CN102986221B

CN102986221B - Picture coding device and integrated circuit thereof and method for encoding images

Info

Publication number: CN102986221B
Application number: CN201280001281.1A
Authority: CN
Inventors: 清水健二
Original assignee: Panasonic Intellectual Property Corp of America
Current assignee: Panasonic Intellectual Property Corp of America
Priority date: 2011-07-19
Filing date: 2012-04-06
Publication date: 2016-11-30
Anticipated expiration: 2032-04-06

Abstract

When by picture coding, it is possible to image is applied fuzzy according to the distance of subject, and suitably determine the block size as the unit encoded.Use the parallax information of each pixel unit of the image of input or the range information of each pixel unit and arbitrary hazy condition, additional fuzzy to this image, and the fuzzy quantity according to each unit area determines the block size in coded object region, divides the image into as coded object region according to the block size determined and encodes.Bigger by the block size that makes image-region that fuzzy quantity is bigger, it is possible to reduce the search treating capacity of reference block, it is achieved the high speed of process and the reduction for the treatment of capacity.

Description

Picture coding device and integrated circuit thereof and method for encoding images

Technical field

The present invention relates to image adds image coding technique that is fuzzy and that encode.

Background technology

As for showing the relief processing that main subject of sening as an envoy to is obvious, carry out focus only in alignment with image The subject of a part is it is intended that prospect beyond it, the process of blurred background.Additionally, at the stereoscopic image for stereopsis In signal, when the attention of beholder is towards distant view, it may occur that beholder feels that what stronger fatigue or stereopsis destroyed asks Topic (for example, referring to non-patent literature 1).In order to prevent this problem, the distant view carrying out making stereoscopic image signals is fuzzy in order to avoid watching Person notes the process towards distant view.

On the other hand, in picture coding, improve compression coding efficiency to remove aggregative redundancy, by many Set of pixels carries out block segmentation.Thus, in utilizing the relevant predictive coding improving compression ratio between this block, if to complexity Image carry out inter prediction and infra-frame prediction with less block unit, smooth image carried out interframe with bigger block unit Prediction and infra-frame prediction, then complicated image and smooth image can be predicted expeditiously.

As the past case of the code device changing block size according to the feature in coded object region, disclose patent literary composition Offer 1.It is said that in general, in diagnostic ultrasound equipment, the exploring degree being closer to focal position then image is the highest, further away from focus Position then exploring degree is the lowest.Therefore, use range information and focal position information to determine block size, be thus simultaneously achieved analysis The higher image quality in the region that picture degree is higher and the high compression in the relatively low region of exploring degree.

Additionally, as another past case, disclose patent documentation 2.In the technology of patent documentation 2, it is conceived to flatness Situation that higher stagnant zone easily flashes and increase prediction block sizes, thus generate the compression image inhibiting flicker.

Prior art literature

Patent documentation

Patent documentation 1: JP 2007-289556 publication

Patent documentation 2: JP 2007-67469 publication

Non-patent literature

Non-patent literature 1: Honda Jie Fuxiu " new advancement-of stereoscopic image technology-space performance media " CMC goes out Version, July 31 (P61-P62) in 2008

Brief summary of the invention

The problem that invention is to be solved

In the technology of patent documentation 1, determine the block size of coding according to range information and focal position information.But, For various Fuzzy Processing, the block size determined according only to focal position information and range information might not be suitable for The block size of the image after Fuzzy Processing.

On the other hand, in the technology of patent documentation 2, determine block size according to the planarization of image-region.But, only By whether the smooth block size judging to determine might not be suitable for the block size of the image after Fuzzy Processing.

Summary of the invention

The present invention makes in view of the above problems, it is therefore an objective to provide one executing according to parallax information or range information Suitable block size is selected to carry out picture coding device and the method for encoding images encoded after adding various Fuzzy Processing.

Means for solving the above

In order to solve above-mentioned conventional problem, the picture coding device of the present invention is characterised by, including: image obtains Portion, obtains image；Parallax obtaining section, obtains parallax or the distance of each photograph primitive unit cell of each pixel unit of above-mentioned image At least one party in information；Fuzzy quantity determination section, according to the parallax of each above-mentioned pixel unit or each above-mentioned pixel unit Range information, determines the fuzzy quantity of each pixel unit of above-mentioned image；Fuzzy Processing portion, according to above-mentioned fuzzy quantity, to above-mentioned Image implements Fuzzy Processing；Block size determination section, according to above-mentioned fuzzy quantity, determines to be used for from implementing among multiple block sizes State each image after Fuzzy Processing and cut out the block size in coded object region；And encoding section, according to the block size of above-mentioned decision, Image after implementing above-mentioned Fuzzy Processing encodes with block unit.

Invention effect

According to this structure, fuzzy quantity is used to determine block size when by picture coding such that it is able to Fuzzy Processing is used Suitable block size corresponding with fuzzy quantity.

Accompanying drawing explanation

Fig. 1 is the module map of the picture coding device of embodiments of the present invention 1.

Fig. 2 is the figure of photographic optical system and the fuzzy quantity representing embodiments of the present invention 1.

Fig. 3 is the figure of the example of the fuzzy matrix representing embodiments of the present invention 1.

Fig. 4 is the flow chart of the action of the picture coding device representing embodiments of the present invention 1.

Fig. 5 is the action of the generation of the coded object signal of the picture coding device representing embodiments of the present invention 1 Flow chart.

Fig. 6 is the action that the block size in the infra-frame prediction of the picture coding device of embodiments of the present invention 1 determines Flow chart.

Fig. 7 is that the block size in the motion compensated prediction of the picture coding device representing embodiments of the present invention 1 determines The flow chart of action.

Fig. 8 is the dynamic of the block size decision in the disparity compensation prediction of the picture coding device of embodiments of the present invention 1 The flow chart made.

Fig. 9 is the module map of the picture coding device of the variation 1 of embodiments of the present invention 1.

Figure 10 is the module map of the picture coding device of embodiments of the present invention 2.

Figure 11 is the flow chart of the action of the picture coding device representing embodiments of the present invention 2.

Figure 12 is the flow chart of the action of the block size decision of the picture coding device representing embodiments of the present invention 2.

Figure 13 is the module map of the picture coding device of embodiments of the present invention 3.

Figure 14 is the flow chart of the action of the picture coding device representing embodiments of the present invention 3.

Figure 15 is the module map of the picture coding device of embodiments of the present invention 4.

Figure 16 is the subject representing embodiments of the present invention 4 and focus, the figure of the position relationship of imaging apparatus.

Figure 17 is the module map of the image encoding system of embodiments of the present invention 5.

Figure 18 is the module map of the picture coding device of the integrated circuit as embodiments of the present invention 1.

Figure 19 is the module map of the diagnostic ultrasound equipment possessing conventional coding unit.

Figure 20 is the flow chart representing and illustrating conventional coding control method.

Detailed description of the invention

As one of image procossing, the fuzzy process of distance based on subject is important technology.Such as, as with Show relief processing in making main subject obvious, carry out only by focus in alignment with a part being shot of image Body is it is intended that the prospect beyond it and the process of blurred background.In such Fuzzy Processing, have by the use subject degree of depth relatively " dissipating scape (bokeh) " of occurring in the case of shallow optical system photography reproduces and makes the situation of prospect and blurred background.Additionally, Also have various process beyond it, such as: for the fuzzy of background and for prospect fuzzy chien shih vague intensity change or By focus in alignment with being in the prospect of the personage of recent side apart from different multiple personages making and being in the most inboard personage's Blurred background or by focus in alignment with the building both sides at personage and its rear make personage and building both sides obvious and Make the subject beyond it fuzzy.

Additionally, in the stereoscopic image signals of stereopsis, as described in non-patent literature 1, when the attention of beholder During towards distant view, if causing the inconsistent of distance represented by parallax information and focal length, then beholder is had to feel stronger Fatigue or the problem of stereopsis destruction etc..Particularly, at the three-dimensional shadow that will make with the premise in projection to less screen In the case of image signal amplification is shown on bigger screen, also can in right eye and the interval of left eye contemplated by stereoscopic image signals It is exaggerated.Therefore, for a subject, the shot object image in right eye image and the shot object image in left eye image Distance become bigger than the interval of right eye and left eye in the case of (rear dissipates), becoming the parallax of this subject can not become and see The state of the parallax in the case of single object, so stereopsis destroys.In order to prevent this problem, carry out making stereoscopic image The distant view of signal is fuzzy so that the attention of beholder is not towards the process of distant view.

On the other hand, when by picture coding, improve compression coding efficiency to aggregative redundancy be removed, will Many set of pixels carry out block segmentation.Thus, in utilizing the relevant predictive coding improving compression ratio between this block, if right Complicated image carries out inter prediction and infra-frame prediction with less block unit, carries out smooth image with bigger block unit Inter prediction and infra-frame prediction, then complicated image and smooth image can be predicted expeditiously.

Here, inventors are conceived to be applied with stronger fuzzy region is applicable bigger block size, the most multiple Miscellaneous pattern and smooth region.Further, following design is obtained: if the image after processing without analytic fuzzy just can obtain As the region of stronger fuzzy object, then the block size when by picture coding after Fuzzy Processing need not in selecting into The block size in this region of row selects, it is possible to cut down the operand for coding.This is because, if so, be then made without The most again detect at as region such block size selection of stronger fuzzy object according to the image after Fuzzy Processing Reason.

In contrast, in conventional block size selection technique, using encoded object image itself as analysis object.Cause And, in the case of the image before Fuzzy Processing is used conventional block size selection technique, there is selected block size also The problem being not necessarily suited for the coding of the image after Fuzzy Processing.

Patent documentation 1 is the diagnostic ultrasound equipment using range information and focus information to select block size, and Figure 19 is special The module map of the diagnostic ultrasound equipment of profit document 1.This diagnostic ultrasound equipment using as depth information reflected signal as Input image data encodes.Owing to the image exploring degree of diagnostic ultrasound equipment is determined by the distance away from focal position, institute To use range information and focal position information to determine block size, thus it is simultaneously achieved the high image quality in the higher region of exploring degree Change the high compression in the region relatively low with exploring degree.

But, even if distance and focal position it is determined that, but as long as the content determined of Fuzzy Processing, then fuzzy intensity Would not be determined uniquely.That is, to same image, the same subject that is not at focal position obscures In the case of process, the background of situation and only focusing position that the prospect in only focusing position carries out Fuzzy Processing carries out mould In the case of paste processes, the region of this subject in image after Fuzzy Processing in the case of the former situation and the latter when The most different.That is, even if the image before Fuzzy Processing is identical, but as long as the content of Fuzzy Processing is different, then Fuzzy Processing it is suitable for After the suitable block size of coding of image might not be identical.Thus, for various Fuzzy Processing according only to distance and Focal position selects in the method for the patent documentation 1 of block size, might not obtain being suitable for the block of the image after obscuring Size.

Additionally, patent documentation 2 is the coding control method of flatness based on image suppression flicker, Figure 20 is to represent patent The flow chart of the coding control method of document 2.In the technology of patent documentation 2, it is made whether it is the judgement of stagnant zone, connects , to being judged as the region of stagnant zone, calculate the planarization in this region according to the dispersion etc. of pixel value.Further, based on this The planarization calculated and quantization step size, it is judged that flicker level, make the block size in the region that flicker degree is higher increase, from And generate the compression image inhibiting flicker.

But, as described above, even the smooth region of complicated image also can be suitable for less block size.Carry out Although stronger fuzzy region is suitable for bigger block size, but has complexity owing to there may exist in unambiguous region The smooth region of pattern, so according only to whether being that smooth region is to determine the technology of the patent documentation 2 of block size In can not select suitable block size.Additionally, in the technology of patent documentation 2, it is impossible to the image before Fuzzy Processing detects That occur after Fuzzy Processing, be suitable for the stronger fuzzy region of carrying out of bigger block size.

So, inventors complete the present invention, it is characterized by have kept each pixel unit when carrying out Fuzzy Processing Fuzzy intensity, use the fuzzy intensity kept to determine block size.Thus, when addition of the coding after obscuring, right Suitable block size can be selected in being applied with stronger fuzzy region, additionally, carry out the coding of predictive pictures in execution In the case of, it is no longer necessary to carry out using the predictive pictures of less block size, so the image after Fuzzy Processing need not be carried out Resolve the operand that just can cut down coding.

Hereinafter, referring to the drawings embodiments of the present invention are illustrated.

(embodiment 1)

In FIG, stereophotography coding system by can photographic subject in the way of stereopsis, and will be generated by photography Stereoscopic image signals coding.Such stereophotography coding system possesses photography and has the master image of parallax and the photograph of sub-picture Machine 100a and 100b, picture coding device 200, fuzzy parameter input unit 300 and storage part 400.

Photographing unit 100a and 100b the most mutually leaves the distance of the 6.5cm of the equispaced of two as people and joins Put.Photographing unit 100a and 100b photographic subject respectively, by defeated to picture coding device 200 for the signal of video signal obtained by photography Go out.

