CN103517048A

CN103517048A - Adaptive edge enhancement

Info

Publication number: CN103517048A
Application number: CN201210209570.2A
Authority: CN
Inventors: B·A·***
Original assignee: ViXS Systems Inc
Current assignee: ViXS Systems Inc
Priority date: 2012-06-20
Filing date: 2012-06-20
Publication date: 2014-01-15
Anticipated expiration: 2032-06-20
Also published as: CN103517048B

Abstract

The invention relates to adaptive edge enhancement. In at least one embodiment of the invention, equipment used for adaptive edge enhancement of video signals comprises a transient improvement module. The transient improvement module is configured to generate a first adjustment pixel value based on a pixel value window of pixels around a focus pixel which initially has an input pixel value. The equipment comprises an adaptive peaking module which is configured to generate a second adjustment pixel value based on the first adjustment pixel value and the input pixel value. In at least one embodiment of the equipment, the adaptive peaking module comprises a high-pass filter which is configured to generate pixel adjustment based on the first adjustment pixel value. In at least one embodiment of the equipment, the adaptive peaking module also comprises a gain path which is configured to apply at least one adaptive gain value on the pixel adjustment so as to generate an adaptive adjustment value.

Description

Adaptive edge strengthens

Technical field

The application relates to processing system for video, and more particularly, the edge that relates to vision signal strengthens.

Background technology

Usually, image processing system and processing system for video are carried out edge enhancing technique to improve image or hereinafter referred to as the apparent acutance of the video signal frame (that is, frame of video, video image) of " image ".Typical edge enhancement filter (for example, transient state is improved filter) is designed to improve the edge transition (edge transition) of the brightness (luma) of image or frame of video.For example, the sharpen edges border in edge enhancement filter identification frame, the edge between main body and the background of contrastive colours for example, and increase is close to the contrast in peripheral region, edge.But edge enhancement filter is understood overshoot (overshoot) or owed punching (undershoot), thereby produce vision not nature or noise adjustment edge.

Summary of the invention

In at least one embodiment of the present invention, the equipment strengthening for the adaptive edge of vision signal comprises that transient state improves module.Transient state is improved pixel value window that module is configured to concerned pixel based on initially having input pixel value pixel around and is produced first and adjust pixel value.Described equipment comprises the self adaptation peaking module being configured to based on the first adjustment pixel value and input pixel value generation the second adjustment pixel value.In at least one embodiment of described equipment, self adaptation peaking module comprises the high pass filter that is configured to produce based on the first adjustment pixel value pixel adjustment.In at least one embodiment of described equipment, self adaptation peaking module also comprises and is configured to apply at least one adaptive gain value to produce the gain path of self adaptation adjusted value to pixel adjustment.

In at least one embodiment of the present invention, the method strengthening for the edge of vision signal comprises that the transient state that the pixel value window based on concerned pixel pixel around produces concerned pixel adjusts pixel value.Described concerned pixel initially has input pixel value.Described method comprises that high-pass filtering based at least one adaptive gain value, pixel value window and transient state adjust the output that pixel value produces concerned pixel and adjust pixel value.

In at least one embodiment of the present invention, the equipment strengthening for the adaptive edge of vision signal comprises that the chromatic value window of the concerned pixel that is configured to based on having input chromatic value and input brightness value pixel around produces first the first colourity transient state adjusting module of adjusting chromatic value.Described equipment comprises the first control module that is configured to produce based on the first adjustment chromatic value, input chromatic value and input brightness value the output adjustment chromatic value of concerned pixel.

Accompanying drawing explanation

By with reference to accompanying drawing, the present invention may be better understood, and to those skilled in the art, its a large amount of target, feature and advantage will become fairly obvious.

Fig. 1 illustrates the functional block diagram of exemplary video treatment system.

Fig. 2 A and Fig. 2 B illustrate the functional block diagram of the adaptive edge augmented video processing module of various embodiment according to the invention.

Fig. 3 illustrates the functional block diagram that Fig. 2 A of at least one embodiment according to the invention and the transient state of Fig. 2 B are improved module.

The luminance transient that Fig. 4 illustrates Fig. 3 of at least one embodiment according to the invention improves the functional block diagram of module.

Fig. 5 A and Fig. 5 B illustrate the look-up table 402 that passes through respectively Fig. 4 and 410 exemplary functions that realize of at least one embodiment according to the invention.

Fig. 6 illustrates the functional block diagram of Fig. 2 A of at least one embodiment according to the invention and the two-dimensional adaptive peaking module 204 with ring control (ringing control) of Fig. 2 B.

Fig. 7 illustrates the functional block diagram of ring control module 804 of Fig. 6 of at least one embodiment according to the invention.

Fig. 8 illustrates the functional block diagram that Fig. 2 A of at least one embodiment according to the invention and the chromatic component transient state of Fig. 2 B are improved module 206.

The chromatic component transient state that Fig. 9 illustrates Fig. 8 of at least one embodiment according to the invention is improved the functional block diagram of module 504.

Figure 10 A and Figure 10 B illustrate the exemplary functions being realized by soft core module 946 and the look-up table 910 of Fig. 6 respectively of at least one embodiment according to the invention.

Figure 10 C illustrates the control path for generation of weighted factor k of Fig. 9 of at least one embodiment according to the invention.

In different accompanying drawings, with identical Reference numeral representation class like or identical project.

Embodiment

With reference to Fig. 1, exemplary video SOC (system on a chip) (SOC) integrated circuit (for example, video SOC 102) receives one or more video data streams by memory 104 from external source.Video processor 108 and 110 is fetched unpressed video data from memory 104.Video processor 108 and 110 is once only fetched a plurality of parts of the single frame of video data.Video processor 108 and 110 can comprise one or more in following: frame rate converter, display, deinterlacer (deinterlacer), machine for composing, scaler (scaler), edge strengthens module, color operations module, face or object identification module, contrast enhancer, superresolution scaler (superresolution scaler), remove piece device (deblocker), time and/or space noise silencer, the video processing module that other is suitable or their combination arbitrarily.Video processor 108 and 110 by process, unpressed video data again stores memory 104 into for use subsequently, or the video data of processing directly offered to sheet upper bumper, directly offers another video processor or offer sheet external circuit (not shown).

