CN102760296B - Movement analyzing method for objects in multiple pictures - Google Patents
Movement analyzing method for objects in multiple pictures Download PDFInfo
- Publication number
- CN102760296B CN102760296B CN201110110598.6A CN201110110598A CN102760296B CN 102760296 B CN102760296 B CN 102760296B CN 201110110598 A CN201110110598 A CN 201110110598A CN 102760296 B CN102760296 B CN 102760296B
- Authority
- CN
- China
- Prior art keywords
- those
- absolute difference
- block
- picture
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a movement analyzing method for objects in multiple pictures. The movement analyzing method is applicable to an image acquiring device and comprises the following steps of first, obtaining the sum of a plurality of groups of first absolute difference values according to noise of the image acquiring device under a plurality of light sources; obtaining two pictures under the condition that a shooting light source of the light sources is arranged; then calculating the sum of a plurality of second absolute difference values between the two pictures; finding out a plurality of object blocks with object grains; calculating a regional movement vector of each object block according to the sum of the second absolute difference values; calculating a first reliability of each object block according to the sum of the second absolute difference values and the sum of the group of first absolute difference values corresponding to the shooting light source; and estimating the regional movement vectors according to the first reliabilities so as to obtain a universal movement vector. The movement analyzing method can be used for estimating the accurate universal movement vector.
Description
Technical field
The present invention relates to a kind of image processing method, relate in particular to the mobile analytical approach of object in a kind of many pictures.
Background technology
It is a lot of that image noise generates reason, for example, via signal amplifier effect, used heat, camera lens and photo-sensitive cell (sensor) reciprocation, the reasons such as signal transduction process interference all can cause noise, and each photo-sensitive cell coordinates and more has impact in various degree from different camera lenses.At image now, process in application, many pictures (multi-frame) synthetic with the technology of analyzing by extensive discussions and use, but running into a very large difficult problem is how Correct Analysis goes out the kinematic relation between different pictures.The three-dimensional noise reduction (3D noise reduction) of take is example, once mobile analysis result between the different pictures of mistake in using, the image after synthesizing will easily find that there is the situation of image retention (ghost).Analyze the mode that between picture, object moves a lot, but take into account that the analysis speed under high picture update rate (high frame rate) must reach soon, hardware cost was not wished after high factor, simpler efficient analytic approach is the mode of absolute difference and (Sum of Absolute Difference, SAD).
Yet under high noise environment, the words of only analyzing by absolute difference and computing are profiling error very easily.Moreover the logic that the absolute difference of calculating and minimum block is considered as to place, optimal approximation block place also has very large problem.Because, under high noise environment, in fact absolute difference and numerical value not little.For example, if maximum absolute difference and with minimum absolute difference value with while numerically almost there is no difference in Search Area (search region) is that the coordinate points risk that corresponds to is very big the identification of the position of numerical value minimum actually, be easy to make the mistake.In addition, take likelihood function (like lihood function) does object as basis and moves analysis and really can reduce the impact property of noise.But, the needed calculation resources of these class methods is too high, under high picture update rate, with complementary metal oxide semiconductor (CMOS) photo-sensitive cell (Complementary Metal-Oxide-Semiconductor sensor, CMOS sensor), for basic Related product, need especially very high arithmetic speed and usefulness.As this complexity algorithm consuming time be difficult for putting into practice in real image product again.
Summary of the invention
The invention provides the mobile analytical approach of object in a kind of many pictures, can avoid noise object to be moved to the impact of analysis.
The present invention proposes the mobile analytical approach of object in a kind of many pictures, is applicable to an image acquiring device, comprises the following steps.First, the noise according to image acquiring device under a plurality of light sources are set, obtain many groups the first absolute difference and.Each light source set respectively corresponding one group of first absolute difference and.Then, the photographic light sources in these light sources are set obtains one first picture and one second picture by image acquiring device under setting.Then, calculate a plurality of the second absolute differences between the first picture and the second picture with.Afterwards, find out a plurality of object block of the texture process of object in the first picture and the second picture.Afterwards, according to these second absolute differences and, calculate these object block and divide other region motion-vector.Then, according to these second absolute differences and with photographic light sources set corresponding this organize the first absolute difference and, calculate these object block and divide other one first fiduciary level.Then, according to these region motion-vectors of these first Reliability assessments, to obtain a universe motion-vector.