Hereinafter, in order to simplify, the signal of video signal exported respectively from photographing unit 100a and 100b is referred to as stereoscopic image Signal.Additionally, the image represented by the signal of video signal exported from photographing unit 100a is referred to as master image, will be by from photographing unit 100b The image that the signal of video signal of output represents is referred to as sub-picture.

By this master image and sub-picture, the picture of subject is carried out stereopsis.Photographing unit 100a and 100b is respectively identical Timing (moment) generate picture exporting.

Fuzzy parameter input unit 300 be by for the stereoscopic image signals of master image and sub-picture additional with by taking a picture The input of the fuzzy parameter of the scattered scape (Pot け) that scattered scape that the actual photographic lens of machine 100a and photographing unit 100b brings is different User interface, be made up of touch panel and cross key.

As the fuzzy parameter of input, the fuzzy adjustment amount (S of the position of input subject front or behind_n, S_f) and Subject front and each weight coefficient (W at rear_n, W_f), the focal position information (P) of subject, between photographing unit 100a and 100b Base length (l), pixel size (m), lens information different from the information of the photographic lens of reality for focal length fp, i.e. false The focal length fv thought and f-number (F value).

Fuzzy adjustment amount (S to the front and back of subject_n, S_f) and weight coefficient (W_n, W_fFor), standing In the case of stereoscopic, when wanting to suppress the situation that jumps out in front to be presented with the far and near sense of depth, by increasing the mould in front Stick with paste adjustment amount S_nOr weight coefficient W_n, or reduce rear fuzzy adjustment amount S_fOr weight coefficient W_f, increase subject front Fuzzy quantity.On the contrary, make the subject in front obvious in the subject by emphasizing front it is intended that blurred background shows During third dimension, by increasing fuzzy adjustment amount S at rear_fOr weight coefficient W_f, or reduce front fuzzy adjustment amount S_nOr power Weight coefficient W_n, increase the fuzzy quantity at subject rear.

If it addition, increasing fuzzy adjustment amount (S_n, S_f), then can be with subject before independently focusing position Side or rear uniformly apply Fuzzy Processing.On the other hand, if increasing weight coefficient (W_n, W_f), then subject is further away from focus Position is then fuzzy the strongest, it is possible to carry out depending on the Fuzzy Processing of subject distance.

Lens information and entering through of weight coefficient are called setting item menu from touch panel and cross key and set Projects are carried out.Focal position is by the position of the subject in the master image to the photographing unit 100a shown on touch panel Put the operation carrying out specifying to input.

Fuzzy parameter input unit 300 accepts the setting of these fuzzy parameters, exports to picture coding device 200.Thus, exist Fuzzy Processing portion 202 can be carried out the arbitrary scattered scape shape that can not will be generated by the lens of virtual photographic lens or reality The Fuzzy Processing that condition can show.

Storage part 400 is used to preserve view data (the local decoder image described later from picture coding device 200 output Signal) record medium.

Picture coding device 200 is encoded by the stereoscopic image signals that will export from photographing unit 100a and 100b, generates and compiles Code signal of video signal also exports.Here, in the present embodiment, it is assumed that the picture coding device 200 master map to stereoscopic image signals Picture and sub-picture both sides carry out Fuzzy Processing, and stereoscopic image signals are encoded.Additionally, picture coding device 200 will comprise When two signal of video signal in stereoscopic image signals encode, according to each region constituting picture, these signal of video signal are compiled Code.Picture coding device 200 is the signal of video signal exported from photographing unit 100a will to be applied with fuzzy picture signal with H.264 MVC(Multi View Coding) device of form coding, will be contained in each picture in signal of video signal respectively as I picture, P picture and B coding of graphics.

It addition, when by I picture and P coding of graphics, picture coding device 200 carries out intraframe predictive coding.

It addition, when by P picture and B coding of graphics, picture coding device 200 carries out inter prediction encoding (motion compensation Predictive coding).

On the other hand, the signal of video signal exported from photographing unit 100b is applied with fuzzy at coding by picture coding device 200 Picture signal time, carry out disparity compensation prediction coding, will be contained in each picture in signal of video signal and compile respectively as P picture Code.

That is, the secondary figure that picture coding device 200 generates in the timing identical with this picture according to the picture prediction of master image The picture of picture, predicts the outcome the coding of graphics of sub-picture based on this.

Such picture coding device 200 possesses image acquiring section 206, parallax obtaining section 207, Fuzzy Processing portion 202, choosing Select device 203, switch 204a, 204b and 204c, subtractor 205, block size determination section 210a and 210b, block cutting part 212a and 212b, infra-frame prediction portion 213, test section 214, compensation section 215, transformation component 216, quantization unit 217, variable-length encoding part 218, inverse Quantization unit 219, inverse transformation portion 220 and adder 221.

Image acquiring section 206 obtains the picture signal from photographing unit 100a and 100b output, defeated to Fuzzy Processing portion 202 Go out.

Parallax obtaining section 207 obtain in the picture signal that image acquiring section 206 accepts respectively want primitive unit cell, subject with The value of the parallax between photographing unit 100 or the value of distance, export the value of parallax or the value of distance to Fuzzy Processing portion 202.? In present embodiment, parallax obtaining section 207 includes that parallax test section 201, parallax test section 201 generate the value of parallax.

Parallax test section 201 detects from the parallax between the master image and auxiliary image data of photographing unit 100a, 100b input.Depending on Poor detects the correspondence position by obtaining between master image and sub-picture, by the unit area of side's image and the opposing party's image The deviation of the position of the horizontal direction in corresponding region is carried out as pixel count detection.Parallax test section 201 will detect Parallax information d exports to Fuzzy Processing portion 202.

In the present embodiment, the detection of parallax is by using Block Matching Algorithm to carry out.In Block Matching Algorithm, first from than The image in relatively source cuts out the unit area of 16 × 16 pixels.Then, multiple identical 16 are cut out from the search domain of comparison object The region of × 16 sizes, for the regional cut out, obtains the summation of the absolute value of the luminance difference in region with reference source (SAD；Sum of AbsoluteDifference).Further, by find out SAD minimum cut out position, obtain with territorial unit Position corresponding to reference source image.

Fuzzy Processing portion 202 uses from the fuzzy parameter of fuzzy parameter input unit 300 input with from parallax obtaining section 207 The parallax information of input or range information, determine the master image inputted from image acquiring section 206 and each pixel list of sub-picture The fuzzy matrix that position is additional.In the present embodiment, Fuzzy Processing portion 202 uses the parallax letter from parallax obtaining section 207 input Breath.Fuzzy Processing portion 202 by the fuzzy quantity of pixel unit that calculate according to fuzzy matrix, each to block size determination section 210a And 210b output.

Additionally, Fuzzy Processing portion 202 carries out the Fuzzy Processing additional to master image or auxiliary image data based on fuzzy matrix, The master image of result and each blurs image data of sub-picture are exported to selector 203.

It addition, in the present embodiment, pixel unit is made up of a pixel.

The flow process of Fuzzy Processing is as follows.Fuzzy Processing portion 202 use parallax information d obtain from photographing unit to subject away from From distribution, generate fuzzy matrix, by master image according to from the parameter of fuzzy parameter input unit 300 and the distance of subject Or each pixel region superposition fuzzy matrix of auxiliary image data adds fuzzy.Meanwhile, Fuzzy Processing portion 202 is according to fuzzy matrix Calculate fuzzy quantity, export to block size determination section 210a and 210b.

Use Fig. 2 illustrates.First, obtain from photographing unit to subject according to the parallax information d of each pixel unit Range distribution.As shown in Figure 2 (a) shows, if setting parallax information as d, the level side of 1 pixel of imaging apparatus that uses in photography A length of m upwards, then subject is as follows with distance D of photographing unit:

［ numerical expression 1 ］

D = \frac{f_{p} \cdot 1}{m \cdot d}

It addition, as it has been described above, f_pBeing the focal length of lens, l is the base length between photographing unit 100a and 100b.

By whole pixel units being carried out such calculating, obtain being shot in master image or sub-picture of shooting The range distribution of body.

The imaging of the point of the position distance with photographing unit deviateed relative to focal position P i.e. dissipates the circle of scape and is referred to as disperse Circle, but in the present embodiment, use imaginary blur circle diameter σ to determine the size of fuzzy matrix.

Focal position P passes through the parallax information at the P of focal position in the calculating formula of above-mentioned D with distance L of photographing unit It is updated in d obtain the value into D.Equally, arbitrary subject and distance X of photographing unit are by the calculating formula of above-mentioned D It is updated in d obtain the value into D by the parallax information of this subject.

It is positioned at the blur circle diameter at distance X relative to photographing unit and uses the focal length of the characteristic value as optical system Fv, F value F, focal position calculate with distance X of subject as follows with distance L, the photographing unit of photographing unit.

In the present embodiment, F value F uses the imagination value from fuzzy parameter input unit 300 input.Additionally, use from mould Stick with paste fuzzy adjustment amount S of parameter input unit 300 input_f、S_n, weight coefficient W_f、W_n, contrast focal position is rearward and front Blur circle diameter σ is weighted.Fuzzy adjustment amount S_f、S_nWith weight coefficient W_f、W_nAs described above, if S_f＞ S_nOr W_f＞ W_n, then rear is fuzzy relatively strong, and obscuring of front is more weak, if S_f＜ S_nOr W_f＜ W_n, then front is fuzzy relatively strong, rear Fuzzy more weak.If additionally, fuzzy adjustment amount S_f、S_nIt is all 0, then becomes there is no adjustment amount, become and produced by virtual optical system Scattered scape, value is the biggest, more can the front or behind of focusing position uniformly apply to obscure.Additionally, weight coefficient W_f、W_nIf Being all 1, then become not weighting, becoming the scattered scape itself produced by virtual optical system, value is the biggest, produces strongly the most interdependent Obscuring in subject distance, if being all 0, then obscures front or behind and no longer depends on subject distance.

If adding adjustment amount after the calculating formula of disperse diameter being approximated and being weighted, then than focal position rearward Imaginary blur circle diameter σ as follows:

［ numerical expression 2 ］

σ = \frac{W_{f} \cdot {f_{v}}^{2} \cdot (X - L)}{FLX} + S_{f}

As follows than focal position imaginary blur circle diameter σ on the front:

［ numerical expression 3 ］

σ = \frac{W_{n} \cdot {f_{v}}^{2} \cdot (L - X)}{FLX} + S_{n}

If according to each pixel unit in master image and sub-picture to the calculating formula input parallax information d of above-mentioned D Obtain distance X, and the calculating formula transmission range X to this imagination blur circle diameter σ, then can obtain the size of fuzzy matrix.? Fig. 2 (b) represents based on numerical expression 2, the imaginary blur circle diameter σ of several 3 and the relation of distance.Then, as shown in Fig. 3 (a), make The size i.e. inside of the imagination blur circle diameter σ circle as diameter with fuzzy matrix is set to filter factor 1, outside is set to filter The fuzzy matrix of wave system number 0.

Then, add by each pixel region of the fuzzy matrix generated with master image or auxiliary image data is superposed Fuzzy.Now, the first sum of the whole filter factors in calculating fuzzy matrix, carry out by this value except filter factors whole in matrix Division, generate within whole filter factor sums be 1 superposition matrix.This superposition matrix is used to be overlapped.

Additionally, according to the fuzzy matrix of each pixel unit obtained in above-mentioned, calculate the fuzzy of each pixel unit Amount.Fuzzy quantity is the value of the scope that filter factor is more than threshold value in instruction fuzzy matrix.For the value of indicating range, example As being the most then diameter, if square is then cornerwise length.

In the present embodiment, fuzzy matrix is will to be set to filtering with the inside of the imagination blur circle diameter σ circle as diameter Coefficient 1, outside is set to the matrix of filter factor 0, so setting threshold value as 1, as the value of the circle of instruction diameter σ, by fuzzy quantity It is set to σ.

Fuzzy Processing portion 202 by the fuzzy quantity of each pixel unit obtained as described above to block size determination section 210a And 210b output.

Selector 203 obtains the signal of video signal exported respectively from photographing unit 100a and 100b, by these signal of video signal alternately Ground switches and exports to switch 204a, subtractor 205, block cutting part 212a or 212b.

In the present embodiment, selector 203 switches over picture unit.That is, if selector 203 is by Fuzzy Processing After master image obtain from Fuzzy Processing portion 202 with identical timing respectively with sub-picture, first output Fuzzy Processing after Master image, the then sub-picture after output Fuzzy Processing.

Subtractor 205 according to each piece calculate from selector 203 output signal of video signal represented by encoded object image, Difference with the prognostic chart picture represented by the prediction signal from switch 204b output.Subtractor 205 would indicate that the forecast error of this difference Signal is to switch 204a output.

Here, subtractor 205 in each picture of the image (block) that encoded object image is I picture, P picture or B picture, Calculate above-mentioned difference and export predictive error signal.

