocessing unit 28 will reach image boundary with sampling block 18 this moment replaces former reference picture becomes another new reference picture, and new thus reference picture extracts a reference block 16 separately.
Please be simultaneously with reference to Fig. 5 A and Fig. 5 B, Fig. 5 A and Fig. 5 B represent to utilize fixing existing detecting image mobile device 20 of machine and detecting image mobile device 20 of the present invention (for example optical mouse) respectively, when making this detecting image mobile device 20 carry out circular motion, in computer drawing software, write down its movement locus simultaneously.Everybody can extract same corresponding relatively section 30 at the two and do contrast, uses the method that detecting image of the present invention moves as can be known and installs 20, can obtain the motion track of more level and smooth littler complications.
Compared with prior art, prior art is upgraded with reference to image when sampling block 18 has reached image boundary because of it, promptly work as sampling image X (Y) direction and reach the border, but Y (X) direction does not detect when mobile as yet, can't detect minute angle and move, so the smoothness of its track detecting is desirable not to the utmost when carrying out circular motion.If image mobile detection device array is of a size of N, reference block 16 is of a size of M, then the maximum amount of movement of image mobile detection device be ± (N-M)/2.The minimum move angle that can detect is tan
-1(1/ (N-M)), as being example with image size 8 * 8 and reference block 162 * 2, the minimum move angle that can judge is tan
-1(1/ (N-M))=tan
-1(1/ (8-2)), that is slope 1/6 (rough 9.5 degree), thus conventional art can't to detect less than slope be moving of 9.5 degree, so be 0,90,180 and 270 degree at 0 ± 9.5,90 ± 9.5,180 ± 9.5 and 270 ± 9.5 move angles that detected.
Yet if use method of the present invention, then can be before sampling block 18 reach image boundary, in same reference image (Reference frame), according to the result of detecting image last time, dynamically extract different reference block 16 (Referencewindow) with direction in contrast to the image motion-vector, and then in order to increase the maximum amount of movement of detecting horizontal direction.In addition, because method of the present invention is only at reference block 16 and sampling block 18 when having reached image boundary all, just upgrade with reference to image, therefore allowing increases the trickle mobile ability of detecting vertical direction, and then calculates the minute angle variable quantity that image moves by level with the amount of movement of vertical two directions.With above-mentioned, as being example with image size 8 * 8 and reference block 16 2 * 2, the minimum move angle that technology of the present invention can be judged is tan
-1(0.5/ (N-M)), that is slope 0.5/6 (rough 4.8 degree), the sensitivity that can increase the angle variable quantity that detecting image moves compared to prior art the present invention reaches more than the twice.
In sum, the foregoing description is not in order to limit the present invention, any those of ordinary skills, without departing from the spirit and scope of the present invention, when knowing that the present invention, just upgrades with reference to image when having reached image boundary all at reference block and sampling block, therefore permission when image when vertical direction or other direction move, increase the trickle mobile ability of detecting all directions, and then calculate the minute angle variable quantity that image moves by the amount of movement of all directions.
In sum, the present invention has the following advantages at least:
1. utilize the present invention not increase under hardware resource and the minimizing operand dual requirements, strengthen the maximum amount of movement of detecting all directions simultaneously.
2. in addition, the user also can strengthen the trickle mobile ability of detecting by this as utilizing the present invention, and then obtains to double the sensitivity of prior art detecting image move angle variable quantity.
3. but when translational speed during greater than reconnaissance range, the present invention will have the image amount of movement of unreasonable acceleration according to principle of inertia, compensate it with the motion-vector of prediction, and then obtain preferred stability.
Though the mode by preferred embodiment has been openly as above in the present invention; right its is not in order to limit the present invention; any those of ordinary skills; without departing from the spirit and scope of the present invention; when can doing a little change and retouching, so protection scope of the present invention should be with being as the criterion that claim was limited.
Claims (20)
1. the method that detecting image moves comprises the following steps:
(A) read a reference picture, and in this reference picture, extract a reference block;
(B) read a sampling picture;
(C) when this reference block does not reach the border of this reference picture, in this reference picture, extract a update blocks separately, to replace this reference block;
(D) according to this reference block in this sampling picture, seek a sampling block of this reference block of coupling and calculate a motion-vector;
(E) when this sampling block has reached the border of this sampling picture, then replace this reference picture, and in this reference picture, extract this reference block separately with this sampling picture; And
(F) carry out step (B) again.
2. the method that detecting image as claimed in claim 1 moves, wherein step (C) is according to this motion-vector or a translational speed corresponding with this motion-vector, extracts this update blocks in this reference picture separately.
3. the method that detecting image as claimed in claim 1 moves, wherein this update blocks of step (C) is by in contrast to this sampling block moving direction and the extractor.
4. the method that detecting image as claimed in claim 1 moves, the step of wherein seeking this sampling block in the step (D) be in this sampling picture searching be same as the big or small of this reference block and with the most close person of GTG value of this reference block.
