CN101159813B

CN101159813B - Method and device for image anti-shake in digital cameras

Info

Publication number: CN101159813B
Application number: CN2007100062031A
Authority: CN
Inventors: 单继章; 杰西·扬华李; 刘关嵩; 潘辉; 丹尼尔·L·弗拉姆
Original assignee: HAOWEI TECH Co Ltd
Current assignee: Omnivision Technologies Inc
Priority date: 2006-03-31
Filing date: 2007-01-30
Publication date: 2012-04-25
Anticipated expiration: 2027-01-30
Also published as: CN101159813A

Abstract

A method and system for minimizing the effects of image motion in still cameras is provided. In one embodiment, a method of capturing an image in a digital camera is presented. The method includes calculating a sharpness value related to an image input. The method also includes evaluating the sharpness value to determine image motion. The method further includes capturing a next image input data from the image input responsive to evaluating the sharpness value.

Description

The method of the image anti-shake in the digital camera and device

Related application

The present invention relates to the provisional application No.60/763 that the applicant proposed in the U.S. on January 30th, 2006,516, require this provisional application as priority of the present invention here.

Technical field

The present invention relates to a kind of method and apparatus of in digital camera, catching image; More particularly, the present invention relates to a kind of method of metastable picture, a kind of equipment and a kind of digital camera that is used to catch metastable picture that is used for catching under opportunity image of in digital camera, catching at harmonic motion.

Background technology

Image tremble (image shake) be a kind of problem that reduces the digital camera performance.Image trembles and is caused by the mobile of camera user usually, and the vibration that is perhaps come by bracing frame (such as tripod or carriage) transmission causes.The root that another kind of image trembles moves from the object to be imaged.Because transducer becomes littler, and the quantity of the pixel in the imageing sensor is increasing, and image trembles and become more serious problem.

A kind of method of avoiding image to tremble is to build a kind of rotary table (gyroscopic mount) for sensor array.Like this, though around camera during component movement sensor array still keep static.But this method is comparatively expensive; And, can not reduce the adverse effect that moves of the object to be imaged to the compensation of camera motion.

Although camera maybe be always mobile a little, trembling of image shown that the motion of camera changes in the process of taking pictures.Similarly, the motion of object can change, and for example perhaps carries out a unexpected of short duration motion when the football player jumps.Therefore, be necessary to provide a kind of and can utilize harmonic motion (low motion) method and system on opportunity well.In addition, be necessary the influence that image trembles to be dropped to minimum with lower cost.

Summary of the invention

On the one hand, the invention provides a kind of method of in digital camera, catching image, it comprises:

Receive the input of previous picture frame, and form the sharpness value of previous picture frame, wherein, previous picture frame input is a sub-frame of previous picture frame, and previous picture frame is a picture frame before next-door neighbour's current image frame;

Receive the input of current image frame, and form the sharpness value of current image frame, wherein, the current image frame input is a corresponding subframe of current image frame;

Poor according between the sharpness value of the sharpness value of the subframe of the sharpness value of the sharpness value of previous picture frame, current image frame and current image frame and the corresponding subframe of previous picture frame; The motion of estimating next picture frame will be less than the motion of current image frame; Wherein, The subframe of current image frame is last subframe of current image frame, and the absolute value of difference is less than the maximum and the product of selecting the factor of the absolute value of other continuous acutance value difference;

Based on estimating, next picture frame is caught in decision; And catch next picture frame.

Above-mentioned method may further include: receive capture command, wherein, the reception that the previous image of previous picture frame is imported is after receiving this capture command.

Above-mentioned method also can further comprise: receive first image input of first picture frame, and form the sharpness value of this first image input; Wherein, first picture frame is different from previous picture frame, and the reception of first image input is before the reception of image input formerly.Wherein, first image input can be a sub-frame of first picture frame, and previous image input can be a sub-frame of previous picture frame, and the present image input can be a sub-frame of current image frame; Form sharpness value and can comprise that a high frequency extracts filter, and this high frequency extraction filter can be a high pass filter.

Above-mentioned method also can further comprise: receive capture command, wherein, the reception of first image input is before the reception of capture command, and the reception of previous image input is after the reception of capture command.

In the above-mentioned method; From previous image input and present image input, select the image input and form its sharpness value and can comprise: with the image input high pass filter matrix (high pass filter matrix) is carried out the convolution summation, and this high pass filter matrix can have symmetry.

In the above-mentioned method, the frame acutance of the frame acutance of current image frame greater than previous picture frame depended in the decision of catching next picture frame; The frame acutance of the frame acutance of current image frame greater than the arbitrary image frame after the capture command depended in the decision of catching next picture frame; The present image input is a sub-frame of current image frame; The input of previous image be a corresponding subframe of previous picture frame, and the sharpness value of the sharpness value of estimating the subframe that then further depends on said current image frame and the corresponding subframe of said previous picture frame between poor; This previous picture frame can be a picture frame before the said current image frame of next-door neighbour; The subframe of current image frame can be last subframe of current image frame, and the absolute value of difference is less than the maximum and the product of selecting the factor of the absolute value of other continuous acutance value difference; This selection factor depends on poor symbol, and less than 1.

On the other hand; The invention provides a kind of means of in digital camera, catching metastable picture; It comprises that the image motion of estimating next picture frame when will be less than the step of the image motion of current image frame, and according to estimating the means of catching next picture frame; Wherein, estimate and catch by following process completion:

Estimate based on said, next picture frame is caught in decision; And catch next picture frame.

On the one hand, the invention provides a kind of method of in digital camera, catching metastable picture again, it comprises: calculate first sharpness value about an image input; Calculating is about second sharpness value of another image input; Calculating comprises the acutance value difference of first sharpness value and second sharpness value; Estimate the quality of next image according to the acutance value difference; And, from the image input, catch next image according to estimation to next picture quality.In this method, first sharpness value can comprise the sharpness value of a sub-frame, and second sharpness value can comprise the sharpness value of another subframe.

Another aspect the invention provides a kind of equipment that is used for catching under opportunity at harmonic motion image, and it comprises: the means of calculating at least two acutance value differences of image input; According to the acutance value difference, estimate the means of next picture quality; And the means of response estimated image quality, from the image input, catch the means of next image.

On the one hand, the invention provides a kind of digital camera that is used to catch metastable picture again, it comprises: the digital image sensor of output and image related data; The machine readable medium of storage data and instruction; The processor of at least one execution command and operating data; The instruction that is used to carry out; Be attached to the acutance detector of imageing sensor, this acutance detector is used for the acutance of evaluate image; According to acutance, estimate the prediction device of the image motion of next image; And decision is caught next image and next image is captured the determining device in the machine readable medium.