Although decompressor 106 and video processor 108 and 110 are illustrated as directly and Memory Controller 112 couplings, in other the embodiment of SOC, decompressor 106 and video processor 108 and 110 and one or more shared memory bus be coupled.Typical memory 104 is double data rate Synchronous Dynamic Random Access Memory (that is, DDR SDRAM memory or DDR memories).In at least one embodiment, memory 104 is to meet DDR SDRAM memory standards, DDR2SDRAM memory standards, DDR3SDRAM memory standards, other DDR SDRAM standard or the DDR memory of other suitable memory standards.

See figures.1.and.2, at least one embodiment of video system 100, video processor 108 and 110 is image to be carried out to edge enhancing technique to improve the adaptive edge of the apparent acutance of image, strengthen module (for example, self adaptation strengthens module 200).The specific pixel of image is generally for example, by brightness value (Y) and two chromatic value (, C _band C _r) represent.As mentioned here, " pixel value " refers to brightness value or chromatic component value, for example, and for Y, the C of the specific pixel of image _bor C _r.For at least one compound video code system, the brightness in brightness value representative image (that is, the black and white of image or achromaticity part).Carrier chrominance signal and luminance signal transmit the colouring information of image discretely.Usually, carrier chrominance signal is generally represented by two colour-difference composition U=0.492 * (B '-Y) (that is, indigo plant-brightness) and V=0.877 * (R '-Y) (that is, red-brightness).In composite video signal, U and V signal modulates color sub-carrier signal.Result is called as carrier chrominance signal.Phase place and the amplitude of the carrier chrominance signal of modulation are roughly corresponding with the color harmony saturation of color.In at least one embodiment of composite video system, do not use the blueness of vision signal and the absolute value of red composition (for example, B ' and R '), and use colour-difference composition (for example, to be respectively C _band C _r).In at least one embodiment, self adaptation strengthens brightness value and the chromatic value that module 200 strengthens image.But the self adaptation that meets technology described herein strengthens the module value of highlighting, C _bchromatic value, C _rchromatic value or their combination arbitrarily.

With reference to Fig. 2 A and Fig. 2 B, in self adaptation, strengthen at least one embodiment of module 200, transient state is for example improved module (for example, luminance transient improves module 202), for brightness value (, the Y of image _iN) carry out edge enhancing technique, and produce for each brightness value of image brightness value (for example, the z adjusting ₁(x, y)).But luminance transient improves module 202 to be introduced high frequency in the brightness of image.The radio-frequency component of these new introducings of the brightness of additional enhancing module (for example, having the two-dimensional adaptive peaking module 204 that ring is controlled) enhancing image is adjusted brightness value (for example, Y to produce the output of image _oUT).Additional transient state is for example improved module (for example, colourity transient state is improved module 208), for chromatic component (, C _band C _r) in each carry out edge enhancing technique and adjust chromatic component value and (for example, be respectively C to produce output _bOUTand C _rOUT).Therefore usually, the chromatic component of image is because high decrement is noise signal, thereby makes and compare in the corresponding bright of image, is more difficult to edge be detected in the chromatic component of image.Compare independent chromatic component (for example, C with colourity Zhong edge with the brightness of image _b) Zhong edge can not arrive by human eye detection.Because edge enhancing technique can be introduced noise false signal, therefore in self adaptation, strengthen at least one embodiment of module 200, chromatic component transient state is for example improved the edge enhancing technique of module 206, by function (, the Y of the associated luminance value of image _iN) control.For example, when edge being detected for corresponding bright value, only by chromatic component transient state, improve the chromatic component value that module 206 is adjusted image.At adaptive edge, strengthen at least one embodiment of module 200, the edge of chromatic component strengthens the function (Y by the delay form of the corresponding bright value of image _iN') control.

With reference to Fig. 3, at luminance transient, improve at least one embodiment of module 202, pixel value window (for example, brightness value window) is stored in buffer 302.Note, window 302 is asymmetric, and line number is less than columns.7 * 9 windows are only exemplary, and centered by corresponding concerned pixel (that is, current pixel or input pixel) there is different size and/or symmetric other window (for example, 3 * 3,4 * 4,9 * 9) meets technology described herein.Brightness value window comprises the brightness value of concerned pixel those pixels around, and wherein concerned pixel is centered by window.Pixel data can directly be retrieved from memory 104, or can be stored in the frame buffer or other suitable buffer of video SOC 102.

At luminance transient, improve at least one embodiment of module 202, four direction filter module (for example, horizontal edge boostfiltering device module 304, vertical edge boostfiltering device module 306, over against corner edge boostfiltering device module 308He Fu diagonal angle edge enhancement filter module 310) in each by the brightness value window to relevant to concerned pixel, apply the transient state improvement value that respective direction (for example, being respectively level, the negative diagonal angle of vertical, positive diagonal sum) determined in two-dimentional non-line filtering.As mentioned here, " improvement value " or " enhancing value " refers to the adjusted value of the vision that will improve image.