Switch 204a encoded object image be I picture master image, do not carry out infra-frame prediction in the case of, by selector 203 are connected to transformation component 216, would indicate that the signal of video signal of this encoded object image transmits to transformation component 216 from selector 203.

Additionally, switch 204a encoded object image be I picture master image, carry out infra-frame prediction in the case of or Encoded object image be P picture or B picture image (master image or sub-picture) in the case of, subtractor 205 is connected to Transformation component 216, transmits each predictive error signal of I picture, P picture or B picture to transformation component 216 from subtractor 205.

Transformation component 216 obtains signal of video signal or predictive error signal according to each piece via switch 204a to be believed as image Number, this picture signal is carried out the discrete cosine transform (DCT as orthogonal transformation；Discrete Cosine Transform). Thus, image signal transformation is the DCT coefficient as coefficient of frequency by transformation component 216, i.e. by the region of this picture signal from sky Between region be transformed to frequency field, this coefficient of frequency vectorization portion 217 is exported.

Quantization unit 217 obtains coefficient of frequency from transformation component 216, quantifies this coefficient of frequency.That is, quantization unit 217 leads to Cross by coefficient of frequency divided by quantization step generating quantification value.Here, quantization unit 217 is according to image encoding standards, in this enforcement According to H.264 in mode, the quantization step determined is used to be quantified by this coded object block.

<variable-length encoding part 218>

Variable-length encoding part 218 is by vowing the quantized value generated by quantization unit 217 and the motion exported from test section 214 Amount or difference vector carry out Variable Length Code, generate encoded stereoscopic signal of video signal and export.It addition, this Variable Length Code is reversible Coding.

<re-quantization portion 219>

Re-quantization portion 219 generates re-quantization frequency system by the quantized value generated by quantization unit 217 is carried out re-quantization Number.That is, re-quantization portion 219 is by being multiplied by quantized value to the quantization step used in quantization unit 217, generates re-quantization frequency system Number.

It addition, the re-quantization coefficient of frequency now generated is different from the coefficient of frequency generated by transformation component 216, comprise quantization Error.

The re-quantization coefficient of frequency generated by re-quantization portion 219 is carried out carrying out as transformation component 216 by inverse transformation portion 220 The inverse orthogonal transformation of the inverse operation of orthogonal transformation, carries out inverse discrete cosine transform (iDCT in the present embodiment；inverse Discrete Cosine Transform).Thus, re-quantization coefficient of frequency is transformed to picture signal by inverse transformation portion 220, i.e. The region of this re-quantization coefficient of frequency is transformed to area of space from frequency field, this picture signal is exported to adder 221.

Adder 221 the master image that encoded object image is I picture, do not carry out infra-frame prediction in the case of, will be from inverse The picture signal of transformation component 220 output preserves to storage part 400 as local decoder picture signal.

Additionally, adder 221 the master image that encoded object image is I picture, carry out infra-frame prediction in the case of, or Encoded object image be P picture or B picture image (both can be master image can also be sub-picture) in the case of, will be from The picture signal of inverse transformation portion 220 output is added, using its addition result as local with the prediction signal from switch 204c output Decoded image signal exports to infra-frame prediction portion 213.Meanwhile, will add up result to preserve to storage part 400.

Block size determination section 210a is carried out in frame in each image that encoded object image is I picture, P picture or B picture In the case of prediction, obtain the fuzzy quantity in encoded object image region from Fuzzy Processing portion 202, with threshold ratio relatively.

In the present embodiment, use H.264 as coded system as described above, so in infra-frame prediction, can To use the block size of 16 × 16,4 × 4.Here, if threshold value is 4 pixels, the region that fuzzy quantity is bigger than 4 pixels does not use 4 × 4 Block size.

In the case of fuzzy quantity is bigger than threshold value, export the block will predicted by infra-frame prediction portion 213 to block cutting part 212a Unit is limited to the block size information of the bigger block size of 16 × 16.

On the contrary, in the case of fuzzy quantity is less than threshold value, in addition to 16 × 16, also include 4 to block cutting part 212a output The less block size of × 4 is at interior block size information.

Block cutting part 212a, based on the block size information exported from block size determination section 210a, will export from selector 203 Encoded object image according to block size information size split.

Such as, in the case of block size information is restricted to 16 × 16, only with 16 × 16 size by coded object figure As segmentation, segmentation image is exported to infra-frame prediction portion 213.

On the other hand, in the case of block size information is constituted by two kinds of 16 × 16 and 4 × 4, prepare coded object The image that image is split with the block size of 16 × 16 and the image that identical encoded object image is split with the block size of 4 × 4 Prepared two kinds of segmentation images are exported by both respectively to infra-frame prediction portion 213.

<infra-frame prediction portion 213>

Infra-frame prediction portion 213 for the local decoder picture signal from the identical picture that adder 221 exports, to from The position of encoded object image when observing be left, upper, upper right, upper left amount to 4 positions, with export from block cutting part 212a The unit of segmentation block size carries out infra-frame prediction.

Such as, in the case of the block message receiving the block size being restricted to 16 × 16 from block cutting part 212a, to 4 Individual position, the block size with 16 × 16 carries out the search of same or similar piece.

On the other hand, in the case of receiving 16 × 16 and 4 × 4 both block size informations, to 4 positions, with 16 Two Seed-ginger size of × 16 and 4 × 4 carry out the search of same or similar piece.

Then, the most similar prediction signal obtained by search is exported to switch 204b and 204c.

Block size determination section 210b is situation and the P picture of the master image of P picture or B picture at encoded object image In the case of sub-picture, obtain the fuzzy quantity in encoded object image region from Fuzzy Processing portion 202, with threshold ratio relatively.

In the present embodiment, as described above, use H.264 as coded system, so can in inter prediction To use the block size of 16 × 16,8 × 8.Here, if threshold value is 8 pixels, the pixel that fuzzy quantity is bigger than 8 pixels does not use 8 × 8 Block size.It addition, the block size of 16 × 8,8 × 16 can also be used in standard H.264, but the most not Use.

In the case of fuzzy quantity is bigger than threshold value, will be by test section 214 searching motion vector to block cutting part 212b output Or the block unit of difference vector is limited to the block size information of bigger block size of 16 × 16.Here, 8 are used as threshold value Pixel.

On the contrary, in the case of fuzzy quantity is less than threshold value, in addition to 16 × 16, also include 8 to block cutting part 212b output The less block size of × 8 is at interior block size information.

Block cutting part 212b, based on the block size information exported from block size determination section 210b, will export from selector 203 Encoded object image according to block size information size split.

Such as, in the case of being limited to the block size information of 16 × 16, only with 16 × 16 size by coded object figure As segmentation, segmentation image is exported to test section 214.

On the other hand, in the case of receiving the two Seed-ginger size information of 16 × 16 and 8 × 8, prepare coded object Image with 16 × 16 block size split image and use identical encoded object image with 8 × 8 block size split figure As both, two segmentation images are exported to test section 214.

The test section 214 encoded object image after being split by block cutting part 212b block is P picture or the master image of B picture In the case of, for this encoded object image according to each piece of detection motion vector.

That is, other master image (I picture that the expression being saved in storage part 400 has been encoded and has decoded by test section 214 Or P picture) local decoder picture signal as with reference to image reference.

Further, test section 214 by search from this reference image and encoded object image (coded object block) is identical or Similar block, detects the motion of this encoded object image as motion vector.

Further, this motion vector is exported by test section 214 to compensation section 215 and variable-length encoding part 218.

On the other hand, the test section 214 encoded object image after being split by block cutting part 212b block is the secondary figure of P picture In the case of Xiang, to this encoded object image according to each piece of detection difference vector.

That is, the expression being saved in storage part 400 has been encoded, has represented decoded master image (I by test section 214 Picture, P picture or B picture) local decoder picture signal as with reference to image reference.

Further, test section 214 by search from this reference image and encoded object image (coded object block) is identical or Similar block, detects the position relationship between this coded object block and same or similar piece as difference vector (parallax).

It addition, include the picture of the sub-picture of coded object block and dividing with reference to image (reference picture) as master image It it not the picture generated in identical timing by photographing unit 100a and 100b.Further, test section 214 by this difference vector to compensation Portion 215 and variable-length encoding part 218 export.

It addition, the position relationship between above-mentioned piece is expressed as the deviation of the position of left and right directions by difference vector.

Additionally, in the case of motion vector and difference vector both sides, be restricted to receiving from block cutting part 212b In the case of the block message of the block size of 16 × 16, for the region of the surrounding 32 × 32 of encoded object image, only with 16 × 16 Block size carry out the search of same or similar piece.

On the other hand, in the case of receiving 16 × 16 and 8 × 8 both block size informations, for coded object figure The region of the surrounding of picture 32 × 32, carries out the search of same or similar piece with two Seed-ginger size of 16 × 16 and 8 × 8.

Then, most similar motion vector or difference vector are exported to compensation section 215.

Compensation section 215, when achieving motion vector from test section 214, uses this motion vector and transports with reference to image Dynamic compensation, thus generates prediction signal, this prediction signal is exported.That is, will be contained in reference to the above-mentioned identical or class in image As tuber spatially shift according to above-mentioned motion vector.

On the other hand, compensation section 215, when achieving difference vector from test section 214, uses this difference vector and reference Image carries out parallax compensation, thus generates prediction signal, this prediction signal is exported.

That is, will be contained in spatially moving according to above-mentioned difference vector with reference to the above-mentioned same or similar tuber in image Position.

Switch 204c the master image that encoded object image is I picture, do not carry out infra-frame prediction in the case of, in open frame Between prediction section 213 and adder 221 and between open compensation section 215 and adder 221.

Additionally, switch 204c is the image (master image or sub-picture) of I picture, P picture or B picture at encoded object image In the case of when carrying out infra-frame prediction, infra-frame prediction portion 213 is connected to adder 221, by prediction signal from infra-frame prediction portion 213 transmit to adder 221.

On the other hand, switch 204c is the image (master image or sub-picture) of P picture or B picture at encoded object image In the case of, compensation section 215 is connected to adder 221, prediction signal is transmitted to adder 221 from compensation section 215.

Switch 204b is the feelings of the image (master image or sub-picture) of I picture, P picture or B picture at encoded object image When carrying out infra-frame prediction under condition, infra-frame prediction portion 213 is connected to subtractor 205, by prediction signal from infra-frame prediction portion 213 to Subtractor 205 transmits.

On the other hand, switch 204b is the image (master image or sub-picture) of P picture or B picture at encoded object image In the case of, compensation section 215 is connected to subtractor 205, prediction signal is transmitted to subtractor 205 from compensation section 215.

Additionally, be the image (master image or sub-picture) of P picture or B picture at encoded object image, have infra-frame prediction portion 213 and compensation section 215 both sides prediction signal in the case of, pre-by the prediction signal in infra-frame prediction portion 213 and compensation section 215 Certain the more similar prediction signal surveyed in signal is transmitted to subtractor 205.

Fig. 4, Fig. 5, Fig. 6, Fig. 7 and Fig. 8 are the flow charts of the action of the picture coding device 200 representing present embodiment.

The picture that picture coding device 200 will be contained in the stereoscopic image signals of parallax encodes successively.

Picture coding device 200, when coding, first, presents third dimension to make main subject obvious, sets The lens information etc. being used for carrying out making the region blur beyond main subject is arbitrary for carrying out the fuzzy parameter obscured (S10).

Picture coding device 200, after the setting of fuzzy parameter completes, carries out the acquirement (S20) of picture signal, and parallax is examined Survey portion 201 carries out the detection (S30) of the parallax between acquired two image.

Picture coding device 200 is selected as the image (S40) of the object of coding.In the present embodiment, due to master Image and sub-picture both sides add fuzzy and encode, so selecting master image and sub-picture both sides.

Fuzzy Processing portion 202, according to the fuzzy parameter obtained by step S10 and the parallax information obtained in S30, calculates The fuzzy quantity to add stereoscopic image signals, carries out the Fuzzy Processing to the fuzzy quantity that stereoscopic image signals additional calculations goes out (S50).

Picture coding device 200 enters for the stereoscopic image signals constituting the broad image having been added fuzzy quantity by these Row coding, generates coded object signal (S60).

Here, illustrate to generate a series of action of the S60 of coded object signal.First, it determines encoded object image is No is master image (S61).Here, the control portion possessed in picture coding device 200 differentiate.

Here, if differentiating that encoded object image is master image (being) in S61, the most above-mentioned control portion discriminates whether further Tackle this encoded object image and carry out motion compensated predictive coding (S62).Such as, picture coding device 200 is by coded object When image encodes as the block of P picture or B picture, it determines for motion compensated predictive coding (in S62 being) should be carried out, will When encoded object image encodes as the block of I picture, it determines for motion compensated predictive coding (in S62 no) should not be carried out.

If control portion is determined as carrying out motion compensated predictive coding (in S62 being), then switch 204a is controlled And subtractor 205 is connected to transformation component 216, switch 204b is controlled and compensation section 215 is connected to subtractor 205, Switch 204c is controlled and will be connected between compensation section 215 with adder 221.