In the above-mentioned digital camera, machine readable medium can comprise the line buffer that can be used for the relatively small amount line; The acutance detector can be asked convolution to the high pass filter matrix with the data relevant with image.

On the one hand, the invention provides a kind of method of in digital camera, catching image again, it comprises: will be divided at least four sub-frame from a picture frame of digital image sensor, these subframes have the view data of equivalent; Receive capture command; Receive current sub-frame images data from digital image sensor; In line buffer, store the line of the selection quantity of current sub-frame images data; The execution high frequency filters, and it comprises the selection row with current sub-frame images data in the line buffer, and the filter matrix is asked convolution; Filter according to high frequency, assess the acutance of current subframe; Assess the poor of current and previous sharpness value; Confirm the poor of maximum current and previous sharpness value; Assessment comprises the acutance of the picture frame of current sub-frame images data; Poor according to the difference of current and previous sharpness value and maximum current and previous sharpness value estimated the quality of next picture frame; And according to the acutance of estimating and comprising the picture frame of current sub-frame images data, next picture frame is caught in decision.

Method and system of the present invention can utilize harmonic motion opportunity well, with lower cost the influence that image trembles is dropped to minimum.

Present invention is described to pass through accompanying drawing with the mode of example below.Accompanying drawing should be counted as example and be unrestricted, because scope of the present invention is defined by claim.

Description of drawings

Fig. 1 has showed an execution mode that is used for reducing at digital camera a kind of system of image motion influence.

Fig. 2 has showed the subframe that the continuous images frame is divided into the continuous images frame.

Fig. 3 has showed the high-pass filter coefficient of 5x5 matrix.

Fig. 4 A has showed a kind of flow chart that improves the selection step of the execution mode that image catches.

Fig. 4 B has showed the more detailed flow chart of the selection step of the execution mode among Fig. 4 A.

Fig. 5 has showed that the pixel data in the line buffer of an execution mode moves.

Fig. 6 has showed a kind of flow chart that improves the data flow of the execution mode that image catches.

Fig. 7 has showed the curve chart that the image of the improvement of an execution mode is caught.

Fig. 8 has showed an execution mode of digital camera.

Embodiment

The invention provides a kind of system, method and equipment that is used at digital still life camera image anti-shake.The execution mode of describing among this paper is an example of the present invention, and its essence just is used for explanation rather than restriction.

In one embodiment, disclosed a kind of method of in digital camera, catching image.This method comprises that sharpness value is calculated in input based on image.In this embodiment, calculate sharpness value and comprise the high frequency value of confirming about the image input.This method also comprises the quality of estimating next image based on sharpness value.This method further comprises response, and this is estimated (prediction) decision and whether catches next image input data.

In another embodiment, disclosed a kind of digital camera.This camera comprises a processor.This camera also comprises the media that be used for image storage of connection (coupled) to processor.This camera further comprises the imageing sensor that is attached to processor.This camera comprises that also the picture quality detector is used for detection image motion (image motion).This picture quality detector comprises one based on to carry out the acutance detector of high-pass filtering from the view data of digital image sensor.And this camera comprises a prediction device (predictor) that is attached to next image motion of mass detector.In addition, this camera comprises a determining device (decision maker) that is attached to prediction device.This prediction device is used for assessing the output of (evaluate) mass detector and catches the image from digital image sensor, the output of response quality detector at media with this determining device.In some embodiments, this mass detector, prediction device and determining device can be processed the device execution.

In another embodiment, this aspect has disclosed a kind of equipment.This equipment comprises the means (means) that are used to calculate the sharpness value of exporting about present image.This equipment also comprises the means that are used to rely on this next picture quality of sharpness value estimation.This equipment further comprises the means that are used for from this next image input data frame of image input capture.The means response that is used to catch is used to estimate the means of next picture quality and works.

Here a kind of method and apparatus of describing and disclosing can improve picture quality in digital still life camera.This method and apparatus relies on the motion feature of estimated image input data, and catches next image of being sensed subsequently.Like this, confirmed to reach that standard of relatively stable picture when processor, processor can be caught next image subsequently and expected that next image will be relatively stable.This allows to assess apace, and need not store a plurality of images.This can also reduce the cost problem that expensive components causes, such as, movable imageing sensor and lens.

In the following description, in order to explain, various details have been proposed so that provide about thorough of the present invention.Yet clearly, for a person skilled in the art, the present invention can not realize yet when having these details.In addition, construction and device shows with the present invention that avoids confusion with the block diagram form.

" execution mode " or " execution mode " mentioned in this specification, described special characteristic, structure or property bag that expression is associated with this execution mode are contained at least one execution mode of the present invention.Therefore, the phrase " in one embodiment " that occurs of many places is inevitable in this manual all represent same execution mode, certainly neither be uncorrelated or other execution mode of mutual repulsion.

Can further understand various execution modes through accompanying drawing.Fig. 1 has showed a kind of system implementation mode that is used for reducing at digital camera the influence of image motion.System 100 comprises the optics 105 that projects image onto imageing sensor 110.Imageing sensor can be CCD, CMOS or similar image sensor matrix.This array is the rectangular array of embarking on journey into the detector of row layout in many execution modes.The pixel of imageing sensor 110 is made public according to rotating shuttle timing mechanism well-known in the art (rolling shutter timing mechanism).In the execution mode of being showed, first row of pixel is selected and is sequentially outputed to line buffer (line buffer) 150 from imageing sensor from the view data that is selected each pixel in the row.Next, next line is copied to line buffer from image output in the same way.In each ending of scanning fully of entire image sensor array, the blanking interval of the scheduled time appears at before raster scan is repeated.One group of view data that scans fully from each pixel in the imageing sensor 110 is called as a frame (frame).The visual picture transducer is used for response by projected image and output image data.

For a person skilled in the art, at random with " OK " perhaps " row " to come each transducer that is positioned on the line (line) in the presentation video transducer be very conspicuous.Although term " OK " typically refers to a pixel that is parallel to digital camera top and bottom, the present invention does not rely on any concrete appointment and in scope of the present invention and spirit, can advance the determinant raster scan yet.

And partitioned image transducer and output are useful from various other methods of the data of imageing sensor in some embodiments.For example, being selected the artificial frame that submatrix forms and repeatedly scanning according to selecteed order by the image sensor pixel data.And a kind of transducer of pixel addressable frame buffer that comprises can be used to accomplish some scheme of the present invention.

