Background technology
Continuous development along with television technology, images such as 3D three-dimensional film, 3D stereoscopic TV are entering people's life gradually more and more, when appreciating the 3D video image, people are also improving the requirement of 3D video, should watch picture true to nature and abundant, can guarantee the comfort level of human eye again, can not cause vision overtired, so the damage eyes.
When watching the 3D video image, the seen picture of eyes is different behind the 3D glasses owing to wear, and needs eyes to focus on and by " fusion " function of brain two different images of eyes are synthesized the image of a solid.The image content of 3D film mostly is flight, rotation, switches, passes through the moving scene of fluctuating fast, and this has just caused eye muscle constantly to regulate, and is to adapt to the switching of close shot distant view, bigger for eye pressure.Therefore, be vision protection for the parallax adjustment of 3D image, improve the necessary link of 3D image comfort level.
The position difference of same object in the left and right cameras imaging surface formed parallax.For imaging plane, binocular parallax can be divided into negative parallax, positive parallax and parallax free, and as shown in Figure 1, Fig. 1 (a) is depicted as negative parallax, and promptly the object that perceives of eyes seemed before screen; Fig. 1 (b) is depicted as positive parallax, means that the object that eyes perceive seems after screen; Fig. 1 (c) is depicted as parallax free, refers to that the object that perceives seems to be positioned at same plane with screen.Convergent angle is meant the angle that the eyes sight line intersects, and common convergent angle is the easy more fatigue of oxeye more.
As shown in Figure 2, for one group of 3D rendering of presenting to eyes simultaneously, each pixel in this image is owing to have different parallaxes, and makes the observer produce depth perception and stereovision straggly to whole scene.The parallax set of all these pixel correspondences becomes optical parallax field or anaglyph.Minimum and maximum parallax in the optical parallax field has determined the disparity range of 3D stereo-picture, and disparity range is big more, and 3-D effect and stereovision are strong more, otherwise then weak more.
The stereoeffect that is produced by parallax requires human eye to carry out correspondingly focus adjustment according to the position of virtual scene, and this adjusting needs eye muscle group's cooperation, so the stereoeffect that human eye can comfortable observation is limited, is not strong more good more.Usually for static 3D stereo-picture, negative parallax has better 3-D effect, but because the eyes convergent angle causes eye fatigue more greatly and easily, therefore static 3-D view adopts positive parallax more.3 D video for motion, owing to also there is motion when having 3-D effect, focusing burden to human eye is bigger, more be easy to generate sense of fatigue, more obvious when especially having the image grand movement that causes by camera motion, thereby need be controlled to reduce this discomfort the optical parallax field that produces 3-D effect according to movable information.
About the parallax adjustment of 3D image, existing a lot of exploration of forefathers.Wherein more representative by Mel Siegel and Shojiro Nagata proposition " little stereopsis (Microstereopsis) " theory, can be when the analyzing three-dimensional scenery based on people's brain in conjunction with priori, the thought of " less 3D effect promptly enough (Just Enough Reality) " is proposed, be that the 3D image only need provide 3D effect slightly, people's brain just can be experienced out enough three-dimensional informations according to priori.The 3D video camera of Sony Corporation just is based on this principle manufacturing and is used for the shooting of film " A Fanda "." little stereopsis (Microstereopsis) " theoretical problem is that it is that cost guarantees the eyes comfortableness of spectators when watching the 3D program with substantial sacrifice 3D effect.
Another kind of representative way is: regulate parallax and make all virtual scenes all be imaged on after the video screen, promptly be in the zone far away than video screen, i.e. " positive parallax " with respect to eyes.With it accordingly, be imaged on video screen situation before and be " negative parallax ", just being imaged on the screen is " parallax free ".Generally, spectators feel comfortably cool owing to two convergent angle is less when watching " positive parallax ", and two the easy more greatly and more ocular vision fatigue that causes of convergent angle when observing " negative parallax ".The method that parallax is controlled to be " positive parallax " can increase the comfort level of human eye, but does not improve for the situation of the ophthalmic uncomfortable that is caused by motion in the 3D video.