In one embodiment of this invention, the noise according to image acquiring device under these light sources are set, obtain these organize the first absolute difference and step, comprise the following steps.First, under these light sources are set, by image acquiring device, one lamp box is obtained to multiple images respectively.Then, record these images at the first absolute difference of diverse location and.
In one embodiment of this invention, record these images the first absolute difference of diverse location and step, comprise the following steps.Record the first absolute difference of four corner pixels of each block in these images, with the first absolute difference by four corner pixels and interpolation go out in each block picture number the first absolute difference and.
In one embodiment of this invention, record these images the first absolute difference of diverse location and step, comprise the following steps.Record the first absolute difference that in each block of these images, pixel shared and.
In one embodiment of this invention, these light source conditions comprise the light sensitivity of brightness, color and the image acquiring device of lamp box, around in dim light at least one of them.
In one embodiment of this invention, find out the step of these object block, comprise the following steps.First, calculate each block between the first picture and the second picture maximum absolute difference and with minimum absolute difference value with.Then, according to maximum absolute difference in each block and with minimal difference and difference, judge that each block is for these object block one of them or a background block.
In one embodiment of this invention, according to these region motion-vectors of these first Reliability assessments, to obtain the step of universe motion-vector, comprise the following steps.First, according to maximum absolute difference in each block and with minimal difference and difference, calculate these object block and divide other one second fiduciary level.Then, according to these first fiduciary levels and these region motion-vectors of these the second Reliability assessments, to obtain universe motion-vector.
In one embodiment of this invention, according to these first fiduciary levels and these region motion-vectors of these the second Reliability assessments, to obtain the step of universe motion-vector, comprise the following steps.First, according to these first fiduciary levels and these the second fiduciary levels, calculate a plurality of average coherences.Then, according to these region motion-vectors of these average coherence weightings, to obtain universe motion-vector.
In one embodiment of this invention, according to these second absolute differences with photographic light sources, set the corresponding step that this organizes the first absolute difference and calculates these the first fiduciary levels, comprise the following steps.First, calculate this organize the first absolute difference and a standard deviation tolerance.Then, calculate these second absolute differences and with this organize the first absolute difference and a ratio.Come again, than correlative value and standard deviation tolerance, to obtain these the first fiduciary levels.
Based on above-mentioned, the present invention by obtain be in advance subject to the first absolute difference that noise impact produces and, then coordinate the second absolute difference that between two pictures, diverse location calculates and carry out assessment area motion-vector, and can evaluate universe motion-vector comparatively accurately.
For above-mentioned feature and advantage of the present invention can be become apparent, special embodiment below, and coordinate accompanying drawing to be described in detail below.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the mobile analytical approach of object in many pictures of first embodiment of the invention.
Fig. 2 is the schematic diagram of the picture that obtains of image acquiring device.
Fig. 3 is the curve synoptic diagram of relative the second, first absolute difference of the first fiduciary level of Fig. 1 and ratio.
Fig. 4 is the process flow diagram of the mobile analytical approach of object in many pictures of first embodiment of the invention.
Fig. 5 A and Fig. 5 B are respectively the picture view in order to declare record the first absolute difference and two kinds of modes.
Fig. 6 in order to illustrate maximum absolute difference and with minimum absolute difference value and picture view.
Fig. 7 in order to illustrate the second fiduciary level with maximum, least absolute value is poor and the curve synoptic diagram of relation.
Reference numeral:
B, B1: block
D: standard deviation tolerance
E, L: pixel
F: picture
F1 F2 F3: image
M: storer
O1-O7: object block
P1: position
S110-S170, S210-S274: step
SADmax: maximum absolute difference and
SADmin: minimum absolute difference value and
Embodiment
Fig. 1 is the process flow diagram of the mobile analytical approach of object in many pictures of first embodiment of the invention.In the present embodiment, the flow process of Fig. 1 is applicable to the image acquiring device (not shown)s such as digital camera, digital code camera.Please refer to Fig. 1, first carry out step S110, the noise according to image acquiring device under a plurality of light sources are set, obtain many groups the first absolute difference and.Each light source set respectively corresponding one group of first absolute difference and.For instance, can by image acquiring device, take in advance the lamp box of different brightness.After the diverse location of lamp box image is calculated, can obtain the first absolute difference that simple noise causes under different brightness and.