As everyone knows, receive the tremble adverse effect of (being also referred to as image motion here) of image when image quality of digital image when making public.Camera motion causes that image trembles and causes acutance to reduce unintentionally.Yet image motion can also be derived from the motion of the target or the object to be imaged.The characteristic of motion is depended in the relative variation of continuous images frame on acutance.Usually, the acutance of image is bigger when having less motion.Discover that image motion can be estimated according to relative variation of the acutance of subframe (subframes) effectively.The present invention includes the anti-method of trembling of a kind of camera efficiently, it comprises according to subframe estimates that relative acutance changes, then according to changing the acutance of estimating next image relatively.In another kind of scheme of the present invention, the anti-method of trembling of camera is used to reduce the adverse effect of target travel on picture quality.In some other scheme of the present invention, disclosed a kind of anti-digital camera that quivers that adopts the anti-method of trembling.This digital camera can utilize common hardware resource to accomplish, so cost is lower.

With reference to Fig. 1 an execution mode that is used for improving at digital camera a kind of method and system of picture quality of the present invention has been carried out further elaboration.This system 100 comprises that response is from the image of optical path 105 and form the imageing sensor 110 of view data.Although be the image of visible light in this execution mode, other form that other execution mode comprises the electromagnetic radiation that forms images such as infrared ray, x-ray, millimeter wave, or the like.This system 100 comprises that further the machine readable medium that is used for store instruction and data and at least one are used to execute instruction and the processor of deal with data.This system also comprises picture quality detector 120, and it is according to the view data input sensed image motion from imageing sensor 110.

In this embodiment, be moved to the buffer 150 of machine readable medium from the data input of imageing sensor 110.This buffer is coupled to picture quality detector 120.In many execution modes, mass detector 120 comprises the acutance detector.This system also comprises motion prediction device 130 and determining device 140.Instruction and data in the computer-readable media is processed device and is used to carry out acutance detector 120, prediction device 130 and determining device 140.Yet in various other execution modes, the function of detector, prediction device or determining device fully or partly, can adopt control circuit to carry out.In this embodiment, media comprises the line buffer that is used to store with from many lines of the capable corresponding view data of the pixel data input of imageing sensor 110.Although the line of view data is meant the row of data in this embodiment, line can be the row of pixel data equally in other embodiments.And, comprise the detector of non-rectangle pixel structure and/or be used for receiving from the image input of image sensing and with this storage and can be used to realize the present invention in other method of machine readable medium.

In an execution mode of system shown in Figure 1, imageing sensor 110 is XGA resolution transducers, and it has 768 row, and every row comprises 1024 pixels.Therefore each complete picture frame of this transducer comprises 768 pixels of 1024 x.Yet comprise that the picture frame from the subclass of the pixel data of transducer is selected processing usually.For example, the less pel array that comprises interesting target can be selected, and has avoided handling through and through and storing the undesirable background (this is called " digital zoom " in some embodiments) from the image input like this.

Found that picture frame is divided into subframe generally can improve the sensitivity of surveying motion.In execution mode shown in Figure 2, picture frame is divided into four lines (row-wise) subframe.Each subframe is W pixel wide and H pixels tall.Wherein picture frame is the XGA array, and the W value is the number or the columns of the horizontal pixel of every row, is 1024.H is by the number of the pixel column of subframe number divided image frame.Therefore the H value is 256 in this embodiment.Certainly in other embodiments, for example, when picture frame is selected to the pixel subset that has from the XGA transducer, this picture frame will have other resolution and can be divided into different number of sub-frames according to concrete application.Therefore, W and H depend on selected resolution and selected number of sub-frames.

In many execution modes, the two field picture acquisition speed is 15 or 30 frame per seconds, depends on many design constraints and industrial standard.The corresponding picture frame time for exposure will be slightly less than 1/15 second or 1/30 second, because frame per second comprises blanking interval usually between double exposure.

In many execution modes, the continuous numbered line quilt exposure in order of picture frame.Because frame per second (frame rate) reduces, picture frame must be divided into more subframes so that the relative motion of this image of sampling effectively.On the other hand, if the number of subframe is too many, the height that reduces of each subframe can produce a kind of artificial sensitivity to the motion of vertical direction.Therefore can find out and between image acquisition speed and image subframes, exist a balance point.Discover when adopting the frame per second of VGA (640x480) or XGA picture frame resolution and standard 15/ second or 30/ second, use at least four same horizontal subframes can realize effective motional induction.

In an execution mode shown in Figure 1, acutance detector 110 becomes to assign to survey acutance according to the spatial high-frequency of view data.When the image of object has the acutance border (for example, object is not trembled by camera or the target of moment move darken), they often have more high spectrum.Image during same object of which movement often has less high spectrum (for example, motion is from blured defining of border in essence).Usually, the spatial high-frequency of increase spectrum is relevant with acutance in image.Therefore a high frequency extracts the acutance that filter is used to detection image.Also can be used to the sensed image motion yet the detection acutance perhaps detects other method of acutance disappearance on the contrary, this depends on concrete application.

In one embodiment, the acutance detector comprises high-pass filter.The high pass filter that is fit to comprises the 5x 5 high pass filter matrix convolution examples among Fig. 3, wherein view data f (x, y) be extracted the picture data of publishing picture high frequency value g (x, y).The application of this filter is through assessment convolution and (convolution sum):

g (x, y) = Σ_{m = - 2}^{2} Σ_{n = - 2}^{2} w (m, n) f (x + m, y + n),

Wherein, (m n) is an element of high pass filter matrix shown in Figure 3 to w.The center of matrix (m=0, n=0) be positioned in (x, y).Although the high pass filter matrix comprises 25 values, it can only be stored in 6 memory cell expeditiously, its belong to 4-folding symmetry (owing to the 4-fold symmetry): w (m, n)=w (m, n)=w (m ,-n)=w (m ,-n).High pass filter matrix shown in Figure 3 just can be used to extract the example of filter of the HFS of view data.Matrix with symmetry is effective especially.Other high frequency extracts filter and crosses device filter combination such as single band filter or multiband, and like, also can use in some embodiments, and this depends on concrete application.

Acutance detector execution mode is j ^ThI in the frame ^ThSubframe forms subframe sharpness value, j ^ThFrame comprises the average HFS of subframe that is provided by following relation of plane:

s_{ij} = \frac{Σ_{x = 1}^{W} Σ_{y = 1}^{H} | g_{ij} (x, y) |}{W \times H},

Wherein, W is the number along the pixel of subframe Width (being meant the delegation in the frame in this embodiment), and H is the number of the pixel that (is meant the subframe part of entire frame row height) in subframe one row.