Summary of the invention
The invention provides a kind of parallax adjusting method and device of 3D video image, can regulate disparity range, avoid visual fatigue based on the movable information in the video image picture.
The parallax adjusting method of a kind of 3D video image that the embodiment of the invention provides comprises:
Obtain the 3D video signal and calculate optical parallax field, statistics is determined described 3D video image parallax difference scope;
Based on the kinematic parameter in the different picture frames calculating videos constantly in the 3D video image that is obtained;
Calculate motion amplitude according to kinematic parameter, and the disparity range of described 3D video image parallax difference field is adjusted according to this motion amplitude;
Generate new 3D vision signal with reference to the disparity range after the described adjusting.
The parallax adjustment device of a kind of 3D video image that the embodiment of the invention provides comprises:
The parallax information acquiring unit is used to obtain the 3D video signal and calculates optical parallax field, and statistics is determined described 3D video image parallax difference scope;
Computing unit is based on the kinematic parameter in the different picture frames calculating videos constantly in the 3D video image that is obtained;
Adjustment unit calculates motion amplitude according to kinematic parameter, and according to this motion amplitude the disparity range of described 3D video image parallax difference field is adjusted;
Generation unit generates new 3D vision signal with reference to the disparity range after the described adjusting.
In sum, the present invention proposes a kind of 3D video image parallax difference control method based on movable information, can regulate disparity range based on the movable information in the video.By the 3D video image is carried out motion analysis, and calculate the optical parallax field of 3D video, the disparity range of 3D vision signal is regulated based on the movable information that motion analysis obtains.According to the present invention, motion picture appears in video image, during especially violent picture, can correspondingly compress disparity range, obtain the disparity range of the most comfortable, improve the purpose that the 3D video image is viewed and admired comfort level.Avoid visual fatigue, thus the protection human eye.The stereo display mode of the fixedly parallax that the present invention can break traditions according to user's request, provides different parallax adjustment patterns, can under the prerequisite of vision protection health, make the user enjoy 3D picture effect true to nature and lively improving users'comfort.
Embodiment
The intrinsic physiological property of human eye is: need to focus on when object observing.When watching the conventional two-dimensional video, human eye is kept all the time and is focused on display plane and get final product, and does not need significantly to adjust, thereby can not cause visual fatigue; When watching the 3D stereo-picture, because the stereoeffect that optical parallax field produces makes virtual scene be presented on the different degree of depth, need eyes to focus on adjustment accordingly, this being adjusted at when watching the three-dimensional motion video can become very frequent along with the variation of scenery, therefore is very easy to cause visual fatigue.
The motion of video can be divided into global motion and part (target) motion two classes according to the shared picture size of motion parts.Global motion is mainly caused by camera motion, and all scenery in the scene all move at this moment; Local motion is mainly caused by the target travel in the scene, and have only the specific objective in the scene to move this moment.When producing motion in the scene, no matter be local motion or global motion, all be accompanied by the variation of scenery, the stereoeffect of these scenery in the 3D video also changes thereupon.When watching the 3D sport video, people's eyes are had to catch simultaneously movable information and stereoscopic parallax information and are focused on again, when human eye is subjected to the physiology restriction and can't catches up with motion amplitude, produce dizzy sense thereby will feel not see picture; Even human eye can be caught up with motion amplitude, to watch for a long time in the process, the also inevitable adjusting that is forced to owing to eyes too frequently causes eye muscle fatigue.
This shows that when watching 3 D video, need the intensity of stereoeffect be controlled according to movable information: movement degree is big more, stereoeffect will weaken relatively, could reduce the burden of eyes; And when scenery is static relatively, then in the comfortable scope of human eye, maximize stereoeffect.So, just can make the physiological property that meets human eye more that presents of stereoeffect, on the basis of protection eye health, allow the onlooker realize best visual effect.
In view of the deficiency that exists in the existing 3D video technique, the present invention proposes a kind of 3 D video optical parallax field control method based on movable information, can regulate disparity range based on the movable information in the video.