In at least one embodiment, the pixel value that the linear filtering of each brightness value that comprises the window based on from centered by concerned pixel in directional filter module 304,306,308 and 310 produces corresponding smoothing (for example, is respectively x _sh(0:8), x _sv(0:6), x _s+45(0:6), x _s-45(0:6) corresponding smoothing filter) (for example, smoothing filter 314,316,318 and 320).The pixel value of smoothing has the transient noise of minimizing along the edge that will be enhanced.In at least one embodiment, smoothing filter unit is respectively to not emphasizing that those brightness values that use same orientation or pay close attention to the monochrome information of direction apply electric-wave filter matrix.At luminance transient, improve at least one embodiment of module 202, smoothing filter 314,316,318 and 320 is 3 * 3 filters of three collinear value of output.On December 16th, 2009 submit to using Bradley Arthur Wallace and the James C.Lynch U.S. Patent application No.12/639 as inventor, the exemplary filters technology of having described the level of 3 * 3 pixel windows, vertical, positive diagonal sum Fu Dui angular direction in 353, adds this application as a reference at this.These instructions can be extended to have and be positioned at concerned pixel (that is, the x that the 3rd row the 4th of pixel window lists ₃₄) concerned pixel centered by 7 * 9 windows.

At luminance transient, improve at least one embodiment of module 202, smoothing filter 314 is 7 * 9 filters of nine level and smooth collinear value of output, and smoothing filter 316,318 and 320 is 7 * 9 filters of seven level and smooth collinear value of output.For example, if the window of brightness value be

The filtering operation on window w can be calculated as:

x＝∑ _i，jH _i，jw _i，j，

Wherein, H _i,jtwo-dimensional filtering kernel function (filter kernel).For example, because the window of available pixel is restricted (, being limited to 7 pixel * 9 pixels) and filter 314 need to provide 9 conllinear output pixel x _kand filter 316,318 and 320 need to provide 7 conllinear output pixel x _k, therefore for each k value restriction filtering core function.For horizontal alignment, use x _kas transient state, improve the input of piece, k=-4～4.But, due to

x _k＝∑ _i，jH _i，jw _i，j+k，

And, if matrix H _{7 * 9}for all i, j has nonzero term, and for index element j+k, window w must be larger so.For example,, when estimating x ₄time, j+k=8+4=12.Therefore, can use different filtering core functions (for example, the kernel function H for each k value ^k).For specific improvement orientation, along orthogonal direction filtered pixel value.In at least one embodiment, the filter 316 of vertically filtering uses different filtering core functions for each k value.For example,, when estimating x for horizontal direction ₄time use following kernel function H ⁴:

H^{4} = [\begin{matrix} 0 & 0 & 0 & 0 & h_{04} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{14} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{24} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{34} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{44} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{54} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{64} & 0 & 0 & 0 & 0 \end{matrix}] .

At luminance transient, improve at least one embodiment of module 202, along the simplification filtering core function with improving the trend pass filtering pixel value of orientation quadrature, comprise at most three non-zeros and specify (tap).

For filter 314, exemplary filtering core function is:

H^{4} = [\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{24} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{34} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{44} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}],

For filter 316, exemplary filtering core function is:

H^{4} = [\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & h_{33} & h_{34} & h_{35} & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}],

For filter 318, exemplary filtering core function is:

H^{4} = [\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & h_{25} & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{34} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & h_{43} & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}],

For filter 320, exemplary filtering core function is:

H^{4} = [\begin{matrix} 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & h_{23} & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & h_{34} & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & h_{45} & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}] .

Although can use other filter coefficient,, [the h of filter 316 ₂₄h ₃₄h ₄₄], [h of filter 318 ₂₅h ₃₄h ₄₃] and [h of filter 320 ₂₃h ₃₄h ₄₅] exemplary coefficient be [0.25 0.5 0.25].

In at least one embodiment of smoothing filter 314,316,318 or 320, when smoothing filter does not operate on the brightness value of corresponding conllinear input vector, smoothing filter is sent to not level and smooth brightness value and improves filter (for example, transient state is improved filter 324,326,328 or 330) by corresponding transient state subsequently.At luminance transient, improve at least one embodiment of module 202, each in directional filter module 304,306,308 and 310 is configured to based on corresponding smooth pixel value (x _sh(0:8), x _sv(0:6), x _sD+ (0:6), x _sD-(0:6)) nonlinear filtering for the brightness value of the concerned pixel in pixel window buffer 302, produce the transient state adjusted value (z of corresponding specific direction _a).For example, the transient state of horizontal edge boostfiltering device module 304 improve filter 324 based on [x _sh(0), x _sh(1), x _sh(2), x _sh(3), x _sh(4), x _sh(5), x _sh(6), x _sh(7), x _sh(8)] the nonlinear filtering of corresponding [z (4), z (3), z (2), z (1), z (0), z (1), z (2), z (3), z (4)], produces the transient state adjusted value z in horizontal direction _ah; The transient state of vertical edge boostfiltering device module 306 improve filter 326 based on [x _sv(0), x _sv(1), x _sv(2), x _sv(3), x _sv(4), x _sv(5), x _sv(6)] the nonlinear filtering of corresponding [z (3), z (2), z (1), z (0), z (1), z (2), z (3)], the transient state adjusted value z of the nonlinear filtering based on pixel window in generation vertical direction _av; Over against the transient state of corner edge boostfiltering device module 308 improve filter 328 based on [x _sD+(0), x _sD+(1), x _sD+(2), x _sD+(3), x _sD+(4), x _sD+(5), x _d+(6)] the nonlinear filtering of corresponding [z (3), z (2), z (1), z (0), z (1), z (2), z (3)], generation for example, over against the transient state adjusted value z of the nonlinear filtering based on pixel window on angular direction (,+45 degree) _aD+; The transient state of negative diagonal angle edge enhancement filter module 310 improve filter 330 based on [x _sD-(0), x _sD-(1), x _sD-(2), x _sD-(3), x _sD-(4), x _sD-(5), x _sD-(6)] the nonlinear filtering of corresponding [z (3), z (2), z (1), z (0), z (1), z (2), z (3)], produce with Fu Dui angular direction (for example ,-45 degree) over against angular direction quadrature on the transient state adjusted value z of the nonlinear filtering based on pixel window _aD-.