Prediction device 130 is estimated the expection image motion of next frame according to the previous subframe sharpness value of the current subframe sharpness value of present frame and previous picture frame.A kind of prediction device method of showing in one embodiment is that picture frame is divided into four sub-frame of running in such a way.Current j ^ThI in the frame ^ThSubframe sharpness value and corresponding previous (j-l) ^ThDifference between the subframe sharpness value of frame is calculated according to following formula:

D _ij＝s _ij-s _i，j-1

Wherein, i is current number of sub frames (i=1,2,3,4).It is to be noted that the subframe of different frame is called as corresponding subframe, and if only be if they have identical number of sub frames.And, D _Max, the acutance value difference D of before continuous subframe _IjMaximum value after capture command, obtain according to following formula:

D _max＝max(|D _ij|)，

Wherein, subscript i comprises sub-frame number (i=1,2,3,4) and subscript j is included in the image frames numbers from first picture frame to current image frame after the capture command.

Prediction device 130 is estimated the image motion of next picture frame through following two propositions of assessment (proposition):

Proposition 1:D _4j＞0 and D _4j＜k ₁* D _MaxAnd D _4j＞D _3j

Proposition 2:D _4j＜0 and|D _4j|＜k ₂* D _MaxAnd D _4j＞D _3j

Wherein, k ₁And k ₂It is the factor of selecting.As if at least one establishment in two propositions, then the next frame image motion will will be outputed to judge module 140 less than estimating of current image frame motion.Otherwise the next frame image motion will be equal to or greater than estimating and will being outputed to determining device of current image frame motion.In application, find, work as k ₁Select constant 2/3 and k ₂When selecting constant 1/2, very effectively.k ₁And k ₂Other selection also can be used, although empirical tests work as k ₁And k ₂Selective value better less than 1 o'clock.

The sub-frame images data of the most recent of a frame are considered to provide about more Duoing of motion in the future is accurate than old data to be estimated.Therefore proposition 1 is based on D with proposition 2 in execution mode _4jThis is because the data of subframe 4 are last and are the parts from the most recent of the image frame data input of imageing sensor.Yet other estimate that according to the acutance of image frame data the method for motion can be used in the various execution modes, and other future image motion prediction device can in scope of the present invention and spirit, be used.

In an execution mode of method and apparatus shown in Figure 1, judge module 140 calculates the frame acutance S of present frame j _jThe frame acutance S of present frame (j) _jAdopt following formula to calculate:

S_{j} = Σ_{i = 1}^{4} s_{ij} .

Yet other tolerance (metric) based on the frame acutance of other formula also can be used, and this depends on concrete application.

Whether determining device 140 decisions are caught next frame and it are kept at video memory (memory) 115 from imageing sensor 110.When following three propositions are all set up, decision is caught: 1) receive capture command, 2) prediction device estimates the acutance that next frame will have increase, and 3) S _j＞S _c(for example, present frame is than receiving that first picture frame behind the capture command is clear).

In application, image is caught and must be accomplished in the finite time after receiving capture command.Estimate the acutance or the S of increase at prediction device _j＞S _cBefore, fault or undesired image state possibly cause unacceptable delay.For fear of unacceptable delay, the determining device 140 among Fig. 1 comprises that an overtime compensation is used for catching next picture frame and getting into video memory 115 through buffer 150 from data pick-up 110.Wherein, When receiving capture command and next frame counting will surpass the maximum quantity (" maximum " in the determination step 470 of Fig. 4 A) of predetermined image data frame the time; Determining device 140 is confirmed overtime and instruction processing unit carries out image processing 160 and next picture frame is captured to image memory (storage) 170 from imageing sensor 110, corresponding to 460,465,470 and 480 steps among Fig. 4 A.

Alternatively, when determining device 140 did not determine to catch next image, other subframe of pixel was read from imageing sensor 110 and is handled by mass detector 120 and prediction device 130.Certain scope of the present invention comprises some other execution mode with spirit; Other execution mode adopts other other method and/or algorithm, and other method and/or algorithm are used for determining whether to catch next image according to the quantity of the output of prediction device and frame and/or the efflux behind the capture command.

Fig. 4 A and Fig. 4 B have shown the further details that is used for image evaluation and the execution mode of catching.At first, be used to keep the parameter and variable (for example, the D of mass detector and prediction device _Max, S _c, S _Ij, D _IjDeng) register and number of frames counter be initialised at frame 410.Then, the sub-frame number counter is initialised at frame 420.Received number of frames depends on the accumulation sum of received subframe.Tally function is merged in some embodiments.And, comprise that the counting that increases frame counter 461, test frame counter 470, increase sub-frame number 441 and resetting sub number of frames 447 can adopt other method to accomplish and have the still less method of step such as employing.

Only, comprise another execution mode of method among Fig. 4 A and Fig. 4 B, keep the counting of the accumulation sum of subframe as an example.In above example, the accumulation sum of subframe comprises sub-frame number (it is the total sub-frame number of accumulation of mould that sub-frame number equals with 4) and number of frames (number of frames is the integer part of total sub-frame count of accumulation divided by 4 gained merchants).

Next, frame is obtained at frame 430, and subframe is with the sharpness value of subframe 442 as subframe, and acutance value difference D _MaxQuilt is estimated and is stored in the register 445.In some embodiments, sharpness value is the formula assessment that provides according to above.Yet in scope of the present invention and spirit, various other methods and sharpness value are measured the quality that can be used to evaluate image, comprise the measurement of other dependence filtering image HFS.

Each

module

430 and 435 among Fig. 4 A comprises the step that module 441 is enumerated to module 447 among Fig. 4 B.At first, sub-frame number is increased in 441; In 442, the part of sub-frame data is read from imageing sensor, and depends on that the item (term) of sub-frame data is assessed.In 442-444, the other part of sub-frame data is read and is assessed, and handle up in 445, accomplishing subframe, and the sharpness value characteristic of subframe is assessed and is stored in the processor readable medium.Step 441,442,444,445 is repeated until that whole frames is processed in module 446.After whole frames was processed, in 447, sub-frame number was reset.

When a frame is read and assesses in module 430, another frame is obtained and is assessed up to receiving capture command 450.When receiving capture command, prediction device is estimated the quality of next frame in module 455, and standard is caught in assessment based on estimated quality in module 460.If suitable based on the standard of catching of estimated quality, determining device will be ordered next frame will be captured video memory in module 480, and in module 490 these EPs (end of program).But if improper based on the standard of catching of estimated quality, then frame counter is increased in module 465, and determining device tests whether frame count has arrived predetermined maximum quantity " Max " in module 470 in module 460.If frame counter has arrived maximum, this determining device will be ordered next frame will be captured video memory in module 480, and in module 490 these EPs (end of program).Otherwise, obtained and assessed at module 435 another frames.