Specifically, can from low-light level (LV2) road, photograph high brightness (LV14) to lamp box, and corresponding various nitometer calculate lamp box image the first absolute difference and.In order to save storer, the first absolute difference under can a recording section progression brightness and.For example only record first under LV2, LV5, LV10 certainly to difference and, and the first absolute difference of LV3, LV4 etc. and available interpolation calculation are out.Except brightness, the also situation of image recordable acquisition device under different light sensitivity, for example light sensitivity 100Yi photographed on road light sensitivity 3200.In another embodiment, can also take by image acquiring device contrast colors chart (color chart), to measure in advance the impact of different color blocks on color noise (color noise).
In addition, in another embodiment, also can take into account the camera lens impact of color dim light (lens color shading) around.Around the noise of color can be more serious for image in theory, and we make example with a situation.Under light sensitivity 800, lamp box brightness LV10, to lamp box take lower different position all calculate the first absolute difference and, and record.Because the noise at dim light place, corner is conventionally comparatively serious, therefore the first absolute difference and numerical value should be larger.Along with light sensitivity is higher, the first absolute difference and also should be larger.For convenient, store, lamp box image can be divided into a plurality of blocks, and be recorded in the first absolute difference of calculating block center and.Because noise impact can allow the first absolute difference and have error range, thus also can calculate its mean value and standard deviation by the result of recording under many experiments, to obtain the data of calibration (calibration).For instance, compared to light sensitivity 800, the first absolute difference under light sensitivity 3200 and standard deviation can be higher.
It is worth mentioning that, according to the above-mentioned practice, can first try to achieve image acquiring device photo-sensitive cell noise characteristic and record in advance.In addition, though to same block carry out the first absolute difference and calculating, each time to numerical value also have deviation, therefore when recording the information of this otherness, the otherness of the first absolute difference and variation also can be recorded.In addition, because of the image of analyzing, by dull scene (VERIVIDE COLOR ASSESMENT CABINET), take and obtain again.Therefore impact property that, can the random noise of independent analysis (random noise).
Then carry out step S120, the photographic light sources in these light sources are set obtains one first picture and one second picture by image acquiring device under setting.Then carry out step S130, calculate a plurality of the second absolute differences between the first picture and the second picture with.Fig. 2 is the schematic diagram of the picture that obtains of image acquiring device.Please refer to Fig. 2, after having obtained two continuous picture F, picture F can be divided into a plurality of block B.Then, to these blocks B calculate respectively the second absolute difference and.
Carry out afterwards step S140, find out a plurality of object block of the texture process of object in the first picture and the second picture.Take Fig. 2 as example, in all block B from picture F, can find out 7 object block O1-O7.Carry out again step S150, according to these second absolute differences and, calculate these object block O1-O7 and divide other region motion-vector.In the present embodiment, because the block B beyond object block O1-O7 in picture F belongs to flat region, and these flat regions are vulnerable to noise interference.Therefore, the present embodiment can be ignored flat region, to avoid flat region to disturb.Also or, the weight of object block O1-O7 is improved, also can reduce the interference of flat region.
Then carry out step S160, according to these second absolute differences and with photographic light sources set corresponding this organize the first absolute difference and, calculate these object block O1-O7 and divide other one first fiduciary level.For instance, can for example, for example, according to user's shooting setup parameter and brightness analysis result (being obtained by camera automatic exposure (AE) algorithm) and the setting of hue analysis result (being tried to achieve by Automatic white balance (AWB) algorithm) taking-up photographic light sources.Suppose that the scene brightness analysis result that user is taking is LV10, light sensitivity is set as 800, and the form and aspect of taking block are muted color, enters system storage these are set to lower the first absolute difference and the taking-up of calibration in advance.Take out the first corresponding absolute difference and afterwards, then calculate the second absolute difference and with the first absolute difference and ratio, to obtain the first fiduciary level.