Transient state is improved pixel value z _ah, z _av, z _ad+, z _aD-in each represent the sharpening degree enhanced form in the corresponding direction of concerned pixel in level, vertical, positive diagonal sum Fu Dui angular direction.Summation module 334,336,338 and 340 improves pixel value z by corresponding transient state _ah, z _av, z _ad+, z _aD-with concerned pixel about filtered corresponding filtered version (that is, the x of respective direction _sh(4), x _sh(3), x _sD+and x (3) _sD-(3)) summation, with the transient state improvement value (x of separated specific direction _h, x _v, x _d+, x _d-).At luminance transient, improve at least one embodiment of module 202, the transient state improvement value of specific direction is gained to produce the transient state improvement value u of specific direction _h, u _v, u _{+ 45}and u _-45, some or all in the transient state improvement value of these specific directions combine (for example, by using summation module 342) to initial relevant with concerned pixel original pixel values, to produce the adjustment pixel value (z of concerned pixel ₁(x, y)).Then, for next pixel, repeat this processing.

With reference to Fig. 4, in at least one embodiment, can be that transient state is improved filter 326, transient state and improved filter 328 or transient state and improve filter 330 and can be suitable for transient state and improve luminance transient that filter 324 receives as input z (4), z (3), z (2), z (1), z (0), z (1), z (2), z (3), z (4) and improve module (for example, luminance transient improves filter 400) and comprise the conllinear pixel determined in the pixel window approximate high pass filter 404 of second derivative in respective direction.The signal obtaining (for example, signal c (n)) represents that brightness value is at the rate of change (that is, increase or reduce) of the rate of change of respective direction.

In at least one embodiment, luminance transient improves filter 400 and comprises the conllinear pixel determined in the pixel window approximate high pass filter 406 of first order derivative in respective direction.The signal obtaining (for example, signal 407) represents along the rate of change (that is, slope) of the brightness value of respective direction conllinear.

In at least one embodiment, luminance transient improves filter 400 for example for example, by obtaining gain 407 absolute value from gain control module (, look-up table 410 in memory) obtain corresponding yield value, based on converter (scalar) 407, produce gain (for example, g ₁(n)).In at least one embodiment, determine g ₁(n) to increase radio-frequency component the not obvious enhancing noise in the situation that.At luminance transient, improve at least one embodiment of filter 400, look-up table 410 is realized as the input value of k (n) and the yield value g exporting by gain control module 410 ₁(n) function of Fig. 5 B of the exemplary dependency relation between corresponding output valve.The absolute value of first order derivative approximation k (n) and gain control value g ₁(n) exemplary relation between has the curve of relative bell-shaped, makes gain control value g ₁(n) relatively little or relatively little when relatively large at k (n), and in the situation that k (n) falls between these two kinds of extreme values, gain control value g ₁(n) relatively large.This relation is will be in the further sharpening that there is no a pixel by being favourable time marquis (, in the situation that there is ring or noise), by gain control value g ₁(n) reduce the importance of transient state improvement value.

In at least one embodiment, luminance transient improves filter 400 and uses z (4), z (3), z (2), z (1), z (0), z (1), z (2), z (3), the brightness value of z (4), to improve filter 400 at luminance transient, be that the transient state of horizontal edge boostfiltering device module 304 obtains while improving filter 324 and strengthens pixel, and use z (3), z (2), z (1), z (0), z (1), z (2), the brightness value of z (3), to improve filter 400 at luminance transient, be that other transient state of Fig. 3 obtains while improving in filter and strengthens pixel.Then, luminance transient improves filter 400 and for example, obtains corresponding yield value from gain control module (, look-up table 402).In at least one embodiment, determine g ₂(n) to increase the intensity of signal when mean flow rate higher (that is, pixel is darker), and when mean flow rate lower (that is, pixel is brighter), reduce intensity.At luminance transient, improve at least one embodiment of filter 400, look-up table 402 is realized as mean flow rate and the yield value g exporting by gain control module 402 ₂(n) function of Fig. 5 A of the exemplary dependency relation between analog value.This relation is the impact that will reduce or eliminate overshoot or owe punching.

For example, by using multiplier 412 to apply gain g to c (n) ₁and g (n) ₂(n), for example, to produce the transient state improvement value (, adjusted value a (n)) of the specific direction in respective direction.At luminance transient, improve at least one embodiment of filter 400, summation module (for example, adder 416) by the brightness value of the smoothed version of adjusted value a (n) and concerned pixel (for example, z (0)) summation, for example, to produce the value of highlighting (, z (0)+a (the n)=z of the specific direction of concerned pixel _e(n)).At luminance transient, improve at least one embodiment of filter 400, as the edge value of the highlighting output z of the specific direction of concerned pixel _e(n).But, at luminance transient, improve at least one embodiment of filter 400, impact and suppress further processing z of module 422 _e(n) to reduce the impact of overshoot/owe punching.

In at least one embodiment, impact to suppress module 422 and check overshoots and owe punching, and the value of highlighting of specific direction in response to any overshoot detecting or a little less than owing to eat up part of.At luminance transient, improve at least one embodiment of filter 400, minimum generation module 418 is determined minimum value (for example, the x of the enhancing pixel value of the specific direction of exporting _min), and maximum generation module 420 is determined maximum (for example, the x of the enhancing pixel value of specific direction _max).For example, minimum generation module 418 is determined the minimum value of z (3), z (2), z (1), z (0), z (1), z (2), z (3), and maximum generation module 420 is determined the maximum of z (3), z (2), z (1), z (0), z (1), z (2), z (3).Impact and suppress module 422 use x _minto produce, owe to eat up part of weak pixel value (for example, the D that strengthens _u).For example, in impacting at least one embodiment that suppresses module 422, D _uthe smooth pixel value of the weighted value based on minimum value and concerned pixel:

D _U=(cc _maxmin×x _min)+(cc _current×z(0)),

Wherein, cc _maxminminimum value and peaked pruning coefficient, and cc _currentit is the pruning coefficient of the original pixel values of concerned pixel.Prune coefficient cc _maxminand cc _currentcan be determined by simulation.In at least one embodiment, cc _maxmin=0.75 and cc _current=0.25.Similarly, at least one embodiment, the peaked maximum x that maximum generation module 420 produces as the pixel value of the concerned pixel conllinear with on specific direction _max(for example, x _max=max{z (3), z (2), z (1), z (0), z (1), z (2), z (3) }).Impact and suppress module 422 use x _maxto produce overshoot, weaken enhancing pixel value (for example, D _o).For example, in impacting at least one embodiment that suppresses module 422, D _olevel and smooth brightness value based on peaked weighted value and concerned pixel:

D _O=(cc _maxmin×x _max)+(cc _current×z(0)).