Then, block size determination section 210b, according to the fuzzy quantity of this encoded object image, determines to carry out when motion detects Block size (S64b).First, obtain the fuzzy quantity in encoded object image region from Fuzzy Processing portion 202, with threshold ratio relatively (S211).If it is determined that bigger than threshold value, then export the block list by test section 214 searching motion vector to block cutting part 212b Position is limited to the block size information (S212) of the bigger block size of 16 × 16.On the contrary, if it is determined that less than threshold value, then to block Cutting part 212b output also block size information including the less block size of 8 × 8 (S213) in addition to 16 × 16.

Block cutting part 212b, based on the block size information exported from block size determination section 210b, will export from selector 203 Encoded object image carry out splitting (S65b) according to the size of block size information.

Then, test section 214 is for this encoded object image, identical with the block size search accepted from block cutting part 212b Or similar block, detect motion vector.Now, most similar motion vector (S661) is exported.

In motion vector, in the feelings of the block message receiving the block size being restricted to 16 × 16 from block cutting part 212b Under condition, for the region of the surrounding 32 × 32 of encoded object image, only carry out same or similar piece with the block size of 16 × 16 Search.

On the other hand, in the case of receiving the two Seed-ginger size information of 16 × 16 and 8 × 8, for coded object figure The region of the surrounding of picture 32 × 32, carries out the search of same or similar piece with two Seed-ginger size of 16 × 16 and 8 × 8.Then, Most similar motion vector is exported to compensation section 215.

And then, compensation section 215 uses this motion vector detected to carry out motion compensation, generates for encoded object image The prediction signal (S662) of block.

Additionally, subtractor 205 generates prediction by mistake by deducting the image represented by prediction signal from encoded object image Difference signal (S67b).

On the other hand, control portion encoded object image be I picture, to be determined as should not carrying out motion compensation in S406 pre- When surveying coding and be determined as should carrying out intraframe predictive coding, or it is P picture or B picture at encoded object image, is determined as Should carry out during intraframe predictive coding (S63 is), switch 204a is controlled and subtractor 205 is connected to transformation component 216, To switch 204b control and intraframe coding portion 213 is connected to subtractor 205, to switch 204c be controlled and by intraframe coding It is connected between portion 213 with adder 221.

Then, block size determination section 210a is according to the fuzzy quantity of this encoded object image, when determining to carry out infra-frame prediction Block size (S64a).First, the fuzzy quantity in encoded object image region is obtained from Fuzzy Processing portion 202, it may be judged whether compare threshold value Greatly (S201).If it is determined that bigger than threshold value, then will be carried out block search by infra-frame prediction portion 213 to block cutting part 212a output Block unit is limited to the block size information (S202) of the bigger block size of 16 × 16.On the contrary, if it is determined that less than threshold value, then To block cutting part 212a output also block size information including the less block size of 4 × 4 in addition to 16 × 16 (S203).

Block cutting part 212a, based on the block size information exported from block size determination section 210a, will export from selector 203 Encoded object image carry out splitting (S65a) according to the size of block size information.

In infra-frame prediction, in the feelings of the block message receiving the block size being restricted to 16 × 16 from block cutting part 212a Under condition, during for observing from the position of encoded object image be left, upper, upper right, upper left amount to 4 positions, only with 16 × 16 Block size carry out the search of same or similar piece.

On the other hand, in the case of receiving the two Seed-ginger size information of 16 × 16 and 4 × 4, for 4 positions, with Two Seed-ginger size of 16 × 16 and 4 × 4 carry out the search of same or similar piece.

Then, by most similar prediction signal to switch 204b and 204c output (S66).

Additionally, subtractor 205 by deducting the image represented by prediction signal from encoded object image, generate forecast error Signal (S67a).

On the other hand, if it is sub-picture (no in S61) that control portion is determined as encoded object image, then switch 204a is entered Row controls subtractor 205 is connected to transformation component 216, is controlled switch 204b and compensation section 215 is connected with subtractor 205 Connect, switch 204c is controlled starving and will be connected between compensation section 215 with adder 221.

Then, block size determination section 210b, according to the fuzzy quantity of this encoded object image, determines to carry out when parallax detects Block size (S64c).First, the fuzzy quantity in encoded object image region is obtained from Fuzzy Processing portion 202, it may be judged whether compare threshold value Greatly (S221).If it is determined that bigger than threshold value, then will be searched for the block of difference vector by test section 214 to block cutting part 212b output Unit is limited to the block size information (S222) of the bigger block size of 16 × 16.

On the contrary, if it is determined that less than threshold value, then in addition to 16 × 16, also include 8 × 8 to block cutting part 212b output Less block size at interior block size information (S223).

Block cutting part 212b, based on the block size information exported from block size determination section 210b, will export from selector 203 Encoded object image carry out splitting (S65c) according to the size of block size information.

Then, test section 214 is identical with the block size search accepted from block cutting part 212b for this encoded object image Or similar block, detect difference vector.Now, most similar difference vector (S663) is exported.

In difference vector, in the feelings of the block message receiving the block size being restricted to 16 × 16 from block cutting part 212b Under condition, for the region of the surrounding 32 × 32 of encoded object image, only carry out same or similar piece with the block size of 16 × 16 Search.

On the other hand, in the case of receiving the two Seed-ginger size information of 16 × 16 and 8 × 8, for coded object figure The region of the surrounding of picture 32 × 32, carries out the search of same or similar piece with two Seed-ginger size of 16 × 16 and 8 × 8.Then, Most similar difference vector is exported to compensation section 215.

And then, compensation section 215 uses this difference vector detected to carry out parallax compensation, thus generates for coded object The prediction signal (S664) of the block of image.

Additionally, subtractor 205 by deducting the image represented by prediction signal from encoded object image, generate forecast error Signal (S67c).

Then, the predicted residual signal generated is exported (S70) to transformation component 216 by switch 204a.

Here, multiple predicted residual signal is sometimes generated.Such as, it is sub-picture and for P picture at encoded object image In the case of, owing to being sub-picture, so generating predicted residual signal (S67c) based on parallax compensation, owing to being P picture, so Generating predicted residual signal (S67b) based on motion compensation, owing to intraframe coding can be carried out, compiling based in frame so generating The predicted residual signal (S67a) of code is as a result, amount to and generate 3 predicted residual signal.In the case, multiple pre-by generate Survey predicted residual signal minimum among residual signals to transmit to transformation component 216.

On the contrary, predicted residual signal is not the most generated.Such as, at encoded object image it is the I picture of master image In the case of, owing to infra-frame prediction, motion compensated prediction, any one of disparity compensation prediction are not carried out, so a prediction Residual signals does not generates.In the case, not being predicted residual signals and select, switch 204a will export from selector 203 The encoded object image of main image data export to transformation component 216.

Then, transformation component 216 is just being carried out by encoded object image or the predicted residual signal being exported switch 204a Alternation changes and generates coefficient of frequency (S80).

Quantization unit 217 is by the generating quantification value (S90) by this coefficient of frequency quantization, and variable-length encoding part 218 is by this amount Change value carries out Variable Length Code (S100).

So, in the present embodiment, during the block size when coding determines, distance based on parallax information is not only used Information and focus information, and the fuzzy message using the Fuzzy Processing by also using arbitrary fuzzy parameter to add, come Arbitrary fuzzy situation can be carried out the decision of block size.

Thus, in the coding that can determine block size from multiple candidates, the region limits that fuzzy quantity is the biggest is the biggest Block size, it is possible to improve the compression ratio in the bigger region of the less fuzzy quantity of quantity of information.

Additionally, in the present embodiment, it is possible to arbitrary fuzzy situation is carried out the laggard enforcement of decision of variable-block size Coding with infra-frame prediction.

Thus, in the relevant coding based on infra-frame prediction improving compression ratio between the block in utilizing same picture, In the region that fuzzy quantity is set bigger, it is judged that the smooth figure reduced for the radio-frequency component by stronger Fuzzy Processing Picture, and be limited to bigger prediction block sizes and be predicted, thus reduce the reference that the prediction residual in infra-frame prediction is less The search treating capacity of block, it is possible to carry out high speed and the energy-saving technology processed.Additionally, contrary, fuzzy quantity is less or unambiguous district In territory, less block is also included within interior block and is predicted by utilization, in order to even if it is the most no problem to there is complicated image.This Sample, smooth image and the image of complexity can be predicted expeditiously.

And then, in the present embodiment, it is possible to after arbitrary fuzzy situation is carried out the decision of variable-block size, make With based on motion compensated prediction or the coding of the inter prediction of disparity compensation prediction.

Thus, in utilizing the relevant coding based on inter prediction improving compression ratio between the block of multiple interframe, Fuzzy quantity is set in bigger region and is judged as the smooth image that radio-frequency component reduces by stronger Fuzzy Processing, And be predicted with bigger block unit, thus the search reducing the less reference block of the prediction residual in inter prediction processes Amount, it is possible to carry out high speed and the energy-saving technology processed.On the contrary, fuzzy quantity is less or in unambiguous region, utilizes less Block is also included within interior block and is predicted, in order to even if it is the most no problem to there is complicated image.So, smooth image and multiple Miscellaneous image can be predicted expeditiously.And then, during the search of the reference block in inter prediction processes, multiple pictures are deposited Store up in memorizer and it is carried out reference, so by the prediction being limited to bigger block unit, it is possible to reduce data access Number of times or the data amount of transferring, it is possible to carry out bandwidth of memory savingization.

(variation 1)

In the above-described embodiment, picture coding device 200 generates parallax information, to master map according to master image and sub-picture Picture and sub-picture are additional fuzzy respectively and are encoded but it also may obtain each pixel of single image and this image The range information of unit, adds fuzzy and encodes.

Represent the picture coding device 500 of this variation in fig .9.In fig .9, for the element identical with Fig. 1 Use identical symbol, omit the description.

In fig .9, photographs coding system obtains the distance letter of subject and system while subject being photographed Breath, imparts fuzzy image signal encoding by using range information.Such photographs coding system possesses this variation Picture coding device 500, photographing unit 100, distance acquisition unit 110 and fuzzy parameter input unit 300.

Distance acquisition unit 110 measures photographing unit and the subject of each pixel from the image of photographing unit 100 input Distance.Distance detection by send ultrasound wave or millimeter wave etc., measure from subject echo arrive till time And phase contrast etc. is carried out.

Picture coding device 500 uses from the range information of distance acquisition unit 110 output exporting from photographing unit 100 Picture signal gives fuzzy and encodes, and thus generates coding image signal and exports.

Such picture coding device 500 possesses image acquiring section 206, parallax obtaining section 501, Fuzzy Processing portion 202, choosing Select device 203, switch 204a, 204b and 204c, subtractor 205, block size determination section 210a and 210b, block cutting part 212a and 212b, infra-frame prediction portion 213, test section 214, compensation section 215, transformation component 216, quantization unit 217, variable-length encoding part 218, inverse Quantization unit 219, inverse transformation portion 220 and adder 221.

Parallax obtaining section 501 does not include parallax test section 201, obtains parallax or range information from distance acquisition unit 110, Parallax or range information is exported to Fuzzy Processing portion 202.In the present embodiment, parallax obtaining section 501 is from distance acquisition unit 110 obtain the photographing unit of each pixel from the image of photographing unit 100 input and the distance of subject.

The action of picture coding device 500 uses the range information of input to obscure except not carrying out the detection of parallax The calculating of amount and fuzzy and coding to as if single image beyond, as with above-mentioned embodiment being, so former state Use the flow chart of Fig. 4, only difference is illustrated.

Image acquiring section 206 obtains picture signal (S20) from photographing unit 100.The picture signal accepted is single viewpoint Signal of video signal.

Parallax obtaining section 501 obtains the range information of each pixel unit of picture signal from distance acquisition unit 110, to Fuzzy Processing portion 202 exports (S30).

Picture coding device 500 selects picture signal that image acquiring section 206 accepts as encoded object image (S40).

Fuzzy Processing portion 202 uses the range information of each pixel unit of picture signal to carry out the calculating of fuzzy quantity, right Encoded object image adds fuzzy additional (S50).Now, due to Fuzzy Processing portion 202 from parallax obtaining section 501 obtained away from From information, so not carrying out the process using fuzzy parameter according to disparity computation range information, and from parallax obtaining section 501 former state Range information is used to calculate fuzzy quantity.

About the process of following coding, in addition to the single picture coding that there will be no sub-picture, with S60～ The action of S100 is same, so omitting the description.

So, in this variation, it is possible to the distance to subject is known single image, use and appointed by utilization The fuzzy message that the fuzzy parameter of meaning and the Fuzzy Processing of distance are added selects block size and encodes.

It addition, in this variation, parallax obtaining section 501 obtains range information from fuzzy from distance acquisition unit 110 Reason portion 202 exports, but parallax obtaining section 501 can also accept the input of the parallax value such as anaglyph itself.In the case, depending on Difference obtaining section 501 both acquired parallax can be exported to Fuzzy Processing portion 202, it is possible to use fuzzy parameter be transformed to away from From information, range information is exported to Fuzzy Processing portion 202.Additionally, same, parallax obtaining section 501 can also obtain distance letter Breath, uses fuzzy parameter that range information is transformed to parallax, is exported to Fuzzy Processing portion 202 by parallax.