5. the method that detecting image as claimed in claim 1 moves wherein comprises that also one compensates the method that image moves between step (D) and step (E), and the method that this compensation image moves comprises the following steps:
(a), judge whether to surpass a predetermined motion-vector according to the image acceleration;
(b) should be scheduled to motion-vector, then prediction of output motion-vector when surpassing; And
(c) above being scheduled to motion-vector, judge then whether this acceleration is to reach a pre-determined number in an allowed band, if this acceleration reaches a pre-determined number and then predicts next motion-vector and export this displacement in this allowed band, otherwise only exports this displacement.
6. the method that detecting image as claimed in claim 5 moves, wherein reaching a pre-determined number in this allowed band of step (c) is that continuous at least three measured acceleration are all less than 1pixel/sec
2
7. device that detecting image moves comprises:
One optical sensor unit is in order to read image;
One auxiliary computational unit is in order to calculate the image of seeking coupling;
One microprocessing unit, be coupled to this auxiliary computational unit, this microprocessing unit reads a reference picture by this optical sensor unit, and in this reference picture, extract a reference block, this microprocessing unit reads a sampling picture in addition, this microprocessing unit according to this reference block in this sampling picture in, seek a sampling block of this reference block of coupling and calculate a motion-vector, when this reference block does not reach the border of this reference picture, in this reference picture, extract a update blocks separately, to replace this reference block, when this reference block has reached the border of this reference picture and this sampling block when also reaching the border of this sampling picture, then replace this reference picture, and in this reference picture, extract this reference block separately with this sampling picture.
8. the device that detecting image as claimed in claim 7 moves, wherein this device has an at least one analog and digital signal converting unit and a filtering signals unit, wherein this optical sensor unit is coupled to this analog and digital signal converting unit, and this analog and digital signal converting unit is coupled to this filtering signals unit.
9. the device that detecting image as claimed in claim 7 moves, wherein this update blocks system is according to this motion-vector or a translational speed corresponding with this motion-vector, comes in this reference picture extractor separately.
10. the device that detecting image as claimed in claim 7 moves, wherein this update blocks is by in contrast to this sampling block moving direction and the extractor.
11. the device that detecting image as claimed in claim 7 moves, wherein this sampling block be in this sampling picture, seek the size be same as this reference block and with the most close person of GTG value of this reference block.
12. the device that detecting image as claimed in claim 7 moves, wherein this microprocessing unit comprises that also compensating image moves computing after calculating this motion-vector, and this compensation image moves computing and comprises following action:
(a), judge whether to surpass a predetermined motion-vector according to the image acceleration;
(b) should be scheduled to motion-vector, then prediction of output motion-vector when surpassing; And
(c) above being scheduled to motion-vector, judge then whether this acceleration reaches a pre-determined number in an allowed band, if this acceleration reaches a pre-determined number and then predicts next motion-vector and export this displacement in this allowed band, otherwise only exports this displacement.
13. the device that detecting image as claimed in claim 7 moves, wherein this image detecting device can be an optical mouse.
14. the device that detecting image as claimed in claim 13 moves, wherein reaching a pre-determined number in this allowed band is that continuous at least three measured acceleration are all less than 1pixel/sec
2
15. one kind compensates the method that image moves, comprises the following steps:
(A) read a reference picture, and in this reference picture, extract a reference block;
(B) read a sampling picture;
(C) according to this reference block in this sampling picture, seek a sampling block of this reference block of coupling and calculate a motion-vector;
(D), judge whether to surpass a predetermined motion-vector according to the image acceleration;
(E) should be scheduled to motion-vector, then prediction of output motion-vector when surpassing;
(F) above being scheduled to motion-vector, judge then whether this acceleration reaches a pre-determined number in an allowed band, if this acceleration reaches a pre-determined number and then predicts next motion-vector and export this displacement in this allowed band, otherwise only exports this displacement; And
(G) carry out step (B) again.
16. the method that compensation image as claimed in claim 15 moves, wherein reaching a pre-determined number in this allowed band of step (F) is that continuous at least three measured acceleration are all less than 1pixel/sec
2
17. the method that compensation image as claimed in claim 15 moves, wherein step (E) is also judging whether acceleration drops on the interior count value of allowed band and reset.
18. one kind compensates the device that image moves, comprising:
One optical sensor unit is in order to read image;
One auxiliary computational unit is in order to calculate the image of seeking coupling;
One microprocessing unit, be coupled to this auxiliary computational unit, this microprocessing unit reads a reference picture by this optical sensor unit, and in this reference picture, extract a reference block, this microprocessing unit reads a sampling picture in addition, this microprocessing unit according to this reference block in this sampling picture in, seek a sampling block of this reference block of coupling and calculate a motion-vector, and compensate the computing that image moves, this microprocessing unit is according to the image acceleration, judge whether to reach a predetermined motion-vector, should be scheduled to motion-vector when surpassing, prediction of output motion-vector then, should be scheduled to motion-vector if surpass, then continue and judge whether this acceleration reaches a pre-determined number in an allowed band,, replace this reference picture and in this reference picture, extract this reference block separately with this sampling picture subsequently if this acceleration reaches a pre-determined number and then predicts next motion-vector and export this displacement in this allowed band.