In impacting at least one embodiment that suppresses module 422, if being less than, the pixel value improving owes to eat up part of weak pixel value (that is, the z that strengthense(n)<Du), impacting so inhibition module 422 provides Duenhancing pixel value z as the specific direction of exportingtI(n).If overshoot weakens and strengthens pixel value and be greater than overshoot and weaken maximum (that is, Do<ze(n)), impact so inhibition module 422 D is providedoenhancing pixel value z as the specific direction of exportingtI(n).Otherwise impacting inhibition module 422 provides ze(n) the enhancing pixel value z of the specific direction of conduct outputtI(n).

Usually, the impact inhibition module 422 that luminance transient improves filter 400 has non-linear effect, and radio-frequency component can be incorporated in the output stream that strengthens pixel value.Therefore, in the self adaptation of Fig. 2 A and Fig. 2 B, strengthen at least one embodiment of module 200, for example, by thering is two-dimensional adaptive peaking module 204 that ring controls, further process and strengthen pixel value (z ₁(x, y)) output stream, to strengthen those radio-frequency components in the sharpening image obtaining.

With reference to Fig. 6, in at least one embodiment of two-dimensional adaptive peaking module 204 with ring control, the value of highlighting of concerned pixel (, Y ') for example, by (using two-dimensional high-pass filter, two dimensional filter 802) smoothedization, and produce adjusted value and with the transient state that highlights, improve the radio-frequency component in the output of module 202.In at least one embodiment, two dimensional filter 802 has and meets above 7 * 9 kernel functions technology, corresponding with filtering window of describing for two-dimensional filtering, and produces the single output valve as the window of 7 * 9 pixels of input reception.But for using different filtering window sizes and/or symmetric self adaptation to strengthen other embodiment of module 200, two dimensional filter 802 has the kernel function corresponding with this filtering window size and symmetry.

In thering is at least one embodiment of the two-dimensional adaptive peaking module 204 that ring controls, to the output of two dimensional filter 802, apply one or more yield values (for example, g ₃(n), g ₄and g (n) ₅).In at least one embodiment, yield value (for example, g ₃(n) be) to allow the user-defined gain of sharpening degree horizontal adjusting to control.In at least one embodiment of video system 100, according to individual preference, by the user of video system 100, determine g ₃(n).In at least one embodiment of video system 100, based on picture quality information, by equipment manufacturers, pre-determine g ₃(n).In at least one embodiment of video system 100, g ₃(n) there is the value of the predetermined factor providing based on customer-furnished factor He You equipment manufacturers.In at least one embodiment, for example, by the gain of using two dimensional filter (two dimensional filter 805, with 3 * 3 kernel functions) low-pass filtering to strengthen pixel value, control path 820 and produce one or more yield values (for example, g ₄and g (n) ₅(n)).Low-pass filtering is will prevent from adjusting signal to change to be conducive to slower variation too soon.Gain is controlled path 820 and for example, is obtained corresponding yield value from gain control module (, the look-up table in memory 806).In at least one embodiment, determine g ₄(n) to increase the edge of high average brightness value (that is, bright pixel), strengthen, and reduce the edge enhancing of low average brightness value (that is, dark pixel).In at least one embodiment of two-dimensional adaptive peaking module 204 with ring control, determine g ₅(n) with the detection based on vertical edge and horizontal edge, increasing edge strengthens.The output of two dimensional filter 805 is passed through two dimensional filter (for example, HPF X 808 and HPF Y 810) by high-pass filtering, with detection level edge and vertical edge.In at least one embodiment, HPF X 808 and HPF Y 810 have respectively 3 * 3 kernel functions, but can use other suitable filter.Summation module 811 is by exporting to produce by the absolute value phase Calais junction filter of filter output the yield value that mostly is 1 most.Soft core module 812 is less than at 1 o'clock in combination provides this combination as g ₅, otherwise provide 1 as yield value (n).Adjusted value output to two dimensional filter 802 applies yield value g ₃(n), g ₄and g (n) ₅(n), and by the value of the highlighting z of the gain adjustment value obtaining and concerned pixel ₁(x, y) summation, to produce the second adjustment pixel value z ₂(x, y).

In at least one embodiment of two-dimensional adaptive peaking module 204 with ring control, second adjusts pixel value z ₂(x, y) is further processed with the ring in limited images.In at least one embodiment, ring control module 804 is for concerned pixel z ₁(x, y) determines that by the maximum module 822 of use max pixel value is also by being used the definite minimum pixel value of minimum module 824 on the whole window of adjusting brightness value.Each in maximum module 822 and minimum module 824 is carried out the adjustment pixel value z of concerned pixel ₁the comparison of (x, y) and each original pixel values in relevant window, to determine minimum pixel value and max pixel value.Then, at least one embodiment, ring control module 804 is adjusted these extreme values (that is, minimum and max pixel value), and to allow respectively some to owe punching and overshoot, this can improve the vision of the image obtaining.For example, each in extreme value is gained 2 ^-n, wherein n be programmable integer numerical value (for example, n=3) and with the combination of initial extreme value (for example, with its addition or subtract each other), for example, to produce the extreme value (, being respectively the maximum of adjustment or the minimum value of adjustment) of adjustment.Second adjusts pixel value z ₂(x, y) then compares with adjustment extreme value.If second adjusts pixel value, surpass adjustment maximum, as output, adjust pixel value Y so _oUT(x, y) provides the maximum of adjustment.If second adjusts pixel value, be less than adjustment minimum value, as output, adjust pixel value Y so _oUT(x, y) provides the minimum value of adjustment.Again see figures.1.and.2, at least one embodiment of video SOC 102, the pixel value Y after the processing obtaining of concerned pixel _oUT(x, y) is buffered, and then again stored into the corresponding position in memory 104.