(variation 2)

In the above-described embodiment, the picture coding device 200 stereoscopic image signals to being made up of master image and sub-picture Add fuzzy and encoded but it also may only obscure the side in master image or sub-picture is additional and encodes.

In the case, such as comprise in fuzzy parameter and represent additional fuzzy to which in master image and sub-picture The information encoded, based on this information selected as the image of object of coding in S40 such that it is able to only to as coding The image of object carries out Fuzzy Processing and coded treatment.Or, only to master image additional obscure encode in the case of, it is possible to The picture signal of photographing unit 100b is not inputted to image acquiring section 206 or Fuzzy Processing portion 202, and only to parallax test section 201 inputs.

In this variation, to obscure the image encoded be single, so Fuzzy Processing portion 202 is only to work due to additional Image for the object of coding carries out Fuzzy Processing, is exported to selector 203 by single blurs image data.Fuzzy Processing and As the action of coding with above-mentioned variation 1 is, so omitting the description.

(embodiment 2)

Figure 10 is the module map of the picture coding device 600 of embodiments of the present invention 2.In Fig. 10, for Fig. 1 phase With element use identical symbol, omit the description.

The picture coding device 600 of present embodiment is characterised by, big arbitrary fuzzy situation is carried out variable-block After little decision, carrying out employing the coding of orthogonal transformation, this orthogonal transformation carries out the conversion to frequency content.

In Fig. 10, stereophotography coding system by can photographic subject in the way of stereopsis, will be generated by photography Stereoscopic image signals coding.Such stereophotography coding system possesses the picture coding device 600 of embodiment 2, accepts There are photographing unit 100a and 100b and the fuzzy parameter input unit 300 of the master image of parallax and the input of sub-picture.

Picture coding device 600 uses the parallax from parallax test section 201 output to export from photographing unit 100a and 100b Stereoscopic image signals additional fuzzy and encode, thus generate coded image signal and export.Additionally, picture coding device 600 When two signal of video signal in will be contained in stereoscopic image signals encode, as rest image or moving image encoding.

Such picture coding device 600 possesses image acquiring section 206, parallax obtaining section 207, Fuzzy Processing portion 202, choosing Select device 203, block size determination section 610, block cutting part 611, transformation component 216, quantization unit 217 and variable-length encoding part 218.

It addition, in the present embodiment, as coded system, use and use the wavelet transform as orthogonal transformation Rest image form JPEG2000, motion image formats MotionJPEG2000.

Block size determination section 610, for encoded object image, obtains encoded object image region from Fuzzy Processing portion 202 Fuzzy quantity, with threshold ratio relatively.

In the present embodiment, block size uses the one in 16 × 16,8 × 8 both, and threshold value is set to 8 pixels.

In the case of fuzzy quantity is bigger than threshold value, exports will be carried out orthogonal transformation by transformation component 216 to block cutting part 611 Block unit is limited to the block size information of the bigger block size of 16 × 16.

On the contrary, in the case of less than threshold value, to the block of the less block size that block cutting part 611 export-restriction is 8 × 8 Size information.

Block cutting part 611 is based on the block size information exported from block size determination section 610, by export from selector 203 Encoded object image is split according to the size of block size information.

In the present embodiment, in the case of being limited to the block size information of 16 × 16, only the size with 16 × 16 will Encoded object image is split, and is exported to transformation component 216 by segmentation image.

On the other hand, in the case of receiving the block size information being limited to 8 × 8, only the size with 8 × 8 will coding Object images is split, and is exported to transformation component 216 by segmentation image.

Figure 11, Figure 12 are the flow charts of the action of the picture coding device 600 representing present embodiment.

In fig. 11, identical symbol is used for the action identical with Fig. 4, omits the description.

Block size determination section 610 determines to carry out block size during orthogonal transformation according to the fuzzy quantity of this encoded object image Determine (S310).First, the fuzzy quantity in encoded object image region is obtained from Fuzzy Processing portion 202, it may be judged whether bigger than threshold value (S311).If it is determined that bigger than threshold value, then will carry out orthogonal by transformation component 216 to the output of block size information block cutting part 611 The block unit of conversion is limited to the bigger block size (S312) of 16 × 16.

On the contrary, if it is determined that less than threshold value, then to the less block size that block cutting part 611 export-restriction is 8 × 8 Block size information (S313).

Block cutting part 611 is based on the block size information exported from block size determination section 610, by export from selector 203 Encoded object image carries out splitting (S320) according to the size of block size information.

So, in the present embodiment, in the decision of the block size of the orthogonal transformation being used for when encoding, use logical Cross the fuzzy message that make use of the Fuzzy Processing of arbitrary fuzzy parameter to add.Thereby, it is possible to arbitrary fuzzy situation is carried out The decision of the block size of orthogonal transformation.

Additionally, in the region that fuzzy quantity is bigger, owing to reducing radio-frequency component by stronger Fuzzy Processing, thus logical Cross and block size is limited bigger, it is possible to realize high compression.

It addition, for the picture coding device 600 of embodiment 2, it is also possible to carry out the variation 1 with embodiment 1 or The deformation that variation 2 is same.The action of variation except in the variation of embodiment 1 explanation S10～40 difference, with And the object obscuring and encoding is beyond single picture signal, as with embodiment 2 being.

(embodiment 3)

Figure 13 is the module map of the picture coding device 700 of embodiments of the present invention 3.In fig. 13, for Fig. 1, Element identical for Fig. 9 uses identical symbol and omits the description.

The picture coding device 700 of present embodiment is characterised by, according to an image with corresponding to the quilt of this image Take the photograph the range information of body, generate can two images of stereopsis, the range information using subject additional arbitrary fuzzy and Encode.

In fig. 13, stereoscopic image coding system, while photographic subject, obtains the range information of subject, based on Range information, is generated stereoscopic image signals from the image generated by photography and encodes.Such stereoscopic image coding system has The picture coding device 700 of standby embodiment 3, accept the photographing unit 100 of the input of image, distance acquisition unit 110, fuzzy ginseng Number input unit 300 and storage part 400.

Picture coding device 700 possesses parallax obtaining section 501, image acquiring section 701, Fuzzy Processing portion 202, selector 203, switch 204a, 204b and 204c, subtractor 205, block size determination section 210a and 210b, block cutting part 212a and 212b, Infra-frame prediction portion 213, test section 214, compensation section 215, transformation component 216, quantization unit 217, variable-length encoding part 218, re-quantization Portion 219, inverse transformation portion 220 and adder 221.In addition, image acquiring section 206 possesses stereo-picture generating unit 701.

Fuzzy parameter input unit 300 is by the user interface that stereo-picture generates the input of parameter and fuzzy parameter, should Stereo-picture generates parameter and is used for generating stereo-picture according to the image photographed by photographing unit 100, and this fuzzy parameter is for institute The stereoscopic image signals generated is added.

The parameter of input is as described below.Parameter is generated, except pixel size m and Jiao of photographing unit 100 as stereo-picture Point distance f_pIn addition, base length l, the parallax adjustment amount S between two photographing units imagined to shoot stereo-picture is also inputted_r With weight coefficient W_r.As fuzzy parameter, the fuzzy adjustment amount S of the position of input subject front or behind_n、S_fAnd quilt Take the photograph each weight coefficient W at body front and rear_n、W_f, focal position information P of subject and reality photographic lens information not Same lens information, the most imaginary focal length f_vWith f-number (F value).

It addition, parallax adjustment amount S_rWith weight coefficient W_rIt is to make the depth distance between the distance of subject entirety and subject The coefficient of change, produce following effect: value is the biggest, subject be closer between photographing unit and subject away from photographing unit The difference of distance is the least, is worth the least then subject the biggest further away from the difference of the distance away from photographing unit between photographing unit and subject. Additionally, by parallax adjustment amount S_rIt is set as 0 and by weight coefficient W_rIn the case of being set as 1, generate and be envisioned for from photographing unit The sub-picture of the position photography of the 100 base length l that have left photographing unit.

Accept the setting of these parameters, stereo-picture is generated parameter and exports to stereo-picture generating unit 701, by fuzzy ginseng Number exports to Fuzzy Processing portion 202.

Image acquiring section 701 obtains image from photographing unit 100, generates stereoscopic image signals, defeated to Fuzzy Processing portion 202 Go out.Image acquiring section 701 includes the stereo-picture generating unit 702 generating stereoscopic image signals.

<stereo-picture generating unit 702>

Stereo-picture generating unit 702 is based on the image inputted from photographing unit 100 and the distance from parallax obtaining section 501 input Information, generates by left eye image and the stereoscopic image signals of right eye image construction.

The method generating stereo-picture is illustrated.As in embodiment 1 explanation, if set parallax information as D, then subject is such as numerical expression 2(Fig. 2 (a) with the relation of distance D of photographing unit), so parallax information d is

［ numerical expression 4 ］

d = \frac{f_{p} \cdot 1}{m \cdot D}

By whole subjects is carried out such calculating, obtain the parallax of each subject in stereo-picture Value.

In the present embodiment, the picture original obtained from photographing unit 100 by image acquiring section 701 is as master image, logical Cross and make each pixel-shift of master image utilize

［ numerical expression 5 ］

d_r=W_r·d+S_r

The d obtained_rAnd generate sub-picture.

It addition, according to d_rDistribution, have each pixel on sub-picture corresponding to the multiple pixels in master image situation, with And there is not the situation of pixel in the master image of correspondence.Pixel on sub-picture is corresponding to the multiple pixels in master image In the case of, if corresponding closest to the pixel of photographing unit with among corresponding pixel, then become natural image.On the contrary, close The pixel of the non-existent blank of pixel in corresponding master image, by the data of the neighboring pixel of this pixel are replicated and incite somebody to action Blank filling, it is possible to prevent from exporting factitious sub-picture.

The master image generated and sub-picture are exported by stereo-picture generating unit 702 to Fuzzy Processing portion 202.

Figure 14 is the flow chart of the action of the picture coding device 700 representing present embodiment.

In fig. 14, identical symbol used for the action identical with Fig. 4 and omit the description.

Image processing apparatus 700, after the setting of fuzzy parameter completes, carries out the acquirement (S20) of picture signal, obtain with The range information (S410) that each pixel of acquired image is relevant.

Stereo-picture generating unit 702 is according to the fuzzy parameter obtained in S10 and the distance letter obtained in step S410 Breath, calculates and should be present in the parallax amount in stereoscopic image signals.The figure that stereo-picture generating unit 702 will obtain in step S20 As master image, the stereo-picture carrying out generating the sub-picture to the parallax amount that master image additional calculations goes out generates (S420). The stereo-picture generated is exported by stereo-picture generating unit 702 to Fuzzy Processing portion 202.

So, in the present embodiment, according to an image and the range information of the subject corresponding to this image, generate Can two images of stereopsis, use the range information of subject additional arbitrary fuzzy and encode.Thereby, it is possible to root Fuzzy encoded stereoscopic signal of video signal is addition of with the range information generation of subject according to the image of a viewpoint.

(embodiment 4)

Figure 15 is the module map of the picture coding device 710 of embodiments of the present invention 4.In fig .15, for Figure 13 Identical element uses identical symbol and omits the description.

The picture coding device 710 of present embodiment is characterised by, by the scaling of photographing unit 120 to two focuses away from From switching over, thus obtain the range information between imaging apparatus and subject.

In fig .15, stereoscopic image coding system obtains different two images of focal length, based on two images regards Difference information calculates the distance of subject, based on the range information calculated, generates stereoscopic image signals from an image of photography And encode.Such stereoscopic image coding system possesses the picture coding device 710 of embodiment 4, has zoom function and Jiao The point photographing unit 120 of distance variable, fuzzy parameter input unit 300 and storage part 400.

Picture coding device 710 possesses distance test section 704 in parallax obtaining section 703, and distance test section 704 is connected to In photographing unit 120, fuzzy parameter input unit 300.

Fuzzy parameter input unit 300 accepts the focal length f for generating image_pWith the focus for carrying out range determination Distance f_p'(f_p'>f_p) input.

Two focal length information are exported by fuzzy parameter input unit 300 to distance test section 704.

Distance test section 704 accepts focal length f from fuzzy parameter input unit 300_pWith focal length f_p', for photographing unit 120 alternately set focal length f according to each frame_pWith focal length f_p', measure from photographing unit 120 input image each Pixel, photographing unit 120 and subject distance.Right by the detection different two kinds of images of focal length of the detection of distance The parallax information of the pixel answered, the measurement principle of use triangle measurement are carried out.The distance letter that distance test section 704 will determine Breath exports to stereo-picture generating unit 702 and Fuzzy Processing portion 202.