For above-mentioned two class video motions, global motion is owing to the most of scenery in the scene all moves, and is bigger for the burden of human eye; Local motion is because range of movement is less, and is also less relatively for the burden of human eye, but because the focusing center of human eye can be transferred on this moving target, therefore also need control parallax at the movable information of motion target area.Therefore, no matter be global motion or local motion, control method is similar substantially, the image range difference of being considered when both differences are just regulated: movable information and the parallax information that need consider entire image based on the adjusting of global motion; And only need consider the movable information and the parallax information of moving region based on the adjusting of local (target) motion.
Figure 3 shows that the present invention proposes the principle schematic of a kind of 3D visual parallax field control, the 3D video image is carried out motion analysis, and calculate the optical parallax field of 3D video, the disparity range of 3D vision signal is regulated based on the movable information that motion analysis obtains.When occurring the violent picture of motion in the video image, can correspondingly compress disparity range reducing stereoeffect, thereby the protection human eye is avoided visual fatigue.
With reference to Fig. 4, the parallax adjusting method of a kind of 3D video image that the embodiment of the invention provides comprises:
S101 obtains the 3D video signal and calculates optical parallax field, and statistics is determined described 3D video image parallax difference scope;
A pair of 3D rendering (being left-eye image, eye image) according to current time calculates optical parallax field, and definite disparity range.
Selected for use the method that optical parallax field calculates is a lot, and most typical method is: adopt the method for images match, at each the pixel P in the left-eye image
l, in eye image, seek corresponding match point P
r, P
lWith P
rBetween horizontal displacement D=P
r-P
l, be parallax at pixel Pl; All pixels among the left figure are calculated parallax obtain optical parallax field;
Statistics obtains the maximum D of parallax in the optical parallax field
MaxWith minimum value D
Min, be disparity range [D
Min, D
Max].
S102 is based on the kinematic parameter in the different picture frames calculating videos constantly in the 3D video image that is obtained;
Motion analysis: according to the movable information in the difference picture frame calculating video constantly.
Method of motion analysis can adopt existing several different methods, for example, based on specific motion model (as 3 parameters, 4 parameters, 6 parameters, 8 parameters etc.), according to a left side (or right) eye pattern picture current and previous moment, utilize the method searching matched pixel point of images match right, utilize the method for adding up to determine the kinematic parameter of motion model then.
Calculate the kinematic parameter in the video, specifically comprise:
Select to determine corresponding kinematic parameter based on predetermined motion model;
According to current and left-eye image previous moment, utilize the method for images match to seek matched pixel point in the eye image, utilize statistics to obtain the kinematic parameter of described 3D video image then, usually, kinematic parameter comprises: translation amplitude, angle, speed etc.
S103 calculates motion amplitude according to kinematic parameter, and according to this motion amplitude the disparity range of described 3D video image parallax difference field is adjusted;
According to motion amplitude the disparity range of optical parallax field is adjusted, concrete steps are:
Selected parallax adjustment function: the parallax adjustment function can be selected arbitrarily as required, but need follow following requirement:
Along with motion amplitude increases, the disparity range of permission is progressively dwindled;
Because the easier visual fatigue that causes of negative parallax, the allowed band of negative parallax is less usually.
Need to prove that when calculating motion amplitude M according to kinematic parameter, computational methods depend on the selected model of estimation.Model difference, kinematic parameter meaning difference, motion amplitude computational methods are also just different;
In that being handled, same section video should select a unique model, to guarantee the consistency of treatment effect;
For the different motion parameter different susceptibilitys is arranged when human eye focuses on, when calculating motion amplitude M, need to regulate by applying weight.
Particularly, in the specific embodiment of the invention, motion amplitude is big more, usually corresponding disparity range is narrowed down in the more little scope.
According to decreasing function relationship delta d=f (M) disparity range of described 3D video image parallax difference field is adjusted in the present embodiment, for example, the curve of function Δ d=f (M) is as Fig. 5 (a) and (b), and wherein Δ d is a parallax value, and M is a motion amplitude value in the 3D video image.