Again with reference to Fig. 2 A and Fig. 2 B, in self adaptation, strengthen at least one embodiment of module 200, colourity transient state improve module 206 based on corresponding luminance pixel values for example, to colourity pixel value (, C _band C _rin each) carry out transient state and improve and process.With reference to Fig. 8, in colourity transient state, improve at least one embodiment of module 208, for each chromatic component (C _band C _r) illustration chromatic component transient state improves module 206.With reference to Fig. 8, in chromatic component transient state, improve at least one embodiment of module 206, each in both direction filter module (for example, horizontal edge boostfiltering device module 502 and vertical edge boostfiltering device module 506) by specific pixel (, concerned pixel or current pixel) apply two dimension, nonlinear filtering and determine for example, transient state improvement value in respective direction (, level or vertical).In transient state, improve at least one embodiment of module 206, horizontal edge boostfiltering device module 502 and vertical edge boostfiltering device module 506 comprise that the linear filtering based on concerned pixel produces the pixel value (x of corresponding smoothing _shand x _sv) corresponding two-dimentional smoothing filter unit (for example, being respectively two dimensional filter unit 512 and 514).In transient state, improve at least one embodiment of module 206, two dimensional filter unit 314 and the 316 identical filter functions with Fig. 3 are realized respectively in two dimensional filter unit 512 and 514.

In chromatic component transient state, improve at least one embodiment of module 206, each in directional filter module 502 and 504 (is respectively x based on corresponding smooth pixel value _shand x _sv) and from the nonlinear filtering of a plurality of pixel values around of pixel window, the transient state that produces the corresponding specific direction of concerned pixel in pixel window buffer 302 is improved chromatic value (z _ahand z _av).For example, the transient state of horizontal edge boostfiltering device module 502 improve filter 526 based on [x _sh0, x _sh1, x _sh2, x _sh3, x _sh4, x _sh5, x _sh6, x _sh7, x _sh8] nonlinear filtering of corresponding [z (4), z (3), z (2), z (1), z (0), z (1), z (2), z (3), z (4)], produce the transient state improvement value z in horizontal direction _ah, and the transient state of vertical edge boostfiltering device module 504 improve filter 528 based on [x _sv0, x _sv1, x _sv2, x _sv3, x _sv4, x _sv5, x _sv6] nonlinear filtering of corresponding [z (3), z (2), z (1), z (0), z (1), z (2), z (3)], produce transient state improvement value TI in vertical direction, the nonlinear filtering based on pixel window _v.Transient state is improved pixel value z _ahand z _avin each represent concerned pixel in the horizontal direction with vertical direction in a corresponding direction on sharpening degree enhanced form.Summation module 534 and 536 improves pixel value z by corresponding transient state _ahand z _avwith the smoothed version summation of concerned pixel, with the transient state improvement value of separated specific direction, (be respectively x _hand x _v).In chromatic component transient state, improve at least one embodiment of module 206, the transient state adjusted value of specific direction is gained to produce the transient state improvement value u of specific direction _hand u _v, some or all in the transient state improvement value of these specific directions and the initially relevant original pixel values combination with concerned pixel, to produce the enhancing pixel value of concerned pixel.Then, for next pixel, repeat this processing.

With reference to Fig. 9, in at least one embodiment, colourity transient state (is for example improved filter module, can be that the chromatic component transient state that transient state that transient state is improved filter 528 and can be suitable for receiving as input z (4), z (3), z (2), z (1), z (0), z (1), z (2), z (3), z (4) is improved filter 526 is improved filter 900) comprise the two-dimensional high-pass filter 904 of carrying out with high pass filter 404 same operation of Fig. 4, to produce, represent that chromatic component value increases on specific direction or the signal c (n) of the speed that reduces.In at least one embodiment, colourity transient state is improved filter module 900 and is comprised carrying out with high pass filter 406 same operation of Fig. 4 and represent along the rate of change of the chromatic component value of specific direction conllinear (to produce, the two-dimensional high-pass filter 906 of invariant signal slope) (for example, signal 907).

In at least one embodiment, chromatic component transient state improves that module 206 realizes the module 908 of Figure 10 C function by use and module 910 produces gain (for example, g based on converter 907 ₁(n)).

Module

908 and 910 realizes

module

408 and 410 identical technology and the operations with above-mentioned Fig. 4 and Fig. 5 B, these technology and operate in the g that determines increase radio-frequency component in the situation of not obvious enhancing noise ₁(n) value.

In at least one embodiment, chromatic component transient state is improved filter 900 by using multiplier 912 to apply gain g to c (n) ₁(n), for example, to produce the transient state improvement value (, improvement value a (n)) of the specific direction of respective direction.In colourity transient state, improve at least one embodiment of filter module 900, summation module (for example, adder 916) by a (n) and initial chroma signal component value (for example, z (0)) summation, for example, to produce enhancing chromatic component value (, z (0)+a (the n)=z of the specific direction of concerned pixel _e(n)).In chromatic component transient state, improve at least one embodiment of filter 900, as the edge of the specific direction of concerned pixel, strengthen chromatic component value output z _e(n).But, in chromatic component transient state, improve at least one embodiment of filter 900, impact and suppress further processing z of module 922 _e(n), to reduce the impact of overshoot/owe punching, this meets above about impacting the technology that module 422 is described that suppresses.

Usually, due to chromatic component C _band C _rby high compression and to be tending towards be noise signal, be therefore more difficult at C _band C _redge detected in signal.In addition, when following brightness to change, the variation of colourity is tending towards more obvious.Therefore,, in order to reduce noise false signal, just realize the edge of chromatic component while only edge being detected in the corresponding brightness value at concerned pixel, strengthen.In chromatic component transient state, improve at least one embodiment of filter 900, by the control Path generation yield value (k) of Figure 10 B.For example, in chromatic component transient state, improve at least one embodiment of filter 900, the luminance pixel values of concerned pixel by high-pass filtering with Edge detected.Then the absolute value of the brightness value of filtering is constrained to by the function of Figure 10 A the value that is less than or equal to 1.If the absolute value of the brightness value of filtering is greater than 1, k is constrained to 1 so.The intensity at the edge detecting in brightness value represents by k, and by deducting k from 1, produces the weight corresponding with initial chroma value.Therefore, the chromatic component Zhi edge enhancing value of concerned pixel and the initial chroma signal component value of concerned pixel are weighted and combine, to produce output chromatic component Z _tI.

Again with reference to Fig. 8, for specific chromatic component, transient state is improved pixel value z _ahand z _avin each represent concerned pixel in the horizontal direction with vertical direction in a corresponding direction on sharpening degree enhanced form.Summation module 534 and 536 improves pixel value z by corresponding transient state respectively _ahand z _avwith the smoothed version x of concerned pixel about the filtered chromatic component of respective direction _shand x _svsummation, with the transient state improvement value (x of separated specific direction _hand x _v).In chromatic component transient state, improve at least one embodiment of module 206, the transient state improvement value of specific direction is gained to produce the transient state improvement value u of specific direction _hand u _v, one or two in the transient state improvement value of these specific directions and the initial original pixel values relevant with concerned pixel combination (for example, by using summation module 542), using produce concerned pixel edge enhancing pixel value (as the C of Fig. 2 A and Fig. 2 B _bOUTor C _rOUTc _oUT).Then, next pixel strengthening for Yao Bei edge repeats this processing.Again see figures.1.and.2, at least one embodiment of video SOC 102, the treated pixel value C of the concerned pixel obtaining _bOUT(x, y) and C _rOUT(x, y) is buffered and then again stored into memory 104Zhong relevant position.

Can be by using the software (comprising firmware) that carry out on the processor in being contained in video processing unit 108 or 110 or the combination by software or hardware to realize structure described herein.Can be by Software Coding described herein at least one palpable computer-readable medium.As mentioned above, palpable computer-readable medium at least comprises dish, tape or other magnetic, optics or electronic storage medium.

Although general supposition circuit and physical structure when describing embodiments of the invention, but in modern semiconductor design with in manufacturing, recognize well, can embody physical structure and circuit with design, simulation, test or the computer-readable description form in the fabrication stage being suitable for subsequently.The structure or the parts that can be used as combination are realized the 26S Proteasome Structure and Function providing as discrete parts in exemplary configuration.Various embodiment of the present invention is envisioned for and for example comprises here the circuit describing and limit in the appended claims,, the system of circuit, relevant method and have the palpable computer-readable medium of the coding (, VHSIC hardware description language (VHDL), Verilog, GDSII data, electronic design interchange format (EDIF) and/or Gerber file) of sort circuit, system and method on it.In addition, computer-readable medium can be stored and can be used to realize instruction of the present invention and data.Instruction/data can be relevant to hardware, software, firmware or their combination.

Here the description of the invention of setting forth is indicative, and is not to be limited in the scope of the present invention of setting forth in following claim.For example, although described the present invention in the embodiment about horizontal direction and vertical direction processing chromatic component, but it will be understood by those skilled in the art that and can only by the processing of luminance pixel values or the transient state at edge, improve to utilize the instruction here about other combination of orientation.In addition, although described the present invention in using specific window size and symmetric embodiment, it will be understood by those skilled in the art that can be by other window size and the symmetry instruction that utilizes these.And, although video system 100 comprises for brightness value, apply and there is the two-dimensional adaptive peaking module 204 that ring is controlled, meet video system 100(Fig. 2 B of instruction described herein) other embodiment to chromatic component value C _bOUTand/or C _rOUTapply and there is the two-dimensional adaptive peaking module 204 that ring is controlled.Do not deviating from the situation that the scope and spirit of the present invention of setting forth in following claim, can the description based on setting forth proposing change and the modification of embodiment disclosed herein here.

Claims

1. the equipment strengthening for the adaptive edge of vision signal, comprising:

The pixel value window that is configured to concerned pixel based on initially having input pixel value pixel around produces the first transient state of adjusting pixel value and improves module; With

Be configured to adjust based on self adaptation adjusted value, first the self adaptation peaking module of pixel value and input pixel value generation the second adjustment pixel value.

2. equipment as claimed in claim 1,

Wherein, self adaptation peaking module comprises the high pass filter that is configured to produce based on the first adjustment pixel value pixel adjustment, and

Wherein, self adaptation peaking module also comprises and is configured to apply at least one adaptive gain value to produce the gain path of self adaptation adjusted value to pixel adjustment.

3. equipment as claimed in claim 2, also comprises: be configured to produce the control path of described at least one adaptive gain value, described control path comprises:

Be configured to produce the two-dimensional low-pass filter of the low-pass filtered version of the first adjustment pixel value;

Be configured to amplify in the low-pass filtered version of the first adjustment pixel value at least one two-dimensional high-pass filter at first direction Shang edge at least,

Wherein, described at least one adaptive gain value is exported accordingly based on described at least one two-dimensional high-pass filter at least one; Alternatively, described control path also comprises:

Being configured to intensity based on low-pass filtering input pixel provides the gain control module of the first adaptive gain value;

Wherein, described at least one two-dimensional high-pass filter comprises:

Be configured to amplify in the output of two-dimensional low-pass filter the first two-dimensional high-pass filter at edge in the horizontal direction;

Be configured to amplify in the output of two-dimensional low-pass filter the second two-dimensional high-pass filter at edge in vertical direction;

Wherein, the second output of adaptive gain value based on the first and second two-dimensional high-pass filters.