Fig. 3 is the curve synoptic diagram of relative the second, first absolute difference of the first fiduciary level of Fig. 1 and ratio.Please refer to Fig. 3, the second, first absolute difference and ratio approach very much at 1 o'clock, represent the second absolute difference and operation result and the first absolute difference of in advance calibration and very approximate.Namely, ratio approaches the impact that this block of 1 is only subject to random noise.In addition, because noise has certain variability (variance), so standard deviation tolerance d of definable.Being positioned at the ratio in the interval of 1-d-1+d, is all may be the result that affects of pure random noise.Furthermore, because the sampling of the overwhelming majority in stochastic distribution theory is near still can relatively to approach at ratio be 1, the fiduciary level is here still the highest.Therefore, standard deviation tolerance d can rely on the first absolute difference that step S110 records and otherness set.Under low speed, d value is less; Under ISO, d value is larger.In addition, in practical application, the curve of Fig. 3 can be first through the processing of smoothing, not as limit.
Then carry out step S170, according to these region motion-vectors of these first Reliability assessments, to obtain a universe motion-vector.For instance, the weight of the region motion-vector that the first fiduciary level is high must improve, and the too low region motion-vector of the first fiduciary level is ignored, and therefore can obtain a universe motion-vector result comparatively accurately.
In the present embodiment, we propose to increase the method for the reliability that absolute difference and object analysis move, and can take under situation and adjust and analyze fiduciary level with difference.The example that is applied as with three-dimensional noise reduction, when analyzing, fiduciary level is high, the weight (weighting) of frame filter (temporal filtering) is just heightened, otherwise turns down the weight of frame filter, makes result be approximately pure spatial filtering (spatial filtering).Not only can avoid on the one hand analyzing because mistake moves the generation that causes ghost, improve the image quality after processing simultaneously.
Fig. 4 is the process flow diagram of the mobile analytical approach of object in many pictures of first embodiment of the invention.Please refer to Fig. 1 and Fig. 4, the present embodiment and last embodiment are similar, and its resemblance will repeat no more, and also can quote mutually at its difference place, not as limit.First carry out step S210, the noise according to image acquiring device under a plurality of light sources are set, obtain many groups the first absolute difference and.In the present embodiment, step S210 can comprise two sub-steps such as S212 and S214.First carry out step S212, under these light sources are set, by image acquiring device, one lamp box is obtained to multiple images respectively.In the present embodiment, these light source conditions can comprise the light sensitivity of brightness, color and the image acquiring device of lamp box, around in dim light at least one of them.
Carry out afterwards step S214, calculate and record these images at the first absolute difference of diverse location and.Fig. 5 A and Fig. 5 B are respectively the picture view in order to declare record the first absolute difference and two kinds of modes.Please refer to Fig. 5 A and Fig. 5 B, for letter economizes system storage, the first absolute difference and the available following two kinds of modes that record simplify:
Please refer to Fig. 5 A, the data of supposing storer M only put 64x64 pen the first absolute difference and numeral, but true picture F1 size is 4096x4096.Now, the size of the corresponding block B1 in true picture F1 is for long: (4096/64)=64; High: (4096/64)=64.That is to say, the pixel of the 64x64 in block B1 just share this opposite position P1 in storer M the first absolute difference and.
Refer again to Fig. 5 B, we also can with bilinear interpolation method (bilinear interpolation) obtain each first absolute difference and numerical value.For instance, for whole image F2, can record in advance each block four corner pixels E the first absolute difference and.The first absolute difference that each pixel L is corresponding in wanting calculation block and time, can reach by bilinear interpolation algorithm.For example say that image size is 4096x4096, can take out the corner pixels E of 65x65 end points, each block size is 64x64.Thus, can by the first absolute difference of four corner pixels E and interpolation go out picture number L in each block the first absolute difference and.
Then carry out step S220, the photographic light sources in these light sources are set obtains one first picture and one second picture by image acquiring device under setting.Then carry out step S230, calculate a plurality of the second absolute differences between the first picture and the second picture with.