In the embodiment of the invention, calculate the motion amplitude in the video image, be specially:
Adopt three parameter model, calculate motion amplitude as follows:
A wherein
1The amplitude of representative image convergent-divergent, a
2, a
3Representative is the translation amplitude of two horizontal directions anyhow, λ
1With z be weight factor at image zoom, λ
2And λ
3It is weight factor at image translation.
In the embodiment of the invention, at first according to the allowed band of motion amplitude and parallax adjustment function calculation parallax.Here select the inverse ratio adjustment function shown in Fig. 5 (a) for use, the curve that calculates the parallax allowed band that obtains is shown in horizontal stripe zone among Fig. 8.Fig. 6 is the disparity range curve chart of 3D video image in the embodiment of the invention, and the motion amplitude curve stack by with the curve among Fig. 6 and Fig. 7 obtains parallax adjustment scope shown in Figure 8.The vertical fringe area of Fig. 8 is the part that exceeds allowed band, need carry out parallax adjustment.Because the horizontal range decision of the size of parallax two video cameras when taking can be adopted virtual visual angle technology, generates the less virtual visual angle 3D rendering of spacing and reaches the purpose that reduces parallax.
At last, the crossbar part as shown in Figure 9 of the parallax after the adjusting has realized the variation control disparity range according to motion amplitude like this, thereby reduces the purpose of human eye burden.
S104 generates new 3D vision signal based on the disparity range after regulating;
Generate new 3D vision signal based on the disparity range after regulating, and output to display unit.
With reference to Figure 10, the parallax adjustment device 200 of a kind of 3D video image that the embodiment of the invention provides comprises:
Parallax information acquiring unit 210 is used to obtain the 3D video signal and calculates optical parallax field, and statistics is determined described 3D video image parallax difference scope;
Computing unit 220 is based on the kinematic parameter in the different picture frames calculating videos constantly in the 3D video image that is obtained;
Adjustment unit 230 calculates motion amplitude according to kinematic parameter, and according to this motion amplitude the disparity range of this 3D video image parallax difference field is adjusted;
Generation unit 240 generates new 3D vision signal with reference to the disparity range after the described adjusting.
As preferably, parallax adjustment device 200 in the embodiment of the invention also comprises:
Memory cell 250 is used to preserve different parallax value scope relevant parameters, uses for the subsequent adjustment parallax.
Computing unit 220 specifically comprises:
Parameter is selected module 220a, selects to determine corresponding kinematic parameter based on predetermined motion model;
Counting statistics module 220b according to current and left-eye image previous moment, utilizes the method for images match to seek matched pixel point in the eye image, utilizes statistics to obtain the kinematic parameter of 3D video image then.
Adjustment unit 230 is adjusted the disparity range of described 3D video image parallax difference field according to decreasing function relationship delta d=f (M), and wherein Δ d is a parallax value, and M is a motion amplitude value in the 3D video image.
In sum, the present invention proposes a kind of 3D video image parallax difference control method based on movable information, can regulate disparity range based on the movable information in the video.By the 3D video image is carried out motion analysis, and calculate the optical parallax field of 3D video, the disparity range of 3D vision signal is regulated based on the movable information that motion analysis obtains.According to the present invention, motion picture appears in video image, during especially violent picture, can correspondingly compress disparity range, obtain the disparity range of the most comfortable, to reach vision protection, to improve the purpose that the 3D video image is viewed and admired comfort level.Thereby the protection human eye is avoided visual fatigue.The stereo display mode of the fixedly parallax that the present invention can break traditions according to user's request, provides different parallax adjustment patterns, can under the prerequisite of vision protection health, make the user enjoy best 3D effect improving users'comfort.
Though the present invention with preferred embodiment openly as above; but it is not to be used for limiting the present invention; any those skilled in the art without departing from the spirit and scope of the present invention; can make possible change and modification, so protection scope of the present invention should be as the criterion with the scope that claim of the present invention was defined.