4. equipment as claimed in claim 2 or claim 3, wherein, self adaptation peaking module also comprises and is configured to combine adaptive gain adjusted value and first and adjusts pixel value to produce the summation module of the second adjustment pixel value.

5. the equipment as described in any one in aforementioned claim, also comprises:

Be configured to produce based on the second adjustment pixel value the ring control module that pixel value is adjusted in output, described output is adjusted pixel value in the scope based on minimum pixel value and max pixel value.

6. the equipment as described in any one in aforementioned claim, wherein, it is that luminance transient improves module that transient state is improved module, and described equipment also comprises:

The window that is configured to the chroma pixel value of the input chromatic value based on comprising concerned pixel produces at least one colourity transient state of adjusting chroma pixel value and improves module, and described input pixel value is corresponding brightness value.

7. the equipment as described in any one in aforementioned claim,

Wherein, luminance transient improves module and colourity transient state and improves at least one in module and also comprise to impact and suppress module,

Wherein, impact inhibition module and be configured to adjust when pixel value is greater than maximum adjustment pixel value and select maximum to adjust pixel value as the first adjustment pixel value in transient state,

Wherein, impact inhibition module and be configured to adjust when pixel value is less than maximum adjustment pixel value and select minimum to adjust pixel value as the first adjustment pixel value in transient state, and

Wherein, impact suppressing module is configured to adjust pixel value in transient state and is less than or equal to maximum pixel value and the transient state adjusted and adjusts pixel value and be more than or equal to and minimumly select transient state to adjust pixel value as the first adjustment pixel value while adjusting pixel value.

8. the equipment as described in any one in aforementioned claim, wherein, transient state is improved module and is configured to:

Combination corresponding to a plurality of horizontal directions, vertical direction ,+45Du Dui angular direction and-adjustment and the input pixel value of 45Du Dui angular direction, to produce the first adjustment pixel value; And/or

Combination is along the gain control value of the first approximation of the conllinear pixel value of the pixel of first direction conllinear, the second approximation of conllinear pixel and the average brightness value based on pixel value, and to produce adjusted value, wherein said first adjusts pixel value based on adjusted value.

9. the method strengthening for the edge of vision signal, comprising:

Pixel value window based on concerned pixel pixel around produces the transient state of concerned pixel and adjusts pixel value, and described concerned pixel initially has input pixel value; With

High pass filter based at least one adaptive gain value, pixel value window and transient state are adjusted the output adjustment pixel value that pixel value produces concerned pixel.

10. method as claimed in claim 9, wherein:

Produce output adjustment pixel value and comprise the toroidal effect reducing by linear passivation shade filtering introducing;

When generation output adjustment pixel value comprises outside the transient state of adjustment form is adjusted the scope of pixel value in pixel value, provide the upper limit of output adjustment pixel value or the lower limit of output adjustment pixel value to adjust pixel value as output, the scope of described pixel value is limited by upper and lower bound, and window and the transient state of described upper and lower bound based on pixel adjusted pixel value; And/or

Described at least one adaptive gain is configured to the rate of change that pixel value is adjusted in restriction output.

11. methods as described in claim 9 or 10, also comprise:

Second Window and the transient state of the pixel value of low-pass filtering concerned pixel pixel are around adjusted pixel value; With

Output based on low-pass filtering produces at least one adaptive gain value; Alternatively, producing at least one adaptive gain value comprises:

The output of two-dimensional high-pass filtering low-pass filtering to be to produce the first yield value, any edge on described two-dimensional high-pass filter and amplification first direction;

The output of two-dimensional high-pass filtering low-pass filtering to be to produce the second yield value, any edge in described two-dimensional high-pass filter and amplification second direction, and described second direction is different from described first direction; And

Based on the first and second yield values, produce described at least one adaptive gain.

12. methods as described in claim 9,10 or 11, wherein, produce transient state and adjust pixel value and comprise:

When the first transient state adjustment pixel value is greater than maximum transient state adjustment pixel value, select maximum transient state adjusted value to adjust pixel value as transient state;

When the first transient state adjustment pixel value is less than minimum adjustment pixel value, select minimum transient state adjusted value to adjust pixel value as transient state; With

When the first transient state, adjust pixel value and be less than or equal to maximum transient state and adjust pixel value and the first transient state and adjust pixel value and be more than or equal to minimumly while adjusting pixel value, select the first transient state to adjust pixel value and adjust pixel value as transient state.

13. methods as described in any one in claim 9～12, wherein:

It is that transient state is adjusted brightness value that transient state is adjusted pixel value, and input pixel value is luminance pixel values, and described method also comprises that based on luminance pixel values, producing transient state adjusts chromatic value; Or

It is that transient state is adjusted chromatic value that transient state is adjusted pixel value, and described input pixel value is chroma pixel value.

14. methods as described in any one in claim 9～13, wherein, transient state adjusted value based on about horizontal direction, vertical direction ,+45Du Dui angular direction or-at least one a plurality of adjusted values in 45Du Dui angular direction; Alternatively, producing transient state adjustment pixel value comprises:

Combination corresponding to a plurality of horizontal directions, vertical direction ,+45Du Dui angular direction and-adjustment and the input pixel value of 45Du Dui angular direction adjust pixel value to produce transient state.

15. methods as described in any one in claim 9～14, also comprise:

The first approximation of the conllinear pixel value of the pixel based on along first direction conllinear, the second approximation of conllinear pixel value and the gain control value based on average brightness value, produce transient state and adjust pixel value.