Carry out afterwards step S240, find out a plurality of object block of the texture process of object in the first picture and the second picture.In the present embodiment, step S240 can comprise sub-step S242 and step S244.First carry out step S242, calculate each block between the first picture and the second picture maximum absolute difference and with minimum absolute difference value with.Specifically, Fig. 6 for in order to illustrate maximum absolute difference and with minimum absolute difference value and picture view.Please refer to Fig. 6, supposing has N block in image F3, and each block can calculate corresponding a maximum absolute difference and SADmax and minimum absolute difference value and a SADmin.
Then carry out step S244, according to maximum absolute difference in each block and with minimal difference and difference, judge that each block is for these object block one of them or a background block.In general, there are contour of object or edge process in this region of [SADmax-SADmin] larger expression, and this regional change of [SADmax-SADmin] less expression is too small is likely flat region or the region changing without any movement.Therefore, the present embodiment can preset a threshold value.When [SADmax-SADmin] of a block is greater than threshold value, judge that this block is object block; When [SADmax-SADmin] of a block is less than threshold value, judge that this block is background block.
Carry out again step S250, according to these second absolute differences and, calculate these object block and divide other region motion-vector.Then carry out step S260, according to these second absolute differences and with photographic light sources set corresponding this organize the first absolute difference and, calculate these object block and divide other one first fiduciary level.In the present embodiment, step S260 can comprise sub-step S262-S266.First carry out step S262, calculate this organize the first absolute difference and a standard deviation tolerance.Then carry out step S264, calculate this organize the second absolute difference and with these first absolute differences and a ratio.Carry out again step S266, than correlative value and standard deviation tolerance, to obtain these the first fiduciary levels.
Then carry out step S270, according to these region motion-vectors of these first Reliability assessments, to obtain a universe motion-vector.In the present embodiment, step S270 can comprise sub-step S272 and S274.First carry out step S272, according to the difference of maximum absolute difference and SADmax and minimal difference and SADmin in each block of image F3, calculate these object block and divide other one second fiduciary level.Fig. 7 in order to illustrate the second fiduciary level with maximum, least absolute value is poor and the curve synoptic diagram of relation.Please refer to Fig. 7, because this block of [SADmax-SADmin] larger expression has contour of object or edge process, with this block analysis, move and will have information comparatively reliably so.Therefore, the second fiduciary level of this block can be improved; It is likely flat region or without any the mobile region changing that this block of [otherwise SADmax-SADmin] less expression changes too small.Therefore, if move with this block analysis, cannot obtain reliable information, so second fiduciary level in Jiang Ci district reduces.That is to say, the second fiduciary level that [SADmax-SADmin] is greater than threshold value (turning point in Fig. 7) improves, and [SADmax-SADmin] is less than the second fiduciary level reduction of threshold value.In addition, in practical application, the curve of Fig. 7 can be first through the processing of smoothing, not as limit.
Then carry out step S274, according to these first fiduciary levels and these region motion-vectors of these the second Reliability assessments, to obtain universe motion-vector.Specifically, can, first according to these first fiduciary levels and these the second fiduciary levels, calculate a plurality of average coherences.Then according to these region motion-vectors of these average coherence weightings, to obtain universe motion-vector.It is worth mentioning that, the weight of the region motion-vector that average coherence is high must improve, and the too low region motion-vector of average coherence is ignored.Therefore, can obtain a result for universe motion-vector comparatively accurately.
In sum, the present invention by obtain be in advance subject to the first absolute difference that noise impact produces and, then coordinate the second absolute difference that between two pictures, diverse location calculates and carry out assessment area motion-vector, and can evaluate universe motion-vector comparatively accurately.In addition, due to absolute difference and lower for cost and the demand of hardware.Therefore, the present invention can save the required cost of hardware effectively.In addition, the present invention has the elasticity of height, can change because of size, calculated amount, the operation time of storer the analytical approach of block in algorithm (block-wise) or pixel (pixel-wise).Moreover, under the colour mixture (blending) of many pictures, also can avoid the generation of image retention (ghost).Therefore, can avoid user to be difficult to the defect of accepting.In addition, analyze the accuracy of colour mixture is improved because mobile, can obtain in flat region best noise reduction quality, Er edge also can retain the most correct details and contour of object.