Another program of the present invention is to adopt the method and system of memory relatively in a small amount.Fig. 5 has showed and a kind ofly has been used to store pixel data so that according to the buffer storage of Fig. 1 and method shown in Figure 4 assessment key element.According to this execution mode; The memory space of a relatively little line buffer 510 and a small amount of word register enough is used to carry out acutance detector 120 shown in Figure 1, prediction device 130 and determining device 140; Wherein, relatively little line buffer 510 comprises the memory that is used to store 4 line pixel memories.

In one embodiment, be used to assess sub-frame number S _IjCalculating operation comprise as foregoing accumulation convolution key element g (x, y).In this embodiment, five row and the five row parts 530 (with reference to Fig. 5) that comprise the view data of working as the front of the most recent that receives from imageing sensor of four continuous lines (OK) of convolution matrix coefficient (as shown in Figure 3), present image sub-frame data 540 (with reference to Fig. 5) are adopted in the assessment of key element.

Pixel data when Fig. 5 has also showed the assessment convolution moves.The pixel value 575 of an input, from current subframe when front (being processed row), receive and be stored in five rightmost unit 576 of depositing buffer 530 of short logic from imageing sensor.Four of that line other introducing pixel datas are stored in the contiguous buffer of depositing, so that read (not too new on the left side, on the right up-to-date).Except row section part 520; The continuous lines (OK) of received view data before the lead-in of subframe; Corresponding to by the row (for example, with convolution key element from n=-2 to n=1 corresponding) of pointer y-2, four line buffers 510 have been maintained to the y+1 appointment.Thereby clearly, row section 520 comprise from imageing sensor receive to five row pixel datas deposit buffer 530 data when the row in front (OK).

In " window " 540 of line buffer 510 and short five row deposit pixel data in the buffer 530 enough assess convolution and a key element.Assessment is through with 5 x, the 5 matrix w (m among Fig. 3; N) element multiply by the respective value f (x+m that is stored in the dashed region; Y+n), wherein, dashed region comprises that five of line buffer 540 is listed as and five row are deposited buffer; Thereby and have the product of symbol to be added to accumulation register accumulation these to comprise g (x, 25 key elements y).This assessment only needs: 6 of matrix among Fig. 3 (this matrix has 4-folding symmetry) can distinguish coefficient, from five respective column from the pixel data of the up-to-date reception of imageing sensor in five row 540 of the pixel data of 4 continuous lines of the view data of buffering and the buffer register 530.Key element g (x, y) by the assessment after, its absolute value is formed and g (x, absolute value y) are run up to one and are comprised S _IjMolecule part and register.

G (x, y) by after the assessment, shown in the bottom of Fig. 5, the window of line buffer and the pixel data of the up-to-date reception row in-position 550 of being advanced to the right.The oldest data that [x-2, y-2] locates in the window 540 of this figure the first half no longer need key element to estimate.The data rewriting that its is located by [x-2, y-1] (for example, shown in arrow 561, abandoned effectively and replaced) by data from position shown in the arrow 562.Therefore, each data among the row x-2 of four line buffers 510 are all upwards promoted delegation, shown in arrow 562,564 and 566: [x-2; Y-1] data located are moved to position [x-2, y-2] 562, [x-2; Y] data located are moved to position [x-2, y-1], [x-2; Y+1] data located are moved to [x-1, y] 564, and the oldest pixel data [x-2 that is positioned at five alignment buffer registers, 530 leftmost positions; Y+2] be sent to [x-2, y+1] positions of 4 line data buffers, shown in arrow 568.Next, other value quilt in the five row introducing pixel data buffer registers 530 is to the row that move to left, like 570,571,572,573 signals of arrow.

Next be read the rightest row of buffer register 530 from next pixel data of imageing sensor reception.Next key element, (x+1 y) can be according to the front to g (x, method assessment y) for g.The assessment convolution and key element, promote the pixel data position and advance the to the right program of row of window be repeated until S _IjAll capable key elements of y all accumulated.The window's position on the left side is subsequently restarted and the key element of next line y+1 is assessed and accumulated after the same method.After all provisional capitals of subframe i are processed and S _IjAccomplish.

Yet (x y) needs data to surpass the girth (convolution that in form, is used for [x, y] with need from the data of the row of adjacent two row and all directions) of this subframe in two row that defined a sub-frame border or two row, to assess g.In various execution modes, the standard technique assessment that these boundary values can be through being used for the border (for example, extrapolation, image method, the value of supposing first available rows, intercept method, or the like).In this easy execution mode, filter [W-2 x H-2] on the subframe be limited to minimizing thus be used to assess this key element from the row and column of the physics of imageing sensor.Certainly the pointer and the constant that are used for boundary value can be adjusted according to employed standard technique.And, be appreciated that with term " left side " " right side " " on " the moving of the data described of D score, or the moving of row and column just is used for explaining.These terms can be exchanged in various execution modes, perhaps use other term.Although the data structure of logic be convenient to indicate in these terms, the physical storage unit of this media arranges that with the whole bag of tricks this depends on concrete application usually.

Can find out that the line buffer and a spot of storage register that are used for the relatively small amount line are enough carried out prediction device 130 and determining device 140.In one embodiment at j ^ThComprise four sub-frame in the frame, four sharpness value S _IjThe characteristic motion.Prediction device depends on D in this execution mode _Max, D _3j, D _4j, k ₁, k ₂Determining device 140 depends on S _j, S _c, prediction device output and the frame selected maximum quantity.Therefore can adopt about 15 to be used to store the constant of motion and the register cell execution prediction device and the determining device of value characteristic.Certainly, according to concrete application, other filter method and/or different filters can be used in scope of the present invention and the spirit, comprise the filter convolution matrix of 5 x, 5 matrixes shown in being greater than or less than.According in the execution mode of the example of being showed at one, the quantity (execution mode shown in Figure 5 is four line buffers) that is used for the line buffer of evaluate image acutance is less than the dimension of this convolution matrix (matrix in the execution mode shown in Figure 3 is 5 dimensions).In addition, the little buffer of depositing is used to store some pixel datas, and quantity equals the dimension of filtered matrix.Therefore, the advantage of the execution mode of this method and system is only to need little buffer and register, thereby has reduced cost.