Use Figure 15 illustrates.At focal length f_pTime corresponding to pixel and the optical axis on the imaging apparatus of subject Distance y_p, and the focal length f from imaging apparatus to subject on the basis of optical axis_pTime orientation θ_pRelation be

［ numerical expression 6 ］

\tan θ_{p} = \frac{y_{p}}{f_{p}}

At focal length f_p'Time corresponding to distance y of pixel and optical axis on the imaging apparatus of subject_p', with optical axis On the basis of from imaging apparatus to the focal length f of subject_p'Time orientation θ_p'Relation be equally

［ numerical expression 7 ］

{\tan θ}_{p^{'}} = \frac{y_{p^{'}}}{f_{p^{'}}}

On the other hand, if setting the distance of optical axis and subject as y, focal length f_p'Time from focus to subject away from From for z, then θ_pWith the relation of y, z it is

［ numerical expression 8 ］

{\tan θ}_{p} = \frac{y}{(f_{p^{'}} - f_{p}) + z}

Equally, θ_p'With the relation of y, z it is

［ numerical expression 9 ］

{\tan θ}_{p^{'}} = \frac{y}{z}

According to numerical expression 6～numerical expression 9, distance Z from subject to imaging apparatus is

［ numerical expression 10 ］

Z = f_{p^{'}} + z = \frac{f_{p} \cdot f_{p^{'}} \cdot (y_{p^{'}} - y_{p})}{f_{p} \cdot y_{p^{'}} - f_{p^{'}} \cdot y_{p}}

Position y by each subject detection pixel_p、y_p', according to the position y of the pixel obtained_p、y_p'And use numerical expression 10 such that it is able to obtain distance Z between above-mentioned subject and imaging apparatus.It addition, the position y of pixel_p、y_p'Detection such as Method as can detecting the method for parallax by utilizing Block Matching Algorithm with the parallax test section 201 of embodiment 1 is carried out.

Distance Z-direction stereo-picture generating unit 702 between the subject and the imaging apparatus that this are obtained exports.

<stereo-picture generating unit 702>

Stereo-picture generating unit 702 based on the image inputted from photographing unit 120 and from parallax obtaining section 703 input away from From information, generate by left eye image and the stereoscopic image signals of right eye image construction.About action, with embodiment 3 it is Same, so omitting the description.

The focal length of photographing unit 120, after the setting of fuzzy parameter completes, is set as f by image processing apparatus 710_p, Carry out the acquirement (S20) of picture signal, then, the focal length of photographing unit 120 is set as f by carrying out_p'Picture signal Obtain and the detection of parallax of acquired 2 images, carry out the detection of the range information of each subject based on parallax value (S410).

So, in the present embodiment, another image obtained according to an image and change focal length, detection The range information of subject, uses the range information of subject detected, generate can two images of stereopsis, additional appoint That anticipates obscures and encodes.There is zoom function and the variable photographing unit of focal length according to Jiao thereby, it is possible to use Point addition of fuzzy encoded stereoscopic signal of video signal apart from two images generations of 1 different viewpoints.

(embodiment 5)

In the present embodiment, the system possessing picture coding device for the present invention is illustrated.

Figure 17 is the module map of the structure of the image encoding system representing present embodiment.In fig. 17, for Fig. 1 phase With element use identical symbol omitting the description.

Image encoding system possesses picture coding device 200, storage part 400, photographing unit 100a, photographing unit 100b, display Portion 800 and external record device 900.

Photographing unit 100a possesses lens 101a, imaging apparatus 102a, master control part 103a and lens control portion 104a.Shooting Element 102a is such as by CCD(Charge Coupled Device) constitute, obtain optical signal via lens 101a, this light is believed Number it is transformed to the signal of telecommunication, exports to master control part 103a.Lens control portion 104a, according to the control from master control part 103a, adjusts The focus etc. of whole lens 101a.Master control part 103a is such as by IC(Integrated Circuit) constitute, obtain from shooting unit The signal of telecommunication of part 102a output, exports this signal of telecommunication to picture coding device 200 as signal of video signal.And then, master control part 103a, by controlling imaging apparatus 102a and lens control portion 103a, carries out shutter speed, Gain tuning and focus adjustment etc..

Photographing unit 100b possesses lens 101b, imaging apparatus 102b, master control part 103b and lens control portion 104b.Photograph Above-mentioned each element that these elements that machine 100b possesses possess with photographing unit 100a respectively is identical.

It addition, master control part 103a and 103b coordinate action respectively so that the focus of photographing unit 100a and 100b and Shutter speeds etc. are identical.

Display part 800 such as possesses liquid crystal display etc., obtains stereoscopic image signals from picture coding device 200, display Master image represented by this stereoscopic image signals and sub-picture.

External record device 900 is configured to, such as, install BD-R(Blu-ray Disc Recordable).Further, outward Portion's recording equipment 900 obtains coded image signal from picture coding device 200, is situated between to the record installed by this coded image signal Matter writes.

(1) in embodiment 1, as the detection method of parallax based on Block Matching Algorithm, illustrate to cut out 16 × 16 pictures The region of element, obtain the summation (SAD) of luminance difference, retrieval cuts out the situation of position, but the present invention is not necessarily limited to this feelings Condition.Such as, cutting out region can also be 8 × 8 pixels, it is possible to use what the aspect ratios such as 8 × 16 pixels were different cuts out region.Or Person, it is also possible to replace the summation (SAD) of luminance difference to use the quadratic sum (SSD of such as luminance difference；Sum of Squared Difference) or normalization cross-correlation (ZNCC；Zero-mean Normalized Cross-Correlation).

Or, the method obtaining the correspondence position between master image and sub-picture may not be Block Matching Algorithm, the most also may be used To use pattern cut method.

(2) in embodiment 1, as the detection method of parallax based on Block Matching Algorithm, illustrate to use luminance difference Situation, but the present invention is not necessarily limited to this situation.Such as, in the case of illustrating pixel value with rgb space, can make By the difference of G-value, represent pixel value with YCbCr space and brightness Y in picture substantially certain in the case of, it is possible to use Cb value Difference or the difference of Cr value.

Or, such as, in the case of illustrating pixel value with rgb space, both the value of R, G, B can be calculated SAD respectively Use their minima or meansigma methods, or weighted mean can also be used.For example, it is also possible to by (SAD+2 × the G of R The SAD of SAD+B)/4 as parallax value, what retrieval parallax value was minimum cuts out position.Equally, picture is being illustrated with YCbCr space In the case of element value, both Y, Cb, Cr can be calculated respectively SAD to use minima, it is possible to use meansigma methods or weighting are flat Average.By in such manner, it is possible to utilize the arbitrary method detection parallax employing all or part of pixel value.

(3) in embodiment 1, illustrate that fuzzy parameter input unit 300 accepts fuzzy ginseng from touch panel and cross key The setting of number, specifies the position of the subject in master image and sets the situation of focal position, but the present invention differing Fixed limit schedules this situation.Such as, fuzzy parameter input unit 300 can also be the arbitrary input equipment such as keyboard or dial.Or Person, if such as photographing unit 100a has automatic focusing function, then fuzzy parameter input unit 300 can also take from photographing unit 100a Obtain focal position and focal position information P.

(4) in embodiment 1, as each weight coefficient W of front and back_n、W_f, fuzzy adjustment amount S_f、S_nIn addition Fuzzy parameter, has enumerated the base length l between focal position information P of subject, photographing unit 100a and 100b, focal length f_p、 Pixel size m, imaginary focal length f_vWith f-number (F value), but the present invention is not necessarily limited to this situation.Such as, as Element size m can also obtain from photographing unit 100a, and base length (l) can also be kept by picture coding device 200 in advance.By this Sample, it is possible to obtain fuzzy required enough fuzzy parameters and use.

Or, picture coding device 200 can also keep the right of multiple distance away from focal position and disperse diameter in advance Answer pattern, the image using certain pattern to carry out Fuzzy Processing is prompted the user with, makes user selects to use which pattern.Logical Cross in such manner, it is possible to select to be suitable for the fuzzy of user and implement.

(5) in embodiment 1, the focal length f for the optical system shot is used as fuzzy parameter_pAnd vacation The focal length f thought_v, and use focal length f_pCalculate distance D of photographing unit and subject, but the present invention might not limit Due to this situation.Such as, in the case of not being to utilize actual distance L but utilize parallax value d to obtain focal position P, The focal length f for the optical system shot can not also be used_pAnd with imaginary focal length f_vAlternative.

When calculating distance D of photographing unit and subject according to parallax value d, if as f_pAlternative and use f_v, then root Distance D calculated according to parallax value ' (hereinafter referred to as imagination distance) become the f of actual distance D_v/f_pTimes, so X and L becomes real The f of actual value_v/f_pTimes.Therefore, corresponding to parallax value d imaginary blur circle diameter σ ' value numerical expression 2, numerical expression 3 respective in The 1st f becoming actual value in the right_p/f_vTimes.Therefore, by advance by f_v/f_pCount weight coefficient W_n、W_fIn, it is possible to obtain with The value of the desired imagination blur circle diameter σ that parallax value d is corresponding.Additionally, imagination distance L of photographing unit and focal position P ' also become F for actual distance L_v/f_pTimes, so the imaginary distance away from photographing unit of whole subjects and focal position P becomes actual The f of distance_v/f_pTimes.Thus, before and after subject and focal position relation, subject each other before and after relation relative to reality Distance become identical relation with imagination distance.

That is, except photographing unit and institute imaginary distance a little it is the f of distance of reality_v/f_pAgain, imagination disperse diameter σ ' It is f apart from interdependent composition_p/f_vBeyond Bei, will not occur to replace f_pAnd use f_vImpact.In addition, as long as it has been described above, By weight coefficient W_n、W_fIt is set to f_v/f_pTimes, it becomes possible to offset the impact on σ ', so in fuzzy matrix and fuzzy quantity, it is possible to Eliminate and replace f_pAnd use f_vImpact.

Because above reason, as f_pAlternative can also use f_v.Thus, it is no longer necessary to use f_p, it is possible to will be fuzzy Parameter reduces one.

(6) in embodiment 1, illustrate to calculate imagination blur circle with numerical expression 2 in the case of more remote than focal position straight Footpath σ, calculating the situation of imagination blur circle diameter σ in the case of nearer than focal position with numerical expression 3, but the present invention might not limit Due to this situation.For example, it is also possible to the blur circle diameter of virtual optical system is used as and is set to

［ numerical expression 11 ］

σ = \frac{{f_{v}}^{2} \cdot | X - L |}{F (L - f_{v}) X}

Can not also carry out approximating and the only amount of being adjusted and weight coefficient additional.

Or, it is also possible to the calculating formula of different σ is used by each distance range of subject.Such as, such as Fig. 2 (c) institute Show, in the case of more remote than focal position, it is also possible to the situation more remote than arbitrary distance X1 and closely in the case of use different The calculating formula of σ.

Or, it is also possible to be not to imagine preferable optical system and additional fuzzy, but use with X as variable arbitrary Function calculates σ.For example, it is also possible to as shown in Figure 2 (d) shows, use two focal positions L1, L2(L1 < L2), at ratio focal position L1 Near situation, the situation more remote than focal position L2 and in the case of near, use the meter of different σ than focal position L1 more than L2 Formula.By in such manner, it is possible to carry out by focus in alignment with apart from different multiple subjects, make to be not present in any focal position On the fuzzy process of whole subjects.

Additionally, can also be such as, if weight coefficient is W_gIf adjustment amount is S_g, for the gain G of camera sensor, It is set to

［ numerical expression 12 ］

σ=W_g·G·|X-L|+S_g

Weight coefficient can also be set as W_iso, set adjustment amount as S_iso, for ISO sensitivity ISO of camera sensor, it is set to

［ numerical expression 13 ］

σ = W_{iso} \cdot \log_{2} \frac{ISO}{100} \cdot | X - L | + S_{iso}

By so, for focal position, it is possible to do not add fuzzy and make it obvious, for the position left from focal position Put, it is possible to be additionally implemented for by because of gain, ISO sensitivity is higher and the noise smoothing that occurs and make it the most eye-catching fuzzy.

In such manner, it is possible to it is additional arbitrary fuzzy.

(7) in embodiment 1, as fuzzy matrix, the filter as within Fig. 3 (a) makes imagination blur circle diameter σ has been enumerated Wave system number is 1, makes the matrix that filter factor is 0 of outside, but the present invention is not necessarily limited to this situation.For example, it is also possible to To be closer to the matrix that center then filter factor is the biggest as shown in Figure 3 (b), it is also possible to be make cornerwise length be imagination more The arbitrary matrixes such as the filter factor of square interior dissipating circular diameter σ is 1, make the filter factor of outside be 0.Or, it is possible to With according to the imaginary blur circle diameter σ calculated, use σ_gThe Gaussian filter of=σ/4

［ numerical expression 14 ］

G (x, y) = \frac{1}{2 π σ_{g}} \exp (- \frac{x^{2} + y^{2}}{2 {σ_{g}}^{2}})

Generator matrix.