Although the present invention discloses as above with embodiment, so it is not in order to limit the present invention, and the those of ordinary skill in any affiliated technical field, when doing a little change and retouching, and does not depart from the spirit and scope of the present invention.
Claims (8)
1. more than, a mobile analytical approach for object in picture, is applicable to an image acquiring device, comprising:
Noise according to this image acquiring device under a plurality of light sources are set, by obtain multiple images obtain many groups the first absolute difference and, wherein each light source set respectively corresponding one group of first absolute difference and;
A photographic light sources in those light sources are set obtains one first picture and one second picture by this image acquiring device under setting;
Calculate a plurality of the second absolute differences between this first picture and this second picture and;
Find out a plurality of object block of the texture process of this object in this first picture and this second picture;
According to those second absolute differences and in corresponding to those object block one of them at least one the second absolute difference and, calculate this one of them the region motion-vector of those object block;
Calculate this photographic light sources set corresponding this organize the first absolute difference and a standard deviation tolerance;
Calculate those second absolute differences and with this organize the first absolute difference and a ratio;
Compare this ratio and this standard deviation tolerance, to obtain those object block, divide other one first fiduciary level; And
According to those object block, divide other those object block of the first Reliability assessment to divide other region motion-vector, to obtain a universe motion-vector.
2. the mobile analytical approach of object in many pictures according to claim 1, the noise under those light sources are set according to this image acquiring device wherein, by obtain those images obtain those organize the first absolute difference and step, comprising:
Under those light sources are set, by this image acquiring device, one lamp box is obtained to those images respectively; And
Calculate and record those images at the first absolute difference of diverse location and.
3. the mobile analytical approach of object in many pictures according to claim 2, wherein record those images the first absolute difference of diverse location and step, comprising:
Record the first absolute difference of four corner pixels of each block in those images, with the first absolute difference by these four corner pixels and interpolation go out in each block picture number the first absolute difference and.
4. the mobile analytical approach of object in many pictures according to claim 2, wherein record those images the first absolute difference of diverse location and step, comprising:
Record the first absolute difference that in each block of those images, pixel shared and.
5. the mobile analytical approach of object in many pictures according to claim 2, those light source conditions wherein, comprising:
The light sensitivity of the brightness of this lamp box, color and this image acquiring device, around in dim light at least one of them.
6. the mobile analytical approach of object in many pictures according to claim 1, wherein finds out the step of those object block, comprising:
Calculate each block between this first picture and this second picture maximum absolute difference and with minimum absolute difference value and; And
According to maximum absolute difference in each block and with minimal difference and difference, judge that each block is for those object block one of them or a background block.
7. the mobile analytical approach of object in many pictures according to claim 6, wherein divides other those object block of the first Reliability assessment to divide other region motion-vector according to those object block, to obtain the step of this universe motion-vector, comprising:
According to those blocks one of them maximum absolute difference and with minimal difference and difference, calculate one of them one second fiduciary level of those object block; And
According to those object block, divide other first fiduciary level and those object block to divide other those object block of the second Reliability assessment to divide other region motion-vector, to obtain this universe motion-vector.