Fig. 6 has showed the reduction procedure of data flow of the digital camera of execution mode of the present invention.This camera has processor 650.This camera also has machine readable medium, and machine readable medium comprises: be processed holder, the instruction that is used for implementation quality detector and prediction device and determining device 630 that device is used for the instruction and data of executive control program 640, be used to carry out the filter coefficient of high frequency acutance filter 620 data storage, be used to carry out other function that comprises mass detector and prediction device and determining device selectivity constant data storage and be used for storing and reception value and result's data storing register through processor.And camera also has an imageing sensor.Respond an image, this imageing sensor will comprise that in order the image sensor pixel data 610 of the line of image sensor pixel are sent to line buffer 670.In this embodiment, processor 650 does not directly lead to the data in the line buffer 670.Control circuit automatically is mirrored to mirror registers 660 with the part of line buffer 670.This mirror registers 660 is processed the device inquiry.The data division 660 of this mirror image comprises the pixel data matrix, and the pixel data matrix comprises the assessment S as shown in Figure 5 that is used in

holder

540 and 530 _IjData.

This camera also comprises image-signal processor (ISP) 680; ISP is used for handling the pixel data 610 from the register 675 of media; And convert pixel data to other form; Such as Joint Photographic Experts Group's form (JPEG) of compression, so that in picture storing media 695, catch and store.Vision sensor data through control circuit from imageing sensor by pipeline transmission to buffer 670, register 675 and ISP independently.

In one embodiment, control program direct control in the media 640 and data flow between the program module of mass detector, prediction device and determining device, according to method shown in Figure 4.Processor implementation quality detector, visit are stored in filter matrix coefficient and the constant in the media 620 and the result are kept at register media 665.Prediction device of selecting and determining device constant are also in media 620.On the other hand, according to view data (for example, g (x, y), S _Ij, D _MaxDeng) assessment quality, be stored in register media 665 and from here the visit.When next image was caught in the determining device decision, it exported a lock-on signal to gate control circuit 690 on holding wire 665.Gate control 690 is captured to picture holder 695 with next image from ISP, if and if only if determining device output lock-on signal.

Although mass detector, prediction device and determining device are processed device and carry out and to be used to realize program coding, some in these functions or execution of all these function Be Controlled circuit in other embodiments in one embodiment.And; Although mirror registers 660 is based on 5 x, 5 matrix high-pass filter view data with mass detector 630 and comes the detection image quality in this embodiment; But in various other execution modes; Mass detector by based on wavelet transform or other method that is suitable for surveying acutance carry out, this depends on concrete application.

Fig. 7 has shown when an execution mode that adopts the anti-camera that quivers is taken pictures, frame acutance S _jPoor one by one D between (solid line 710) and the current and previous subframe acutance _Ij=S _Ij-S _Ij-1The differentiation of (dotted line 720).Be used for S _jLeft-hand side vertical axis 722 be used for D _IjRight-hand side vertical axis 724 all indicate arbitrary unit.Trunnion axis unit is the sum of subframe, N, and it is just assessed from zero-time N=0.When camera be placed in be ready to catch the state of image after, it begins to assess subframe.S _jAnd D _IjCurve begin picture from N=4 (frame 1).The N=4 place comprises that four acutancees that are numbered first picture frame 781 of 1 to 4 subframe are assessed.The acutance 781 of picture frame 1 is approximately 167 units.Camera has the frame timeout parameter of 10 preset frames in execution mode shown in Figure 7.

Shown in Figure 7, after picture frame 1 was by assessment, the more images frame was read and assesses.Just handling the 20th ^ThReceive capture command after the subframe 735.The 20th ^ThThe 5th the complete graph picture frame that subframe begins corresponding to the source from the N=0 curve.Therefore picture frame 6 corresponding to point 786, is first picture frame assessed that receives behind the capture command.And, the 10th behind the capture command ^ThPicture frame is the picture frame 15 at 796 places among Fig. 7.This is the picture frame that after receiving capture command, has maximum 10 frame timeouts.Therefore determining device can not caught up to picture frame 14 according to overtime rewriting (timeout override) decision and assessed.

After capture command is received (N=20 in this example), motion has appearred increasing, and this can be by the negative acutance value difference D of subframe 21 to 24 appearance _IjConfirm.Response should increase motion, received that the acutance of next whole frame of locating at N=24 (frame 6) behind the capture command reduces with respect to frame 5.Catch the acutance of instruction first picture frame afterwards, be used as the S in the formula of front by determining device _c=S ₆, have a value S ₆≈-127.In following 3 sub-frame processes, moving continues to increase, and 25-27 is like negative D _Ij(D _Ij＜0) confirms.Less improving (less motion, D have appearred in subframe 28 _Ij＞0) motion that, still increases at subframe 25-27 has surpassed the relatively little improvement at subframe 28 places.Therefore at the motion S at subframe 28 places _C+1(S ₇) tolerance, it comprises the effect from four sub-frame N=25～28, than S ₆(S _c) worsen, from S ₆≈-127 drops to S ₇≈-318 (S _C+1).Because S _C+1≈-318＜S _c≈-127 is not so determining device is caught.

At the relatively stable (D of subframe 29 place's levels of motion _Ij=0) and from subframe 30 setting in motion ranks significantly descends.32 subframe acutancees continue to improve (D from subframe 30 to subframe _Ij＞0).This causes frame S ₈=S _C+2The acutance at 788 places increases fully, and it reaches local maximum about 240.Although at this frame S ₈＞S _cBut the acutance between the continuous subframe in the back of frame 8 is improved speed and is reduced (D _4j＜D _3j, j8).So next image motion is not estimated to become to improve (because proposition 1 all is false with proposition 2) and determining device does not determine to catch next picture frame.

From subframe 34 beginnings, 790 places, motion increases (D again _Ij＜0) causes the deterioration again of frame acutance.At frame 9,792 places, the acutance between two continuous subframes in the back of frame is improved speed and is reduced again, and the acutance that for example increases worsens speed, (D _4j＜D _3j＜0, j=9).Therefore, determining device does not determine to catch next frame.Weaken the D that this can be increased to subframe 40 (label 794) dullness by the subframe 36 (label 793) from frame 10 yet worsen speed afterwards at frame 9 (subframe 36) _IjConfirm.

Assess (subframes 40 at label 794 places) when frame 10 quilts, prediction device is estimated next picture frame motion will be less than the current image frame motion because of 2 establishments of assigning a topic.This can clearly find out from following consideration.D _4j＞D _3j, (j=10) be last key element that proposition 1 and proposition 2 all require.Next, can find out at frame 10 (subframe 40) D _4j＜0.Therefore assigning a topic 1 is false and second key element continuing assessment proposition 2 moved to determine whether that next picture frame moves to be estimated less than current image frame.| D _Ij|＜k ₂* D _Max(k ₂=1/2), through selecting in the spacer segment that after capture command, begins after the subframe of frame 40, to finish from frame 20 | D _Ij| maximum at first confirm D _MaxCan see | D _Ij| arriving relative maximum is in subframe 31 (787) at interval, wherein D _I.j≈ 240.At frame 10, (subframe 40) D _4.jBe approximately-18 units.So inequality: D _4.j≈-18＜k ₂* D _Max≈ 1/2*240=120 satisfies condition at frame 10.Thereby, set up and prediction device is estimated next image motion and will be moved less than current image frame in frame 10 proposition 2.