These matrixes both can be by user's Freely input, it is also possible to by user from preparing in advance at picture coding device Matrix in 200 selects.By in such manner, it is possible to select to be suitable for the fuzzy of user and implement.

(8) in embodiment 1, illustrate that fuzzy quantity is the situation of the value of imagination blur circle diameter σ, but the present invention is not Necessarily it is defined in this situation.Such as in the case of using such as the fuzzy matrix of Fig. 3 (b), it is also possible to by intensive filter factor The diameter of scope of certain proportion (being set to 68.26% here) as fuzzy quantity.

Additionally, equally in the case of using such as the fuzzy matrix of Fig. 3 (b), it is also possible to the meansigma methods of filter factor is made For threshold value, using the diameter of the scope of value more than this threshold value intensive as fuzzy quantity.Alternatively, it is also possible to threshold value to be set to filtering The median of coefficient or the half of the maximum of filter factor.

Or, in the case of the matrix of the Gaussian filter employing numerical expression 14 expression of above-mentioned variation (7) in utilization, Can also be using (σ/2) as threshold value.

By in such manner, it is possible to according to the property calculation fuzzy quantity of matrix.

(9) in embodiment 1, illustrate that the pixel unit of Fuzzy Processing is the situation of a pixel, but the present invention is also It is not necessarily limited to this situation.For example, it is also possible to be made up of 4 pixels that pixel unit is 2 × 2 pixels.By in such manner, it is possible to Process being suitable for the unit of Fuzzy Processing, be furthermore possible to cut down treating capacity.

(10) in embodiment 1 and embodiment 2, illustrate block size determination section 210a, 210b, 610 threshold value set Be 4 pixels or 8 pixels, the feelings of two Seed-ginger size IOB size information according to 16 × 16 and 4 × 4 or 16 × 16 and 8 × 8 Condition, but the present invention is not necessarily limited to this situation.

Such as, the pixel count of threshold value when cornerwise length of the imaging surface of the sensor as photographing unit is set to H, Allow blur circle a diameter of (H/N) value of 1000～1500 (N be), so blur circle diameter (H/1300) can also be will allow for 2 times of pixel counts being set to threshold value.

In addition it is also possible to by be exceed do not carry out approximating, adjustment amount and the additional blur circle diameter of weight coefficient Maximum (f²/ (FL)) 2 times value, be the arbitrary value of above-mentioned allow blur circle diameter (H/1300) more than 2 times again It is set to the pixel count of threshold value.By in such manner, it is possible to the pixel count of sensor size decision threshold according to photographing unit, IOB is big Little information.

The pixel count of threshold value is not limited to this, it is also possible at random set the pixel count of threshold value by user.

(11) in embodiment 1, illustrate that threshold value is set to 4 pixels, according to 16 × 16 and 4 by block size determination section 210a The situation of the two Seed-ginger size IOB size information of × 4, but the present invention is not necessarily limited to this situation.Such as, can also In the case of enough using the block size of 8 × 8, it is also possible to use 4 pixels and 8 pixel the two as threshold value, according to 16 × 16 With 8 × 8,4 × 43 Seed-ginger size IOB size information.By in such manner, it is possible to make the image-region that fuzzy quantity is big Then make block size the biggest.

Or, in the case of fuzzy quantity is less than threshold value, it is also possible to export-restriction be minimum block size, such as 4 × 4 Block size information.By so, for the region that fuzzy quantity is less, it is no longer necessary to multiple block sizes are implemented infra-frame prediction, energy Enough cut down treating capacity.

(12) in embodiment 1, illustrate that threshold value is set to 8 pixels, according to 16 × 16 and 8 by block size determination section 210b The situation of the two Seed-ginger size IOB size information of × 8, but the present invention is not necessarily limited to this situation.

For example, it is also possible to use 4 pixels as threshold value, for being determined as the block of 8 × 8, it is below 4 pixels in fuzzy quantity In the case of be determined as 4 × 4 sub-block.By in such manner, it is possible to change multistage for block size according to fuzzy quantity.

Or, in the case of fuzzy quantity is less than threshold value, it is also possible to export-restriction be minimum block size, such as 8 × 8 Block size information.By so, for the region that fuzzy quantity is less, it is no longer necessary to multiple block sizes are implemented motion prediction, regards Difference prediction, it is possible to cut down treating capacity.

Additionally, due to the block size of 16 × 8,8 × 16 can also be used in standard H.264, so such as can also Threshold value is set to 8 pixels, and the pixel that fuzzy quantity is bigger than 8 pixels does not use the block of 8 × 8,16 × 8,8 × 16.

And then, such as fuzzy quantity can also be made bigger than 8 pixels by setting the threshold to 8 pixels and 6 pixel the two Pixel does not use the block of 8 × 8,16 × 8,8 × 16, and the pixel making fuzzy quantity bigger than 6 pixels does not use 8 × 8.

(13) in embodiment 1, illustrate that the region of the surrounding 32 × 32 of encoded object image is carried out by test section 214 The situation of the search of same or similar piece, but the present invention is not necessarily limited to this situation.Can also be to bigger region, example As around 64 × 64 region carry out the search of same or similar piece.By in such manner, it is possible to search for more similar block.

Or, it is also possible to for the block in less region such as 8 × 8, the region to around 16 × 16 carries out identical or class As the search of block.By in such manner, it is possible to cut down treating capacity.

(14) in embodiment 1, illustrate that motion compensated prediction and disparity compensation prediction are all inter predictions, carry out phase Same or similar process, so shared block size determination section 210b and block cutting part 210b, test section 214 and compensation section 215 are entered The situation that row processes, but the present invention is not necessarily limited to this situation.For example, it is also possible to by block size determination section 210b in motion In compensation prediction and disparity compensation prediction separately.By in such manner, it is possible to the threshold value of the fuzzy quantity that block size determines will be used for by often Individual process is set as different values, can determine suitable block size for each process.

(15) in embodiment 1, illustrate test section 214 to encoded object image to accept from block cutting part 212b Block size searches for same or similar piece, the situation of detection difference vector, but the present invention is not necessarily limited to this situation.Example As, it is possible to use the parallax information that parallax test section 201 detects calculates difference vector.By in such manner, it is possible to omit block and search Rope processes, it is possible to cut down treating capacity.

(16) in embodiment 1, illustrate that transformation component 216 and inverse transformation portion 220 carry out the situation of DCT, iDCT respectively, But the present invention is not necessarily limited to this situation.Can also be such as that transformation component 216 carries out wavelet transform (DWT； Discrete Wavelet Transform), inverse transformation portion 220 carries out inverse discrete wavelet transform.By in such manner, it is possible to utilize and make By the arbitrary coded system of orthogonal transformation.

(17) in embodiment 1, illustrate that picture coding device 200 is by utilizing disparity compensation prediction and motion compensation H.264MVC the situation carrying out encoding of prediction the two mode, but the present invention is not necessarily limited to this situation.Such as, it is possible to To use the coded system H.264 or not having disparity compensation prediction as MPEG-4.By so, as long as block size does not limits It is set to the coded system of a kind, it becomes possible to use arbitrary coded system.

(18) in the variation 1 of embodiment 1, illustrate that distance acquisition unit 110 uses ultrasound wave or millimeter wave etc. Measure the situation of each pixel, photographing unit and subject distance from the image of photographing unit 100 input, but the present invention is not Necessarily it is defined in this situation.For example, it is also possible to be, distance acquisition unit 110 obtain photographing unit 100 image and with photograph The image of the viewpoint that machine 100 is different, utilizes parallax to obtain distance.Additionally, for example, it is also possible to be, distance acquisition unit 110 uses The distance determination unit of the photographing unit with plural focal length as described in embodiment 4.By so, depending on Difference obtaining section 501 obtains the parallax or range information obtained by arbitrary method, and picture coding device 500 can obscure Process and coding.

(19) in embodiment 2, illustrate that threshold value is set to 8 pixels, according to 16 × 16 and 8 by block size determination section 610 The situation of the two Seed-ginger size IOB size information of × 8, but the present invention is not necessarily limited to this situation.Such as, big at block Can also use the arbitrary sizes such as 4 × 4,32 × 32,8 × 16,64 × 32 in little, threshold value can also be according to the block size used And use arbitrary value.

Or, such as, can also be to use 4 pixels and 8 pixel the two as threshold value, according to 16 × 16 and 8 × 8,4 The 3 Seed-ginger size IOB size information of × 4.By in such manner, it is possible to make the image-region that fuzzy quantity is big make block size The biggest.

(20) in embodiment 2, illustrate if rest image then use JPEG2000 as coded system, if Being that moving image then uses Motion JPEG2000 as the situation of coded system, but the present invention is not necessarily limited to this feelings Condition.As long as allow multiple block size, the coded system of use orthogonal transformation, then can use arbitrary coded system.

(21) in embodiment 3, picture coding device 700 is by the photographing unit 100a of picture coding device 200, photograph Machine 100b replaces with photographing unit 100, parallax obtaining section 207 replaces with parallax obtaining section 501, is replaced by image acquiring section 206 For the structure of image acquiring section 701, but the present invention is not necessarily limited to this situation.Such as, the picture coding dress of embodiment 3 Putting can also be the photographing unit 100a of picture coding device 600, photographing unit 100b to replace with photographing unit 100, by parallax acquirement Portion 207 replaces with parallax obtaining section 501, image acquiring section 206 replaces with the structure of image acquiring section 701.By so, right In the stereoscopic image signals that the range information according to an image with corresponding to the subject of this image generates, it is possible to use Carry out the coding of the orthogonal transformation of conversion to frequency content.

Additionally, same, the picture coding device 710 of embodiment 4 can also be by the photographing unit of picture coding device 600 100a, photographing unit 100b replace with photographing unit 120, parallax obtaining section 207 replace with parallax obtaining section 703, by image acquirement Portion 206 replaces with the structure of image acquiring section 701.

(22) in embodiment 4, the focal length f for distance detection is illustrated_p′Than the focus generated for image Distance f_pLong situation (f_p'>f_p), but the present invention is not necessarily limited to this situation.It can also be the focus for distance detection Distance f_p'Than the focal length f generated for image_pShort situation (f_p'<f_p).By so, if can set with for scheming As the focal length f generated_pDifferent focal length f_p', it becomes possible to carry out distance inspection according to two images that focal length is different Survey.

(23) in embodiment 5, illustrate that image encoding system includes the situation of picture coding device 200, but this Bright it is not necessarily limited to this situation.Such as, image encoding system can also replace picture coding device 200 to include that image is compiled Code device 600.Additionally, such as, image encoding system can also replace photographing unit 100a, 100b and picture coding device 200 and Including the picture coding device 500 being connected with photographing unit 100 and distance acquisition unit 110 or picture coding device 700.Additionally, Such as, image encoding system can also replace photographing unit 100a, 100b and picture coding device 200 to include and photographing unit 120 The picture coding device 710 connected.

(24) in embodiment 5, illustrate that display part 800 obtains stereoscopic image signals from picture coding device 200, shows Show the situation of master image and the sub-picture represented by this stereoscopic image signals, but the present invention is not necessarily limited to this situation.Example As, display part 800 can also obtain local decoder picture signal from picture coding device 200, and display is by this local decoder image Master image that signal represents and sub-picture.By in such manner, it is possible to monitor the picture signal after encoding.

(25) in embodiment 5, illustrate external memory 900 using BD-R as record medium situation, but this Invention is not necessarily limited to this situation.Such as, as storage medium, it is possible to use CD-R(Compact Disc Recordable), MO(Magnet Optical disk), DVD-R(Digital Versatile Disk Recordable), Or the semiconductor memory such as SDHC storage card, SDXC storage card can also be used.

(26) picture coding device of above-mentioned each embodiment can also be typically as integrated circuit i.e. LSI(Large Scale Integration) realize.Both can be using each circuit individually as a chip, it is also possible to include whole circuit Or the circuit of a part and form a chip.Such as, Fuzzy Processing portion 202 both can be integrated into and other electricity as shown in figure 18 In the integrated circuit that road portion is identical, also can also become different integrated circuits.

Here, be recited as LSI, but according to the difference of integrated level, also have referred to as IC(IntegratedCircuit), system LSI, super LSI, the situation of ultra-large LSI.

Additionally, the method for integrated circuit is not limited to LSI, it is also possible to realized by special circuit or general processor. The FPGA(Field ProgrammableGate Array that can program after LSI manufactures can also be utilized), LSI can be reconstructed The connection of internal circuit unit and the reconfigurable processor of setting.

And then, if because the integrated circuit of replacement LSI occur in the progress of semiconductor technology or the other technologies of derivation Technology, then can certainly use this technology to carry out the integrated of functional module.

(27) above-mentioned each embodiment and the explanation of each variation are only the illustrations of the present invention, it is possible to without departing from this Bright scope and carry out various improvement or deformation.

Hereinafter, to about the picture coding device of embodiment and its integrated circuit and the structure of method for encoding images And effect illustrates.