19. the device that compensation image as claimed in claim 18 moves, wherein this microprocessor is in that surpass should be predetermined during motion-vector, also judging whether acceleration drops on the interior count value of allowed band and reset.
20. the device that compensation image as claimed in claim 19 moves, wherein this microprocessing unit is all less than 1pixel/sec according to continuous at least this accekeration that records for three times
2And judge that this acceleration is whether in this allowed band.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004101031766A CN100395690C (en) | 2004-12-31 | 2004-12-31 | Method and apparatus for detecting image shift |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004101031766A CN100395690C (en) | 2004-12-31 | 2004-12-31 | Method and apparatus for detecting image shift |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1801060A true CN1801060A (en) | 2006-07-12 |
| CN100395690C CN100395690C (en) | 2008-06-18 |
Family
ID=36811099
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004101031766A Expired - Fee Related CN100395690C (en) | 2004-12-31 | 2004-12-31 | Method and apparatus for detecting image shift |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100395690C (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102117136B (en) * | 2010-01-04 | 2013-08-14 | 原相科技股份有限公司 | Displacement detecting device and method |
| US8515129B2 (en) | 2009-12-18 | 2013-08-20 | Pixart Imaging Inc. | Displacement detection apparatus and method |
| CN103576845A (en) * | 2012-08-06 | 2014-02-12 | 原相科技股份有限公司 | Environment light detection device and method and interactive device adopting environment light detection device |
| CN111080985A (en) * | 2018-10-19 | 2020-04-28 | Tcl新技术（惠州）有限公司 | Wearable device based anti-fatigue alarm method, wearable device and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08335142A (en) * | 1995-06-08 | 1996-12-17 | Fujitsu Ltd | Move controller for mouse pointer |
| TW472206B (en) * | 1998-03-30 | 2002-01-11 | Agilent Technologies Inc | Seeing eye mouse for a computer system |
| TW529309B (en) * | 2001-10-31 | 2003-04-21 | Pixart Imaging Inc | Displacement measuring method of image sensor |
| CN1180378C (en) * | 2001-12-29 | 2004-12-15 | 原相科技股份有限公司 | Moving distance detecting method of image sensor |
-
2004
- 2004-12-31 CN CNB2004101031766A patent/CN100395690C/en not_active Expired - Fee Related
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8515129B2 (en) | 2009-12-18 | 2013-08-20 | Pixart Imaging Inc. | Displacement detection apparatus and method |
| CN102117136B (en) * | 2010-01-04 | 2013-08-14 | 原相科技股份有限公司 | Displacement detecting device and method |
| CN103576845A (en) * | 2012-08-06 | 2014-02-12 | 原相科技股份有限公司 | Environment light detection device and method and interactive device adopting environment light detection device |
| CN111080985A (en) * | 2018-10-19 | 2020-04-28 | Tcl新技术（惠州）有限公司 | Wearable device based anti-fatigue alarm method, wearable device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100395690C (en) | 2008-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1278277C (en) | Finger mobile detection method and appts. thereof | |
| US7796120B2 (en) | Method and apparatus for detecting image movement | |
| CN1825260A (en) | Programmable lift response for an optical navigation device | |
| CN1503198A (en) | Method for processing light mouse data | |
| CN1744673A (en) | Video electronic flutter-proof device | |
| CN1304928C (en) | Method for counting photoelectric mouse moving value and photoelectric mouse using the same method | |
| CN1635541A (en) | Photoelectric detection positioning system and method for computer touch screen | |
| CN1956547A (en) | Motion vector estimating device and motion vector estimating method | |
| CN1916826A (en) | System and method for optical navigation device | |
| CN1758132A (en) | Imaging system and imaging method | |
| CN101068309A (en) | Method and device for automatic rotating mobile terminal camera head shooting results | |
| CN1885946A (en) | Motion error compensator, and method for detecting and compensating motion error | |
| MX2007002073A (en) | Image pickup device, image pickup result processing method and integrated circuit. | |
| CN101063911A (en) | Method and apparatus for detecting motion of image in optical navigator | |
| CN1674040A (en) | Apparatus for controlling the position of a screen pointer with low sensitivity to particle contamination and method thereof | |
| CN1879078A (en) | Information processing device and method, and program | |
| CN1272702C (en) | Navigation system and navigation method | |
| CN1235127C (en) | Optical mouse and method for avoiding error action | |
| CN1801060A (en) | Method and apparatus for detecting image shift | |
| CN103809348B (en) | It is applied to the automatic focusing method of camera model | |
| CN1627825A (en) | Motion estimation method for motion picture encoding | |
| CN110332886A (en) | A kind of precision visual method for rapidly positioning | |
| CN1852442A (en) | Layering motion estimation method and super farge scale integrated circuit | |
| CN103809817B (en) | Optical touch system and object position judgment method thereof | |
| CN1157957C (en) | Device and method for searching motion vector |
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 |
Granted publication date: 20080618 Termination date: 20161231 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |