CN102238313A - Method for generating image transformation matrix as well as image transformation method and device - Google Patents

Abstract

The invention provides a method for generating an image transformation matrix as well as an image transformation method and device. The image transformation device can generate one three-dimensional rotation matrix in accordance with a three-dimensional rotation angle; the three-dimensional rotation matrix is projected to a two-dimensional plane to generate a two-dimensional rotation matrix; and according to the two-dimensional rotation matrix and an affine transformation matrix, a target position in a target image is calculated to generate a source position corresponding to the target position, wherein the source position is located in a source image, and the affine transformation matrix is used for carrying out displacement, zoom and rigid body deformation on the target image. The source position and a source pixel nearest to the source position are utilized to generate pixel data. The pixel data are received to be taken as a target pixel value of the target position so as to generate the target image.

Description

Produce method, image conversion method and the device thereof of video conversion matrix

Technical field

The invention relates to a kind of video conversion technology, and particularly relevant for a kind of instant conversion image, and make the image of conversion have the bidimensional image switch technology of approximate stereoeffect.

Background technology

Electronic product uses the diagram in user's interface often or shows two dimension (two-dimensional, 2D) images such as photo, film.The portions of electronics product can be rotated image, moves, three-dimensional (three-dimensional, 3D) conversion allow the user appreciate and more convenient when browsing image, or increase novel sense for convergent-divergent etc.

Existing 3-dimensional image switch technology mainly is to utilize business software to handle image, as open graphic package storehouse (Open Graphics Library, OpenGL) etc., it can allow image have 3-dimensional image effects such as the depth of field, rigid body rotation (solid body rotation), three-dimensional convergent-divergent.But when handling image, must additionally increase processor, memory, buffer or the like many elements, thereby increase cost by the complex software in open graphic package storehouse and so on.In addition, it is longer to utilize software to handle the required time of image, thereby is difficult to instantaneity ground conversion image.

The mode of another kind of conversion image is to handle bidimensional image by hardware unit.Though present image processor is simple relatively, cost is lower, only image can be done image processing such as two-dimensional plane rotation, convergent-divergent, does not comprise the function of 3-dimensional image conversion.

Summary of the invention

The invention provides a kind of image conversion method, the video conversion matrix and its inverse matrix that are calculated by the three-dimensional rotation matrix that is projected to two dimensional surface, just can use the hardware structure of two-dimensional transformations, allow the image after changing have the effect of approximate three-dimensional convergent-divergent, distortion and rotation.

With another viewpoint, the invention provides a kind of video conversion device, the video conversion matrix and its inverse matrix that are calculated by the three-dimensional rotation matrix that is projected to two dimensional surface, just can use the hardware structure of two-dimensional transformations, allow the image after changing have the effect of approximate three-dimensional convergent-divergent, distortion, displacement and rotation.

The present invention proposes a kind of image conversion method, and it comprises the following steps.At first, utilize will originate a plurality of end points of image of video conversion matrix to come the source position to be converted to a plurality of end points target location of target image, to define the zone of source video conversion to target image.According to the inverse matrix of video conversion matrix, the source position of coming that produces corresponding target location is calculated in the target location in the target image afterwards, and above-mentionedly come the source position to be arranged in the source image.Then, in the basis source image come source position and the most approaching predetermined number that comes the source position come source pixel to produce pixel data.And, with the target pixel value of above-mentioned pixel data, to produce target image as the target location.Wherein, above-mentioned target image is shown in the display unit.

In one embodiment of this invention, the step that produces this pixel data comprises utilizing above-mentionedly comes source pixel to make interpolation to calculate, to produce pixel data.

In one embodiment of this invention, above-mentioned predetermined number is one, and the step that produces pixel data comprises: directly use and the most approachingly come the pixel value that comes source pixel of source position as pixel data.

In one embodiment of this invention, above-mentioned video conversion matrix is tried to achieve according to two-dimentional spin matrix and affine transition matrix, and two-dimentional spin matrix is the three-dimensional rotation matrix that the three-dimensional rotation angle is formed to be projected to two dimensional surface produce.

In one embodiment of this invention, the above-mentioned video conversion matrix product that is two-dimentional spin matrix and affine transition matrix.

Also comprise the following steps: in one embodiment of this invention target image is covered on the background video, to produce composite image, wherein composite image is shown in the display unit.

In one embodiment of this invention, the target location in the target image is calculated the source position of coming that produces corresponding target location and also comprise the following steps: at first check the source position whether to be positioned at the source image.When coming the source position not to be positioned at the source during image, the pixel data of corresponding target location is set at default background pixel value.When coming the source position to be positioned at the source during image, subsequent steps then, the source pixel of coming of coming source position and the most approaching predetermined number that comes the source position in the basis source image is used the generation pixel data.

In another angle, the present invention proposes a kind of video conversion device, and it comprises video conversion unit and output unit.The video conversion unit produces the three-dimensional rotation matrix according to the three-dimensional rotation angle, and with the projection of three-dimensional rotation matrix in two dimensional surface to produce two-dimentional spin matrix, and according to two-dimentional spin matrix and affine transition matrix, utilize the target location in the target image to calculate the source position of coming that produces corresponding target location, wherein, come the source position to be arranged in the source image.And the source pixel of coming of coming source position and the most approaching predetermined number that comes the source position in the basis source image is used the generation pixel data.Output unit receives pixel data with the target pixel value as the target location, uses the generation target image, and wherein target image is shown in the display unit.

In one embodiment of this invention, above-mentioned video conversion unit multiply by two-dimentional spin matrix and affine transition matrix to calculate mutually tries to achieve the video conversion matrix, then calculate again and try to achieve the inverse matrix of video conversion matrix, and utilize inverse matrix and target location to calculate to produce the source position.

In one embodiment of this invention, the video conversion device also comprises the edge calculations unit, it utilizes will originate a plurality of end points of image of video conversion matrix to come the source position to be converted to a plurality of end points target location of target image, to define the zone of source video conversion to target image.

In one embodiment of this invention, the video conversion device also comprises the image mixed cell, and it is covered in target image on the background video, and to produce composite image, wherein, above-mentioned composite image is shown in the display unit.

With another viewpoint, the present invention proposes a kind of method that produces the video conversion matrix, and it comprises the following steps: at first, utilizes the three-dimensional rotation angle to produce the three-dimensional rotation matrix.Then, with the projection of three-dimensional rotation matrix in two dimensional surface, to produce two-dimentional spin matrix.Afterwards, try to achieve the video conversion matrix according to two-dimentional spin matrix and affine transform matrix calculations.And, calculate the inverse matrix of trying to achieve the video conversion matrix, wherein, when obtaining the source image, just can produce target image, so that target image is shown in the display unit according to inverse matrix and source image.And the target pixel value of target location can utilize and come the pixel value of source position to produce in the corresponding target location in the target image, thereby uses the generation target image, and comes the source position to be arranged in the source image.

In one embodiment of this invention, image conversion method also comprises the following steps: to come the source position to be converted to a plurality of end points target location of target image by will originate a plurality of end points of image of video conversion matrix, to define the zone of source video conversion to target image.

Based on above-mentioned, embodiment of the invention utilization is projected to video conversion matrix and its inverse matrix that the three-dimensional rotation matrix and the affine matrix of two dimensional surface are calculated, can calculate the source by the target location in the target image and come the source position producing corresponding pixel value in the image, but make video conversion not need buffer and the instant playback image.Meanwhile, also can be in the converting structure of two dimension, allow target image after the conversion have the effect of approximate perspective transformations.

Description of drawings

Fig. 1 is the calcspar according to a kind of video conversion device of first embodiment of the invention;

Fig. 2 is according to the source video conversion of the first embodiment of the invention schematic diagram to target image;

Fig. 3 is the flow chart according to a kind of image conversion method of first embodiment of the invention;

Fig. 4 is the calcspar according to a kind of video conversion device of second embodiment of the invention;

Fig. 5 is the schematic diagram according to the composite image method of second embodiment of the invention;

Fig. 6 is the flow chart according to a kind of image conversion method of third embodiment of the invention.

Drawing reference numeral

100,400: video conversion device BI: background video

110: video conversion cells D I: target image

120: output unit DP: target location

130: edge calculations unit MI: composite image

150: display unit SI: the source image

460: image mixed cell SP: come the source position

BA: borderline region TM: video conversion matrix

E1, E2, E3, E4: four end points of source image come the source position

E1 ', E2 ', E3 ', ' E4 ': four end points of the target image of corresponding source image come the source position

TM ^-1: the inverse matrix of video conversion matrix

RA: the hatched example areas between borderline region and the target image

S310～S390: the step of image conversion method

S610～S640: the step of image conversion method

Embodiment

For above-mentioned feature and advantage of the present invention can be become apparent, embodiment cited below particularly, and cooperate appended graphic being described in detail below.

Please refer to Fig. 1 and Fig. 2, Fig. 1 is the calcspar according to a kind of video conversion device 100 of first embodiment of the invention, and Fig. 2 is the schematic diagram that is converted to target image DI according to the source image SI of first embodiment of the invention.Video conversion device 100 comprises video conversion unit 110 and output unit 120.Video conversion unit 110 produces the three-dimensional rotation matrix according to the three-dimensional rotation angle, with the projection of three-dimensional rotation matrix in two dimensional surface to produce two-dimentional spin matrix.And video conversion unit 110 is according to two-dimentional spin matrix and affine transition matrixes, by the target position D P in the target image DI with calculate produce corresponding target position D P come source position SP, wherein come source position SP to be arranged in the image SI that originates.In addition, video conversion unit 110 basis source position SP and the most approaching source position SP's of coming comes source pixel to produce pixel data.Output unit 120 receives the target pixel value of pixel data as target position D P, and with generation target image DI, and target image DI can be shown in the display unit 150.

In present embodiment, video conversion device 100 also comprises edge calculations (bound-boxingcalculation) unit 130, it utilizes will the originate end points of image SI of above-mentioned video conversion matrix to come the source position to be converted to the end points target location of corresponding target image DI, to define the transition region that source image SI is converted to target image DI.

Usually know that for this area is had the knowledgeable can understand the present invention more, the operation workflow and the principle of the image conversion method of the embodiment of the invention are described at this, please be simultaneously with reference to Fig. 1, Fig. 2 and Fig. 3.Fig. 3 is the flow chart according to the image conversion method of first embodiment of the invention.Wherein, video conversion device 100 is in order to the image conversion method of execution graph 3.The target of present embodiment is to utilize video conversion matrix T M and its inverse matrix TM ^-1Target image DI will be converted to with will originating image SI instantaneity.Wherein, video conversion matrix T M is by the three-dimensional rotation matrix T _3DBe projected to the two-dimentional spin matrix T of two dimensional surface _2DAnd the product of affine transition matrix AT, therefore allow the image of conversion can have approximate three-dimensional effect.And owing to come source position SP can utilize video conversion matrix T M to be converted on the target position D P, target position D P just can be by the inverse matrix TM of video conversion matrix T M ^-1Try to achieve source position SP with backwards calculation.Thus, video conversion device 100 does not just need additionally to increase buffer and keeps in image, and can directly utilize the inverse matrix TM of video conversion matrix T M ^-1Calculate the pixel value that each is positioned at the target position D P of target image DI, with instant generation target image DI.

Above-mentioned three-dimensional rotation matrix T _3DFormed by the three-dimensional rotation angle.Affine transition matrix AT then comprises source image SI and is converted to the convergent-divergent of target image DI, all multi-parameters that rigid body is out of shape and move the position, and these parameters and the anglec of rotation can be set according to user's demand.For example, video conversion unit 110 can according to the user the given parameter and the anglec of rotation produce video conversion matrix T M, also can produce video conversion matrix T M by the parameter and the anglecs of rotation default in the video conversion unit 110.

Describe the flow process of image conversion method in detail at this.In step S310, the video conversion unit 110 in the video conversion device 100 at first according to the three-dimensional rotation angle to produce the three-dimensional rotation matrix T _3DThe three-dimensional rotation angle comprises X-axis anglec of rotation α, Y-axis anglec of rotation β and Z axle anglec of rotation θ in present embodiment.The three-dimensional rotation matrix T _3DShown in equation (1) and (2):

$T_{3D}={\mathrm{RM}}_Z\×{\mathrm{RM}}_Y\×{\mathrm{RM}}_X=\left(\begin{array}{ccc}\cos \underset{\‾}{\mathrm{\θ}}& -\sin \underset{\‾}{\mathrm{\θ}}& 0\\ \sin \underset{\‾}{\mathrm{\θ}}& \cos \underset{\‾}{\mathrm{\θ}}& 0\\ 0& 0& 1\end{array}\right)\×\left(\begin{array}{ccc}\cos \underset{\‾}{\mathrm{\β}}& 0& \sin \underset{\‾}{\mathrm{\β}}\\ 0& 1& 0\\ -\sin \underset{\‾}{\mathrm{\β}}& 0& \cos \underset{\‾}{\mathrm{\β}}\end{array}\right)\×\left(\begin{array}{ccc}1& 0& 0\\ 0& \cos \underset{\‾}{\mathrm{\α}}& -\sin \underset{\‾}{\mathrm{\α}}\\ 0& \sin \underset{\‾}{\mathrm{\α}}& \cos \underset{\‾}{\mathrm{\α}}\end{array}\right)..\left(1\right)$