In addition, in frame 10 (subframe 40), the determining device standard that is used to catch next picture frame meets, because: 1) received capture command, 2) prediction device is estimated acutance increases, and 3) the frame acutance of frame 10 is greater than S _c: S ₁₀≈-54＞S _c=S ₆≈-127.After frame 10 is confirmed to catch, next frame 796, the frame 11 at subframe 44 places is hunted down and is stored in the imageing sensor.D _IjCalculated value and captive frame after some continuous subframes also in Fig. 7, showed to compare.Can find out that the characteristic of frame 11 is that its acutance is greater than any frame before or subsequently.Therefore in this example, method of the present invention shown in the time period (70 frames in about 2,3 seconds, such as, per second 30 frames) catch frame with maximum sharpness.

Fig. 8 has shown an a kind of example of personal device, and it can be used as digital camera or similar device.Such device can be used to carry out various functions, according to concrete application, such as telephonic communication, bidirectional paging communication, individual tissue, global positioning system or similar functions.This comprises the computer part of the device of digital camera computer system 800 expressions.Computer 800 is connected to external system through communication interface 820.This interface generally be some forms of cable and/or radio interface so that be used with a direct available PC, and can comprise radio interface so as with network communication such as 802.11 wireless networks.Certainly the diverse network and the means of communication are such as bluetooth, infrared interface, mobile phone interface and other, and this depends on concrete application.

Computer system 800 comprises processor 810; It can be the combination of conventional microprocessor such as Intel Pentium microprocessor, IBM power PC microprocessor, Texas tool count code signal processor or various types of processors, and this depends on concrete application.Memory 840 is coupled to processor 810 through bus 870.Memory 840 can be dynamic random access memory (DRAM) and can comprise static memory (SRAM), flash memory, magnetic memory and other type that this depends on concrete application.Bus 870 is attached to memory 840, non-volatile storage 850, display controller 830 and i/o controller (I/O) 860 with processor 810.In some embodiments, the various combinations of these elements are integrated in the integrated circuit or are integrated in the combination of a plurality of integrated circuits that are arranged in same encapsulation.It is pointed out that display controller 830 and I/O controller 860 often are integrated in together, and this display can also provide output.

Display controller 830 is with traditional demonstration control mode control display device 835, and display device 835 generally is LCD (LCD) or similar flat-panel monitor, little form factor display.Input/output device 855 can comprise keyboard or stylus and touch-screen; Sometimes can also expand hard disk drive, printer, scanner and other input and output device; Comprise mouse or other pointing device, such as being connected to some docking stations or PC when camera.Display controller 830 can adopt traditional well-known technology to carry out with I/O controller 860.Digital image input unit 865 can be the digital camera that comprises an execution mode of the present invention, its be coupled to I/O controller 860 or through independent connection so that allow image to be imported into this device 800.

Non-volatile storage 850 is flash memory or read-only memory normally, perhaps both combinations.The holder that is used for a large amount of storage datas of magnetic disc, laser disc or other form also can be used in some execution modes, the general permanent components that is installed to be this device 800 of repelling of the form factor of the device of even now.Certainly, the big storage capacity device on other computer generally can install the use that combines of 800 comparatively limited holder with this.Handle write memory 840 through direct memory access in the software execute process of in these data some in the device 800 of being everlasting.Those skilled in the art can recognize at once that this term " machine readable medium " or " computer-readable media " comprise the accumulator apparatus of be processed device 810 visits of any type, and comprises the carrier wave of encoded data signal.

Device 800 is multiple examples that have in possibly the installing of different characteristic.For example, have multiple bus usually based on the device of Intel microprocessor, wherein one can be I/O (I/O) bus that is used for ancillary equipment and another direct connection processing device 810 and memory 840 (so-called memory bus).Each bar bus links together through the bridge parts, and owing to different bus protocols, the bridge parts are carried out necessary conversion arbitrarily.

In addition, device 800 is by operating system software control, and operating system software can comprise file management system, and such as disc operating system, it is the part of operating system software.An example with operating system of associated with management system software is the family of Windows

operating system of the Microsoft in usually said washingtonian Lei Mengde district, and their related file management systems.Another example with operating system of associated with management system software is

operating system and its related file management system.Yet a common problem of digital camera is that they have the document management software and related user interface that falls behind very much.File management system generally is stored in the non-volatile storage 850 and makes various operations that processor 810 executive operating systems require so that input and output data and storing the data in the memory are included in store files on the non-volatile storage 850.Other operating system can be provided by the manufacturer of device, and those operating systems generally can have device-specific characteristic, and these device-specific characteristics are not the parts of the similar operations system on the similar device.Similarly, or

operating system is suitable on specific device to specific device performance.

Install 800 in some embodiments and can be integrated on the chip or on the core assembly sheet, and generally be installed into little form factor so that as personal device.Like this, all be integrated on the chip processor, bus, one-board memory and display-I/O controller also unrare.Alternatively, function can be divided into a little to putting interconnective several chip, like this, observes from actual device or relevant sketch map, and bus is logically clear but physically not obvious.

A part of describing in detail above be adopt the term of algorithm and in computer memory the symbolic representation to the operation of data bit carry out.These arthmetic statements and expression are that the technical staff of data processing field reports their the work conventional means of essence to the technical staff in other field.The algorithm here, general, be contemplated that the self-consistent operating sequence of leading to required result.This operation is the physical operations of those physical quantitys that need.Usually, although unnecessary, this tittle adopts and can be stored, transmits, combines, relatively and the form of the electrical or magnetic signal of other operation.Referring to these signals with position, value, element, symbol, characteristic, key element, numeral or analog, mainly is the convenience when being directed against common the use.

But, should be kept in mind that all these terms or similar term should related with suitable physical quantity and just be used in the label that makes things convenient on these physical quantitys.Remove non-exclusively statement in addition; Otherwise be to be understood that to running through this specification; The discussion of adopting term to carry out such as " processing " or " calculating " or " confirming " or " demonstration " or " assessment " or similar vocabulary; Be meant the behavior or the processing of computer system or similar computing electronics, it perhaps converts the data manipulation of representing with physical quantity in the RS of computer system to the memory of computer system or other data represented with physical quantity similarly in register or out of Memory storage, transmission or the display unit.Sending out equipment like this can implement in media.

Although those skilled in the art will appreciate that the example of this system and method and execution mode are showed, do not deviating from the spirit and scope of the present invention and can conversion go out various versions.For example, execution mode of the present invention can be used for many dissimilar image-taking systems, imaging device, database, application program and other system.And the characteristic of an execution mode can be incorporated into another execution mode, even those characteristics are not described in one of present specification independent execution mode together.Therefore, scope of the present invention is limited its claim.

Claims

1. method of in digital camera, catching image comprises:

Receive the input of previous picture frame, and form the sharpness value of said previous picture frame, wherein, described previous picture frame input is a sub-frame of said previous picture frame, and said previous picture frame is a picture frame before next-door neighbour's current image frame;

Receive the input of current image frame, and form the sharpness value of said current image frame, wherein, described current image frame input is a corresponding subframe of said current image frame;

Poor according between the sharpness value of the sharpness value of the subframe of the sharpness value of the sharpness value of said previous picture frame, said current image frame and said current image frame and the corresponding subframe of said previous picture frame; The motion of estimating next picture frame will be less than the motion of current image frame; Wherein, The subframe of said current image frame is last subframe of said current image frame, and the absolute value of said difference is less than the maximum and the product of selecting the factor of the absolute value of other continuous acutance value difference;

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

2. the method for claim 1, it further comprises: receive capture command, wherein, the input that receives said previous picture frame is after receiving this capture command.

3. the method for claim 1, it further comprises: receive the input of first picture frame, and form the sharpness value of this first picture frame input; Wherein, said first picture frame is different from said previous picture frame, and the reception of said first picture frame input is before the reception of said previous picture frame input.

4. method as claimed in claim 3; It further comprises: receive capture command; Wherein, the reception of said first picture frame input is before the reception of said capture command, and the reception of said previous picture frame input is after the reception of said capture command.

5. method as claimed in claim 3; Wherein, Said first picture frame input is a sub-frame of said first picture frame, and said previous picture frame input is a sub-frame of said previous picture frame, and said current image frame input is a sub-frame of said current image frame.

6. method as claimed in claim 3 wherein, forms sharpness value and extracts the filter completion through a high frequency.

7. method as claimed in claim 6, wherein, it is a high pass filter that described high frequency extracts filter.

8. the method for claim 1; Wherein, From said previous picture frame input and the input of said current image frame, select the picture frame input, and form the sharpness value of this picture frame input, its process comprises with this picture frame input carries out the convolution summation to the high pass filter matrix.

9. method as claimed in claim 8, wherein, described high pass filter matrix has symmetry.

10. the frame acutance of the frame acutance of said current image frame greater than said previous picture frame depended in the decision of the method for claim 1, wherein catching said next picture frame.

Depend on the frame acutance of the frame acutance of said current image frame 11. the method for claim 1, wherein catch the decision of said next picture frame greater than the arbitrary image frame after the said capture command.

12. a method of in digital camera, catching image comprises:

Poor according between the sharpness value of the sharpness value of the subframe of the sharpness value of the sharpness value of said previous picture frame, said current image frame and said current image frame and the corresponding subframe of said previous picture frame; The motion of estimating next picture frame will be less than the motion of current image frame; Wherein, The subframe of said current image frame is last subframe of said current image frame, and the absolute value of said difference is less than the maximum and the product of selecting the factor of the absolute value of other continuous acutance value difference, wherein; The described selection factor depends on the symbol of said difference, and the described selection factor is less than 1;

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

13. a method of in digital camera, catching image comprises:

Poor according between the sharpness value of the sharpness value of the subframe of the sharpness value of the sharpness value of said previous picture frame, said current image frame and said current image frame and the corresponding subframe of said previous picture frame; The motion of estimating next picture frame will be less than the motion of current image frame; Wherein, the subframe of said current image frame is last subframe of said current image frame;

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

14. a method of in digital camera, catching image comprises:

Poor according between the sharpness value of the sharpness value of the subframe of the sharpness value of the sharpness value of said previous picture frame, said current image frame and said current image frame and the corresponding subframe of said previous picture frame; The motion of estimating next picture frame will be less than the motion of current image frame; Wherein, the absolute value of said difference is less than the maximum and the product of selecting the factor of the absolute value of other continuous acutance value difference;

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

15. a device of in digital camera, catching metastable picture, this device comprise that the image motion of estimating next picture frame when will be less than the parts of the image motion of current image frame, and estimate the parts of catching said next picture frame according to this; Wherein, described estimate and catch by following process accomplish:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

16. a device of in digital camera, catching metastable picture, this device comprise that the image motion of estimating next picture frame when will be less than the parts of the image motion of current image frame, and estimate the parts of catching said next picture frame according to this; Wherein, described estimate and catch by following process accomplish:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

17. a device of in digital camera, catching metastable picture, this device comprise that the image motion of estimating next picture frame when will be less than the parts of the image motion of current image frame, and estimate the parts of catching said next picture frame according to this; Wherein, described estimate and catch by following process accomplish:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

18. a digital camera that is used to catch metastable picture comprises:

The digital image sensor of output and image related data;

The machine readable medium of storage data and instruction;

At least one is carried out said instruction and operates the processor of said data;

Described instruction can be used to carry out;

Be attached to the acutance detector of said imageing sensor, said acutance detector is used for the acutance of evaluate image;

According to said acutance, estimate the prediction device of the image motion of next image; And

Decision is caught next image and said next image is captured the determining device in the said machine readable medium;

Wherein, described estimating by following process accomplished:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

19. digital camera as claimed in claim 18, wherein, described machine readable medium comprises can be used for the less line buffer of line number.

20. digital camera as claimed in claim 18, wherein, described acutance detector is asked convolution with the said data relevant with image to the high pass filter matrix.

21. a digital camera that is used to catch metastable picture comprises:

The digital image sensor of output and image related data;

The machine readable medium of storage data and instruction;

Described instruction can be used to carry out;

Wherein, described estimating by following process accomplished:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

22. digital camera as claimed in claim 21, wherein, described machine readable medium comprises can be used for the less line buffer of line number.

23. digital camera as claimed in claim 21, wherein, described acutance detector is asked convolution with the said data relevant with image to the high pass filter matrix.

24. a digital camera that is used to catch metastable picture comprises:

The digital image sensor of output and image related data;

The machine readable medium of storage data and instruction;

Described instruction can be used to carry out;

Wherein, described estimating by following process accomplished:

Estimate based on said, next picture frame is caught in decision; And

Catch said next picture frame.

25. digital camera as claimed in claim 24, wherein, described machine readable medium comprises can be used for the less line buffer of line number.

26. digital camera as claimed in claim 24, wherein, described acutance detector is asked convolution with the said data relevant with image to the high pass filter matrix.