A the picture coding device of () embodiment is characterised by, including: image acquiring section, obtain image；Parallax obtains Portion, obtains at least one party in the parallax of each pixel unit of above-mentioned image or the range information of each pixel unit；Fuzzy Amount determination section, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, determines above-mentioned image The fuzzy quantity of each pixel unit；Fuzzy Processing portion, according to above-mentioned fuzzy quantity, implements Fuzzy Processing to above-mentioned image；Block is big Little determination section, according to above-mentioned fuzzy quantity, determines that from multiple block sizes each image being used for after implementing above-mentioned Fuzzy Processing is cut Go out the block size in coded object region；And encoding section, according to the block size of above-mentioned decision, will implement above-mentioned fuzzy with block unit Picture coding after process.

Additionally, the integrated circuit of embodiment is characterised by, including: image acquiring section, obtain image；Parallax obtains Portion, obtains at least one party in the parallax of each pixel unit of above-mentioned image or the range information of each pixel unit；Fuzzy Amount determination section, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, determines above-mentioned image The fuzzy quantity of each pixel unit；Fuzzy Processing portion, according to above-mentioned fuzzy quantity, implements Fuzzy Processing to above-mentioned image；Block is big Little determination section, according to above-mentioned fuzzy quantity, determines to cut out for the image after implementing above-mentioned Fuzzy Processing from multiple block sizes The block size in coded object region；And encoding section, according to the block size of above-mentioned decision, above-mentioned fuzzy place will be implemented with block unit Picture coding after reason.

Additionally, the method for encoding images of embodiment is characterised by, including: image acquisition step, obtain image；Parallax Acquisition step, obtains at least in the parallax of each pixel unit of above-mentioned image or the range information of each pixel unit Side；Fuzzy quantity deciding step, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, determines The fuzzy quantity of each pixel unit of above-mentioned image；Fuzzy Processing step, according to above-mentioned fuzzy quantity, implements fuzzy to above-mentioned image Process；Block size deciding step, according to above-mentioned fuzzy quantity, determines for after implementing above-mentioned Fuzzy Processing from multiple block sizes Image cut out the block size in coded object region；And coding step, according to the block size of above-mentioned decision, will be real with block unit Execute the picture coding after above-mentioned Fuzzy Processing.

Additionally, the image encoding system of embodiment, including: for the photographing unit of photographs, by above-mentioned picture coding Picture coding device and be used for record by above-mentioned picture coding device coding generate image compression data record be situated between Matter, this image encoding system is characterised by, above-mentioned picture coding device includes: image acquiring section, obtains image；Parallax obtains Portion, obtains at least one party in the parallax of each pixel unit of above-mentioned image or the range information of each pixel unit；Fuzzy Amount determination section, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, determines above-mentioned image The fuzzy quantity of each pixel unit；Fuzzy Processing portion, according to above-mentioned fuzzy quantity, implements Fuzzy Processing to above-mentioned image；Block is big Little determination section, according to above-mentioned fuzzy quantity, determines that from multiple block sizes the image being used for after implementing above-mentioned Fuzzy Processing cuts out The block size in coded object region；And encoding section, according to the block size of above-mentioned decision, above-mentioned fuzzy place will be implemented with block unit Picture coding after reason.

According to these said structures, fuzzy quantity is used to determine block size when by picture coding such that it is able to use and scheme The block size that the fuzzy quantity of each pixel unit of picture is the most suitable.

B () is additionally, the picture coding device of above-mentioned (a) of embodiment can also be that above-mentioned fuzzy quantity is by above-mentioned image The parallax of each pixel unit or the fuzzy parameter of the range information of each pixel unit and above-mentioned image determine.

Thus, parallax information is not only used, additionally it is possible to use such as focus information, the fuzzy quantity etc. of each subject distance Arbitrary fuzzy message carries out multi-Fuzzy process.

C (), additionally, the picture coding device of above-mentioned (b) of embodiment can also be, above-mentioned fuzzy parameter includes and claps The optical characteristic value that the characteristic value of the optical system taking the photograph above-mentioned image is different.

Thereby, it is possible to use fuzzy message to set virtual optical system, it is not restricted to be actually used in the optical system of shooting System and make and addition of the image of scattered scape situation based on arbitrary optical system, or made by Fuzzy Processing be envisioned for by The image of the optical system shooting being difficult to.

D () is additionally, the picture coding device of above-mentioned (c) of embodiment can also be that above-mentioned fuzzy parameter includes one The focal position information of above subject.

Thereby, it is possible to carry out using the Fuzzy Processing of focus information, such as, can also carry out there are multiple focuses etc. and pass through The actual irrealizable Fuzzy Processing of optical system.

(e) additionally, the picture coding device of above-mentioned (a) of embodiment can also be, above-mentioned block size determination section for The pixel unit that above-mentioned fuzzy quantity is bigger than threshold value, being determined as among multiple block size is not the block size of smallest block size.

Thus, in the coded system that can use two or more block sizes, the region that fuzzy quantity is the biggest, will be used Block size be limited to the biggest block size, it is possible to improve the compression ratio in quantity of information is less, fuzzy quantity is bigger region. Additionally, in the case of being predicted coded treatment, it is not necessary to the block size contrasting the block size determined little is predicted block The selection of size, it is possible to realize the further reduction for the treatment of capacity.

(f) additionally, the picture coding device of above-mentioned (a) of embodiment can also be, above-mentioned block size determination section for The pixel unit that above-mentioned fuzzy quantity is less than threshold value, including minimum block size be determined as in multiple block sizes with On block size.

Thus, less block is also included within interior by region that is less in fuzzy quantity or that be not blurred uses Variable-block, it is possible to prevent the reduction of quantity of information.If additionally, being determined as using multiple in the case of being predicted coded treatment Block size, then can use bigger block size to carry out expeditiously in the case of reference block is restricted to bigger block size Predictive coding processes.

G () is additionally, the picture coding device of above-mentioned (e) or above-mentioned (f) of embodiment can also be that above-mentioned encoding section exists In the case of carrying out motion compensated prediction, disparity compensation prediction or infra-frame prediction, use that above-mentioned block size determination section determines is each Block size carries out the search of similar block, selects the block of most similar piece an of image that can obtain and split by above-mentioned block unit Size, generates the residual signals for selected block size.

Thus, in the case of having carried out the process of motion prediction, parallax prediction or infra-frame prediction during coding, energy Enough decision from the variable-block determined is suitable for each block size processed, it is possible to make residual signals minimize, it is possible to carry out High compression.

H (), additionally, the picture coding device of above-mentioned (a) of embodiment can also be, above-mentioned image acquiring section obtains use The left eye image shown in stereo-picture and right eye image；Above-mentioned parallax obtaining section includes detecting above-mentioned left eye image The parallax test section of the parallax in the region that region is corresponding with above-mentioned right eye image.

Thereby, it is possible to generate each pixel unit according to the right eye image shown for stereo-picture and left eye image Parallax information, for generated parallax information can be used by the stereo-picture of right eye image and left eye image construction Carry out Fuzzy Processing and use the coding of suitable block size.

J () is additionally, the picture coding device of above-mentioned (a) of embodiment can also be that above-mentioned image acquiring section includes standing Body image production part, this stereo-picture generating unit according to an acquired image and above-mentioned parallax obtaining section obtain above-mentioned The parallax of image or range information, generate by left eye image and the stereo-picture of right eye image construction.

Thereby, it is possible to generate for stereopsis according to the range information of the subject involved by an image and this image Two images, use the range information of subject carry out Fuzzy Processing and use suitable block size to the stereo-picture generated Coding.

Industrial applicibility

The present invention relates to, by the picture coding device of image signal encoding, particularly to carry out the distance phase with subject Suitable fuzzy the adding and the decision of block size corresponding with fuzzy quantity, the most high efficiency coding and process answered The reduction of amount.

Thus, for processing the image recorder of digital video camera or the digital still camera etc. of signal of video signal It is effective.In addition, also it is effective for image transmission devices and image editing apparatus etc..

Symbol description

100,100a, 100b, 120 photographing units

101a, 101b lens

102a, 102b imaging apparatus

103a, 103b master control part

110 distance acquisition unit

200,500,600,700,710 picture coding device

201 parallax test sections

202 Fuzzy Processing portions

203 selectores

204a, 204b, 204c switch

205 subtractors

206,701 image acquiring section

207,501,703 parallax obtaining section

210a, 210b, 610 block size determination sections

212a, 212b, 611 pieces of cutting parts

213 infra-frame prediction portions

214 test sections

215 compensation section

216 transformation components

217 quantization units

218 variable-length encoding part

219 re-quantization portions

220 inverse transformation portions

221 adders

300 fuzzy parameter input units

400 storage parts

702 stereo-picture generating units

704 distance test sections

800 display parts

900 external memories

Claims

1. a picture coding device, it is characterised in that including:

Image acquiring section, obtains the image shot by photographing unit；

Parallax obtaining section, obtain the parallax of each pixel unit of above-mentioned image or each pixel unit, above-mentioned photographing unit with At least one party of the range information of subject；

Fuzzy quantity determination section, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, determines The fuzzy quantity of each pixel unit of above-mentioned image；

Fuzzy Processing portion, according to above-mentioned fuzzy quantity, implements Fuzzy Processing to above-mentioned image；

Block size determination section, according to above-mentioned fuzzy quantity, determines for after implementing above-mentioned Fuzzy Processing from multiple block sizes Each image cuts out the block size in coded object region；And

Encoding section, according to the block size of above-mentioned decision, the image after implementing above-mentioned Fuzzy Processing encodes with block unit.

2. picture coding device as claimed in claim 1, it is characterised in that

Above-mentioned fuzzy quantity is by the parallax of each pixel unit of above-mentioned image or the range information of each pixel unit and above-mentioned The fuzzy parameter of image determines.

3. picture coding device as claimed in claim 2, it is characterised in that

Above-mentioned fuzzy parameter includes the optical characteristic value different from the characteristic value of the optical system shooting above-mentioned image.

4. picture coding device as claimed in claim 3, it is characterised in that

Above-mentioned fuzzy parameter includes the focal position information of more than one subject.

5. picture coding device as claimed in claim 1, it is characterised in that

Above-mentioned block size determination section, for the above-mentioned fuzzy quantity pixel unit bigger than threshold value, is determined as among multiple block size not It it is the block size of minimum block size.

6. picture coding device as claimed in claim 1, it is characterised in that

Above-mentioned block size determination section, for the above-mentioned fuzzy quantity pixel unit less than threshold value, is determined as including among multiple block size Minimum block size at interior more than one block size.

7. the picture coding device as described in claim 5 or 6, it is characterised in that

Above-mentioned encoding section, in the case of carrying out motion compensated prediction, disparity compensation prediction or infra-frame prediction, uses by above-mentioned piece Each block size that size determination section determines carries out the search of similar block, selects one can obtain and splits by above-mentioned block unit The block size that image is most similar piece, generates the residual signals corresponding to selected block size.

8. picture coding device as claimed in claim 1, it is characterised in that

Above-mentioned image acquiring section obtains the left eye image and right eye image shown for stereo-picture；

Above-mentioned parallax obtaining section includes parallax test section, and this parallax test section detects the region of above-mentioned left eye image and the above-mentioned right side The corresponding interregional parallax of ophthalmically acceptable image.

9. picture coding device as claimed in claim 1, it is characterised in that

Above-mentioned image acquiring section includes stereo-picture generating unit, this stereo-picture generating unit according to an acquired image and by The parallax of the above-mentioned image that above-mentioned parallax obtaining section obtains or range information, generate by left eye image and right eye image construction Stereo-picture.

10. an integrated circuit, it is characterised in that including:

Image acquiring section, obtains the image shot by photographing unit；

Block size determination section, according to above-mentioned fuzzy quantity, determines for after implementing above-mentioned Fuzzy Processing from multiple block sizes Image cuts out the block size in coded object region；And

11. 1 kinds of method for encoding images, it is characterised in that including:

Image acquisition step, obtains the image shot by photographing unit；

Parallax acquisition step, obtains parallax or each pixel unit, above-mentioned photographing unit of each pixel unit of above-mentioned image At least one party with the range information of subject；

Fuzzy quantity deciding step, according to parallax or the range information of above-mentioned each pixel unit of above-mentioned each pixel unit, certainly The fuzzy quantity of each pixel unit of fixed above-mentioned image；

Fuzzy Processing step, according to above-mentioned fuzzy quantity, implements Fuzzy Processing to above-mentioned image；

Block size deciding step, according to above-mentioned fuzzy quantity, determines for after implementing above-mentioned Fuzzy Processing from multiple block sizes Image cut out the block size in coded object region；And

Coding step, according to the block size of above-mentioned decision, the image after implementing above-mentioned Fuzzy Processing encodes with block unit.

12. 1 kinds of image encoding systems, including the photographing unit for shooting image, fill the picture coding of above-mentioned picture coding The record medium of the image compression data put and generated by the coding of above-mentioned picture coding device for record, its feature It is,

Above-mentioned picture coding device includes:

Image acquiring section, obtains image；