$=\left(\begin{array}{ccc}\left(\cos \underset{\‾}{\mathrm{\θ}}\right)\left(\cos \underset{\‾}{\mathrm{\β}}\right)& -\left(\sin \underset{\‾}{\mathrm{\θ}}\right)\left(\cos \underset{\‾}{\mathrm{\α}}\right)+\left(\cos \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\β}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)& \left(\sin \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)-\left(\cos \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\β}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)\\ \left(\sin \underset{\‾}{\mathrm{\θ}}\right)\left(\cos \underset{\‾}{\mathrm{\β}}\right)& \left(\cos \underset{\‾}{\mathrm{\θ}}\right)\left(\cos \underset{\‾}{\mathrm{\α}}\right)+\left(\sin \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\β}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)& -\left(\cos \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)+\left(\sin \underset{\‾}{\mathrm{\θ}}\right)\left(\sin \underset{\‾}{\mathrm{\β}}\right)\left(\cos \underset{\‾}{\mathrm{\α}}\right)\\ -\left(\sin \underset{\‾}{\mathrm{\β}}\right)& \left(\cos \underset{\‾}{\mathrm{\β}}\right)\left(\sin \underset{\‾}{\mathrm{\α}}\right)& \left(\cos \underset{\‾}{\mathrm{\θ}}\right)\end{array}\right)..\left(2\right)$

Can find out the three-dimensional rotation matrix T by equation (1) _3DIn present embodiment Z axis rotation matrix RM _Z, Y-axis spin matrix RM _YWith X-axis spin matrix RM _XProduct.X-axis spin matrix RM _XExpression points to the source vector V (not illustrating) that comes of source position SP to rotate X-axis anglec of rotation α, Y-axis spin matrix RM for X-axis from the initial point of rectangular coordinate system _YThen expression comes source vector V for Y-axis rotation Y-axis anglec of rotation β, and Z axis rotation matrix RM _ZThen expression comes source vector V for Z axle rotation Z axle anglec of rotation θ.Equation (2) is the Z axis rotation matrix RM with equation (1) _Z, Y-axis spin matrix RM _YWith X-axis spin matrix RM _XThe three-dimensional rotation matrix T that multiplies each other and launched _3D

Then enter step S320, video conversion unit 110 is with the three-dimensional rotation matrix T _3DBe projected to two dimensional surface, to produce two-dimentional spin matrix T _2D, so just can use two-dimentional framework to reach approximate three-dimensional conversion effect.In present embodiment, the three-dimensional rotation matrix T _3DThe step that is projected to two dimensional surface is with the three-dimensional rotation matrix T _3DFirst row the 3rd row, second row the 3rd row, the third line first row be made as " 0 " with the operator of the third line secondary series, and with the three-dimensional rotation matrix T _3DThe tertial operator of the third line be made as " 1 ", in the hope of two-dimentional spin matrix T _2DTwo dimension spin matrix T _2DShown in equation (3):

$T_{2D}=\left(\begin{array}{ccc}(\cos \_\mathrm{\θ})(\cos \_\mathrm{\α})& -(\sin \_\mathrm{\θ})(\cos \_\mathrm{\β})+(\cos \_\mathrm{\θ})(\sin \_\mathrm{\α})(\sin \_\mathrm{\β})& 0\\ (\sin \_\mathrm{\θ})(\cos \_\mathrm{\α})& (\cos \_\mathrm{\θ})(\cos \_\mathrm{\β})+(\sin \_\mathrm{\θ})(\sin \_\mathrm{\α})(\sin \_\mathrm{\β})& 0\\ 0& 0& 1\end{array}\right)...\left(3\right)$

Afterwards, video conversion unit 110 two-dimentional spin matrix T of foundation in step S340, S350 and S370 _2DWith affine transition matrix AT, come source position SP with what calculate to produce corresponding target position D P by the target position D P in the target image DI.Affine transition matrix AT comprises that in present embodiment X-axis zooming parameter SX, Y-axis zooming parameter SY, X-axis rigid body deformation parameter SHX, Y-axis rigid body deformation parameter SHY, X-axis moving parameter TX and y-axis shift move parameter TY, and it can be out of shape blurring, convergent-divergent and rigid body by the parameter that affine transition matrix AT provides.By this, just affine transition matrix AT shown in equation (4):

$\mathrm{AT}=\left(\begin{array}{ccc}\mathrm{SX}& \mathrm{SHX}& \mathrm{TX}\\ \mathrm{SHY}& \mathrm{SY}& \mathrm{TY}\\ 0& 0& 1\end{array}\right)...\left(4\right)$

In present embodiment, at first video conversion unit 110 calculates at step S340 and tries to achieve video conversion matrix T M.Wherein, video conversion matrix T M is by the three-dimensional rotation matrix T _3DThe two-dimentional spin matrix T of projection _2DAnd the product of affine transition matrix AT, shown in equation (5):

$\mathrm{TM}=T_{2D}\×\mathrm{AT}=$

$\left(\begin{array}{ccc}(\cos \_\mathrm{\θ})(\cos \_\mathrm{\α})& -(\sin \_\mathrm{\θ})(\cos \_\mathrm{\β})+(\cos \_\mathrm{\θ})(\sin \_\mathrm{\α})(\sin \_\mathrm{\β})& 0\\ (\sin \_\mathrm{\θ})(\cos \_\mathrm{\α})& (\cos \_\mathrm{\θ})(\cos \_\mathrm{\β})+(\sin \_\mathrm{\θ})(\sin \_\mathrm{\α})(\sin \_\mathrm{\β})& 0\\ 0& 0& 1\end{array}\right)\×\left(\begin{array}{ccc}\mathrm{SX}& \mathrm{SHX}& \mathrm{TX}\\ \mathrm{SHY}& \mathrm{SY}& \mathrm{TY}\\ 0& 0& 1\end{array}\right)$

$=\left(\begin{array}{ccc}A& B& C\\ D& E& F\\ 0& 0& 1\end{array}\right)...\left(5\right)$

Because video conversion matrix T M is via two-dimentional spin matrix T _2DOperator after multiplying each other with affine transition matrix AT is too numerous and diverse, represents with A to F with operator of second row at this first row with video conversion matrix T M.The operator of video conversion matrix T M the third line can calculate simply to try to achieve and be (0,0,1).

Therefore, just come source position SP to convert target position D P to by video conversion matrix T M, shown in equation (6):

$\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)=T_{2D}\×\mathrm{AT}\×\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)=\mathrm{TM}\×\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)...\left(6\right)$

(SP wherein _X, SP _Y) be the coordinate that comes source position SP, and (DP _X, DP _Y) be the coordinate of target position D P.

In order to calculate the pixel value of trying to achieve each target position D P, above-mentioned equation (6) can be with the inverse matrix TM that takes advantage of video conversion matrix T M in about equal sign ^-1Try to achieve following equation (7) and derive, to obtain the coordinate of source position SP by target position D P.Its derivation is as follows:

$\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)=\mathrm{TM}\×\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)...\left(6\right)$

$\⇒{\mathrm{TM}}^{-1}\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)={\mathrm{TM}}^{-1}\mathrm{TM}\×\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)=\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)$

$\⇒\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)={\mathrm{TM}}^{-1}\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)...\left(7\right)$

Therefore when program S350, video conversion unit 110 convenient calculating with video conversion matrix T M are tried to achieve its inverse matrix TM ^-1, shown in equation (8):

${\mathrm{TM}}^{-1}={\left(\begin{array}{ccc}A& B& C\\ D& E& F\\ 0& 0& 1\end{array}\right)}^{-1}=\frac{1}{\mathrm{AE}-\mathrm{BD}}\left(\begin{array}{ccc}E& -B& \mathrm{BF}-\mathrm{CE}\\ -D& A& -\mathrm{AF}+\mathrm{CD}\\ 0& 0& \mathrm{AE}-\mathrm{BD}\end{array}\right)$

$=\left(\begin{array}{ccc}\frac{E}{\mathrm{AE}-\mathrm{BD}}& \frac{-B}{\mathrm{AE}-\mathrm{BD}}& \frac{\mathrm{BF}-\mathrm{CE}}{\mathrm{AE}-\mathrm{BD}}\\ \frac{-D}{\mathrm{AE}-\mathrm{BD}}& \frac{A}{\mathrm{AE}-\mathrm{BD}}& \frac{-\mathrm{AF}+\mathrm{CD}}{\mathrm{AE}-\mathrm{BD}}\\ 0& 0& \frac{\mathrm{AE}-\mathrm{BD}}{\mathrm{AE}-\mathrm{BD}}\end{array}\right)...\left(8\right)$

By combining of equation (7) and (8), just can pass through target position D P and inverse matrix TM ^-1Calculating tries to achieve source position SP, shown in equation (9):

$\left(\begin{array}{c}{\mathrm{SP}}_X\\ {\mathrm{SP}}_Y\\ 1\end{array}\right)={\mathrm{TM}}^{-1}\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)=\left(\begin{array}{ccc}\frac{E}{\mathrm{AE}-\mathrm{BD}}& \frac{-B}{\mathrm{AE}-\mathrm{BD}}& \frac{\mathrm{BF}-\mathrm{CE}}{\mathrm{AE}-\mathrm{BD}}\\ \frac{-D}{\mathrm{AE}-\mathrm{BD}}& \frac{A}{\mathrm{AE}-\mathrm{BD}}& \frac{-\mathrm{AF}+\mathrm{CD}}{\mathrm{AE}-\mathrm{BD}}\\ 0& 0& \frac{\mathrm{AE}-\mathrm{BD}}{\mathrm{AE}-\mathrm{BD}}\end{array}\right)\left(\begin{array}{c}{\mathrm{DP}}_X\\ {\mathrm{DP}}_Y\\ 1\end{array}\right)...\left(9\right)$

Equation (9) can also convert the simultaneous equations shown in equation (10) to:

$\left\{\begin{array}{c}{\mathrm{SP}}_X=\left(\frac{E}{\mathrm{AE}-\mathrm{BD}}\right)\×{\mathrm{DP}}_X+\left(\frac{-B}{\mathrm{AE}-\mathrm{BD}}\right)\×{\mathrm{DP}}_Y+\left(\frac{\mathrm{BF}-\mathrm{CE}}{\mathrm{AE}-\mathrm{BD}}\right)\\ {\mathrm{SP}}_Y=\left(\frac{-D}{\mathrm{AE}-\mathrm{BD}}\right)\×{\mathrm{DP}}_X+\left(\frac{A}{\mathrm{AE}-\mathrm{BD}}\right)\×{\mathrm{DP}}_Y+\left(\frac{-\mathrm{AF}+\mathrm{CD}}{\mathrm{AE}-\mathrm{BD}}\right)\end{array}\right....\left(10\right)$

At this moment, just need to define the zone that source image SI is converted to target image DI, to obtain target position D P one by one.Therefore in step S360, edge calculations unit 130 comes source position E1, E2, E3 and E4 to be converted to four end points target location E1 ', E2 ', the E3 ' and E4 ' of corresponding target image DI respectively by will originate four end points of image SI of video conversion matrix T M, is converted to the borderline region BA of target image DI to define source image SI.Wherein, borderline region BA is the minimum rectangle that comprises target image DI, and four end points target location E1 ' of target image DI, E2 ', E3 ' all are positioned on the square boundary of borderline region BA with E4 '.So just do not need to calculate the pixel value of whole image, only need the central pixel value of computation bound area B A, can improve video conversion efficient.

In addition, because present embodiment is an example with the source image SI of rectangle, therefore the borderline region BA among the step S360 also is a rectangle, should come to produce corresponding borderline region BA according to the source image SI of other shapes by visual its design requirement but use embodiment of the invention person, the present invention should be as limit.In some application examples, the borderline region BA among the step S360 also can directly be considered as the quadrangle (being target image DI zone) that E1 ', E2 ', E3 ' and E4 ' are surrounded; If or source image SI is N limit shape (N is integer and N 〉=3), then edge calculations unit 130 can be converted to its end points among the borderline region BA of N limit shape.Also or when source image SI was circular image, then borderline region BA was the Minimum Area that comprises target image DI.

Just enter step S370 after obtaining borderline region BA, a target position D P is obtained in video conversion unit 110 from borderline region BA, and will try to achieve the corresponding source position SP that comes with calculating in its substitution equation (10).Then in step S380, video conversion unit 110 checks source position SP whether to be positioned at source image SI.If come source position SP not to be positioned at source image SI, represent that this target position D P is not arranged in target image DI, but in Fig. 2 borderline region BA to the hatched example areas RA between the target image DI.Video conversion unit 110 is convenient among the step S381 pixel data of corresponding target position D P is set at default background pixel value, or this target position D P skipped over disregards.

If the source position SP that comes that tries to achieve is positioned at source image SI, then enter step S382, come that source position SP and the most approaching source position SP's of coming come source pixel to produce the pixel data of corresponding target position D P among the video conversion unit 110 basis source image SI.In order to eliminate or issuable edge sawtooth (aliasing) phenomenon when being reduced in video conversion, to use the present embodiment person and can look its design requirement and carry out corresponding action.In the present embodiment, video conversion unit 110 is carried out and is adapted to bicubic interpolation (adaptivebicubic interpolation) automatically, comes source pixel to try to achieve pixel data with calculating according to a plurality of of the predetermined number of the most approaching source position SP of coming.Wherein, adapt to the bicubic interpolation method automatically and be this area and have and know that usually the knowledgeable can understand and implement easily, do not add to give unnecessary details at this.

In other embodiment, this area has knows issuable edge crenellated phenomena when the knowledgeable can also utilize its interpolation computing method of knowing with elimination or minimizing video conversion usually, and the interpolation computational methods of present embodiment are not in order to limit the present invention.If pay no attention to the edge crenellated phenomena, video conversion unit 110 also can directly use the pixel data that the pixel value that comes source pixel of the most approaching source position SP of coming produces as institute's desire in another embodiment, and only use one to come source pixel to produce pixel data this moment.

Then in step S390, the pixel data that output unit 120 receiving step S381 or S382 are produced is perhaps ignored this target position D P according to step S381 is described as the target pixel value of target position D P.Output unit 120 just can produce target image DI according to the target pixel value of these target position D P, it is shown in display unit 150.In step S391, check whether target position D P all in the borderline region BA and its pixel value are disposed then.If all target position D P and its pixel value in the borderline region BA are not disposed as yet, video conversion unit 110 just comes back to step S370 and obtains next target position D P among the borderline region BA with repeating step S370 to S390, produces the pixel value that is positioned at all target position D P of borderline region BA by this.

The second embodiment of the present invention below is described, please refer to Fig. 4 and Fig. 5.Fig. 4 is the calcspar according to a kind of video conversion device 400 of second embodiment of the invention.Fig. 5 is the schematic diagram of mixing (blending) image method according to second embodiment of the invention.Wherein, present embodiment is similar to above-mentioned first embodiment, so same apparatus and process description do not repeat them here.Its difference is that video conversion device 400 also comprises image mixed cell 460, and it is covered in target image DI on the background video BI, and to produce composite image MI, wherein composite image MI can be shown in the display unit 150.

Describe the embodiment of the invention at this with another viewpoint, please refer to Fig. 6 (and asking simultaneously) with reference to Fig. 1 and Fig. 3.For the image SI of source, when video conversion device 100 obtained this source image SI, its video conversion unit 110 should calculate finished video conversion matrix T M and its inverse matrix TM ^-1(that is completing steps S310 to S350).Therefore it is as described below that the image SI that originates can directly carry out switch process:

Step S610 comes the source position to be converted to a plurality of end points target location of target image DI according to will originate a plurality of end points of image SI of video conversion matrix T M, to define the zone of source video conversion to target image.

Step S620 is according to video conversion matrix T M and its inverse matrix TM ^-1, the target location in the target image DI is calculated the source position of coming that produces corresponding target location.

In the step S630 basis source image come source position and the most approaching predetermined number that comes the source position come source pixel to produce pixel data.

Step S640 is with the target pixel value of pixel data as the target location, and to produce target image, wherein target image is shown in the display unit.

Can be according to above-mentioned steps corresponding in other application examples, video conversion unit 110 also may store many default video conversion matrix T M and its inverse matrix TM ^-1, in order to conversion source image SI.Therefore when realizing present embodiment, step S310～S350 differs and is decided to be the prerequisite step of conversion source image SI.Therefore, video conversion device 100 can be in obtaining at first to enter step S610 (promptly can be considered step S360) behind the image SI of source, edge calculations unit 130 comes source position E1, E2, E3 and E4 according to the end points of video conversion matrix T M and source image SI, uses and defines the borderline region BA that source image SI is converted to target image DI.

Enter step S620 (promptly can be considered step S370) afterwards, video conversion unit 110 comes source position SP according to video conversion matrix T M with what the target position D P in the target image DI calculate to produce corresponding target position D P.Then, (can be considered the blanket of step S380, step S381 and step S382) in step S630, video conversion unit source position SP in 110 future comes source pixel to produce pixel data with the most approaching source position SP's of coming.And in step S640 (can be considered step S390 and S391 blanket) with the target pixel value of this pixel data as target position D P.Then, video conversion unit 110 checks whether the target position D P in the borderline region BA is disposed by step S391, does not then get back to step S610 to continue the conversion image if dispose.When just the target position D P in the borderline region BA disposes, just can in step S640, export target image DI to display unit 150, use that target image DI is presented in the display unit 150.The thin portion of other of present embodiment flow process is included in the various embodiments described above, so do not repeat them here.

In sum, embodiments of the invention utilize the video conversion matrix to obtain the transfer equation formula of source image and target image, wherein the video conversion matrix is to utilize the three-dimensional rotation matrix that is projected to two dimensional surface to calculate and get with the affine matrix that move the position with the three-dimensional convergent-divergent of representative, rigid body distortion, to allow the image of conversion have the effect of approximate stereo rotating, distortion and convergent-divergent by lower-cost two-dimensional transformations framework.Embodiments of the invention utilize the borderline region of video conversion matrix computations video conversion, so do not need to calculate the pixel value of whole image, only need the central pixel value in computation bound zone, can improve video conversion efficient.In addition, by the inverse matrix of video conversion matrix, above embodiment can directly calculate the pixel value of trying to achieve target location in the target image, need not use buffer with instant conversion image.

Though the present invention discloses as above with embodiment; right its is not in order to limit the present invention; have in the technical field under any and know the knowledgeable usually; without departing from the spirit and scope of the present invention; when doing a little change and retouching, so protection scope of the present invention is worked as with being as the criterion that claim was defined.

Claims (14)

1. an image conversion method is characterized in that, described image conversion method comprises:

Come the source position to be converted to a plurality of end points target location of a target image a plurality of end points of one source image according to a video conversion matrix, to define the zone of described source video conversion to described target image;

According to the inverse matrix of described video conversion matrix, the target location in the described target image calculated produce one of corresponding described target location and come the source position, wherein saidly come the source position to be arranged in described source image;

Come source pixel to produce a pixel data according to what described in the described source image came source position and the most approaching described predetermined number that comes the source position; And

With the target pixel value of described pixel data as described target location, to produce described target image, wherein said target image is shown in the display unit.

2. image conversion method as claimed in claim 1 is characterized in that, the step that produces described pixel data comprises utilizing above-mentionedly comes source pixel to make interpolation to calculate, to produce described pixel data.

3. image conversion method as claimed in claim 1 is characterized in that, described predetermined number is one, and the step that produces described pixel data comprises that direct use is the most approaching and describedly comes the described of source position to come the pixel value of source pixel as described pixel data.

4. image conversion method as claimed in claim 1, it is characterized in that, described video conversion matrix is tried to achieve according to a two-dimentional spin matrix and an affine transition matrix, and described two-dimentional spin matrix is the three-dimensional rotation matrix that a three-dimensional rotation angle is formed to be projected to two dimensional surface produce.

5. image conversion method as claimed in claim 4 is characterized in that, described affine transition matrix comprises that an X-axis zooming parameter, a Y-axis zooming parameter, an X-axis deformation parameter, a Y-axis deformation parameter, an X-axis moving parameter and a y-axis shift move parameter.

6. image conversion method as claimed in claim 4 is characterized in that, described video conversion matrix is the product of described two-dimentional spin matrix and described affine transition matrix.

7. image conversion method as claimed in claim 1 is characterized in that, described image conversion method comprises also described target image is covered on the background video that to produce a composite image, wherein said composite image is shown in the described display unit.

8. image conversion method as claimed in claim 1 is characterized in that, one of the corresponding described target location of generation is calculated in the target location in the described target image come the source position also to comprise:

Check describedly come the source position whether to be positioned at described source image;

Come the source position not to be positioned at described source image when described, the pixel data of the described target location of correspondence is set at default background pixel value;

Come the source position to be positioned at described source image when described, then come source pixel to produce a pixel data according to what described in the image of described source came source position and the most approaching described predetermined number that comes the source position.

9. a video conversion device is characterized in that, described video conversion device comprises:

One video conversion unit, produce a three-dimensional rotation matrix according to a three-dimensional rotation angle, with the projection of described three-dimensional rotation matrix in two dimensional surface to produce a two-dimentional spin matrix, according to a described two-dimentional spin matrix and an affine transition matrix, come the source position by the target location in the target image to calculate one of the corresponding described target location of generation, wherein saidly come the source position to be arranged in a source image, and come source pixel to produce a pixel data according to what described in the image of described source came source position and the most approaching described predetermined number that comes the source position; And

One output unit receives the target pixel value of described pixel data as described target location, and to produce described target image, wherein said target image is shown in the display unit.

10. video conversion device as claimed in claim 9, it is characterized in that, described video conversion unit multiply by described two-dimentional spin matrix and described affine transition matrix to calculate mutually tries to achieve a video conversion matrix, calculate again and try to achieve an inverse matrix of described video conversion matrix, and utilize described inverse matrix and described target location to calculate that generation is described to come the source position.

11. video conversion device as claimed in claim 10 is characterized in that, described video conversion device also comprises:

One edge computing unit, it utilizes described video conversion matrix to come the source position to be converted to a plurality of end points target location of described target image a plurality of end points of described source image, to define the zone of described source video conversion to described target image.

12. video conversion device as claimed in claim 9 is characterized in that, described video conversion device also comprises:

One image mixed cell is covered in described target image on one background video, and to produce a composite image, wherein said composite image is shown in the described display unit.

13. a method that produces the video conversion matrix is characterized in that, the method for described generation video conversion matrix comprises:

Utilize a three-dimensional rotation angle to produce a three-dimensional rotation matrix;

With the projection of described three-dimensional rotation matrix in two dimensional surface, to produce a two-dimentional spin matrix;

Try to achieve a video conversion matrix according to described two-dimentional spin matrix and described affine transform matrix calculations; And

An inverse matrix of described video conversion matrix is tried to achieve in calculating,

Wherein, when image is originated in acquisition one, produce a target image according to described inverse matrix and described source image, so that described target image is shown in the display unit, and one of the corresponding described target location of a target pixel value utilization of a target location of described target image comes the pixel value of source position to produce, use producing described target image, describedly come the source position to be arranged in described source image.

14. method as claimed in claim 13 is characterized in that, described method also comprises:

Come the source position to be converted to a plurality of end points target location of described target image a plurality of end points of described source image by described video conversion matrix, to define the zone of described source video conversion to described target image.

Images