8. the mobile analytical approach of object in many pictures according to claim 7, wherein according to those object block, divide other first fiduciary level and those object block to divide other those object block of the second Reliability assessment to divide other region motion-vector, to obtain the step of this universe motion-vector, comprising:
According to those first fiduciary levels and those the second fiduciary levels, calculate a plurality of average coherences; And
According to those region motion-vectors of those average coherence weightings, to obtain this universe motion-vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110110598.6A CN102760296B (en) | 2011-04-29 | 2011-04-29 | Movement analyzing method for objects in multiple pictures |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110110598.6A CN102760296B (en) | 2011-04-29 | 2011-04-29 | Movement analyzing method for objects in multiple pictures |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102760296A CN102760296A (en) | 2012-10-31 |
| CN102760296B true CN102760296B (en) | 2014-12-10 |
Family
ID=47054749
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110110598.6A Active CN102760296B (en) | 2011-04-29 | 2011-04-29 | Movement analyzing method for objects in multiple pictures |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102760296B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104811584B (en) * | 2014-01-29 | 2018-03-27 | 晨星半导体股份有限公司 | Image-processing circuit and method |
| CN109741363B (en) * | 2019-01-11 | 2023-07-14 | 湖南国科微电子股份有限公司 | Motion judgment method and device based on block difference value and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1187988C (en) * | 1997-12-25 | 2005-02-02 | 三菱电机株式会社 | Motion compensating apparatus, moving image coding apparatus and method |
| US6985172B1 (en) * | 1995-12-01 | 2006-01-10 | Southwest Research Institute | Model-based incident detection system with motion classification |
| CN1806444A (en) * | 2004-05-10 | 2006-07-19 | 三星电子株式会社 | Adaptive-weighted motion estimation method and frame rate converting apparatus employing the method |
| CN101026761A (en) * | 2006-02-17 | 2007-08-29 | 中国科学院自动化研究所 | Motion estimation method of rapid variable-size-block matching with minimal error |
| TW201101841A (en) * | 2009-06-30 | 2011-01-01 | Pixel Technologies Inc U | Device and method for adaptive blending motion compensation interpolation in frame rate up-conversion |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7480334B2 (en) * | 2003-12-23 | 2009-01-20 | Genesis Microchip Inc. | Temporal motion vector filtering |
-
2011
- 2011-04-29 CN CN201110110598.6A patent/CN102760296B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6985172B1 (en) * | 1995-12-01 | 2006-01-10 | Southwest Research Institute | Model-based incident detection system with motion classification |
| CN1187988C (en) * | 1997-12-25 | 2005-02-02 | 三菱电机株式会社 | Motion compensating apparatus, moving image coding apparatus and method |
| CN1806444A (en) * | 2004-05-10 | 2006-07-19 | 三星电子株式会社 | Adaptive-weighted motion estimation method and frame rate converting apparatus employing the method |
| CN101026761A (en) * | 2006-02-17 | 2007-08-29 | 中国科学院自动化研究所 | Motion estimation method of rapid variable-size-block matching with minimal error |
| TW201101841A (en) * | 2009-06-30 | 2011-01-01 | Pixel Technologies Inc U | Device and method for adaptive blending motion compensation interpolation in frame rate up-conversion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102760296A (en) | 2012-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10997696B2 (en) | Image processing method, apparatus and device | |
| EP2881915B1 (en) | Techniques for disparity estimation using camera arrays for high dynamic range imaging | |
| TWI424377B (en) | Method for analyzing object motion in multi frames | |
| CA2969482C (en) | Method and apparatus for multiple technology depth map acquisition and fusion | |
| CN110692083B (en) | Block-matched optical flow and stereoscopic vision for dynamic vision sensor | |
| US20180096487A1 (en) | Systems and methods for fusing images | |
| US9558543B2 (en) | Image fusion method and image processing apparatus | |
| CN105453136B (en) | The three-dimensional system for rolling correction, method and apparatus are carried out using automatic focus feedback | |
| CN110717942B (en) | Image processing method and device, electronic equipment and computer readable storage medium | |
| CN109712192B (en) | Camera module calibration method and device, electronic equipment and computer readable storage medium | |
| CN101690236B (en) | Apparatus and method for estimating motion due to hand trembling and image pickup device using the same | |
| WO2015195317A1 (en) | Local adaptive histogram equalization | |
| CN109685853B (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium | |
| CN109559353B (en) | Camera module calibration method and device, electronic equipment and computer readable storage medium | |
| CN108053438B (en) | Depth of field acquisition method, device and equipment | |
| CN111371986A (en) | Apparatus and method for focus adjustment and depth map determination | |
| US10157439B2 (en) | Systems and methods for selecting an image transform | |
| US20190355101A1 (en) | Image refocusing | |
| JP6353233B2 (en) | Image processing apparatus, imaging apparatus, and image processing method | |
| CN102760296B (en) | Movement analyzing method for objects in multiple pictures | |
| US10893223B2 (en) | Systems and methods for rolling shutter compensation using iterative process | |
| CN111699511A (en) | Image processing method, device and storage medium | |
| CN105488845A (en) | Method for generating three-dimensional image and electronic device | |
| Alasal et al. | Improving passive 3D model reconstruction using image enhancement | |
| Mustaniemi et al. | Parallax correction via disparity estimation in a multi-aperture camera |
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 |