CN104299204A - Histogram local image contrast enhancing method and histogram local image contrast enhancing device - Google Patents

Abstract

The invention relates to a histogram local image contrast enhancing method and a histogram local image contrast enhancing device. After the histogram distribution of one image or video image is obtained, a gray space of a non-key-brightness region of a histogram is compressed by adopting a mode of selecting the brightness range, so that the pixel of the non-key-brightness region is extruded to a small range; the pixel in a key-bright region can obtain a greater brightness gray value equalization space; then, the gray range of an image key block is expanded through specific HS (Histogram Stretching) based method or HE (Histogram Equalization) based method technology; more obvious contrast enhancing and image sharpness improving effects can be obtained; and the problem of distortion of contents of the whole image due to simultaneous distribution out-of-balancing of the brightness gray value of the pixel in other regions of the image or the video image when the brightness gray value of a local part of the whole image or the video image needs to be enhanced in the prior art is solved.

Description

The method and apparatus of histogram local image Contrast enhanced

Technical field

A kind of device strengthening image contrast of the present invention, espespecially a kind of histogram range expanding local image is to strengthen the device of image contrast.

Background technology

The quality that the contrast of existing individual image (photo) or continuous image (film) distinguishes this quality of image for human eye occupies considerable status, if the environment of shooting is undesirable, the contrast of acquired image may be caused not enough, produce overall image and show tired situation, for solving the problem of image contrast's deficiency, the operation method of various enhancing contrast just arises at the historic moment.

Utilize the expansion to image contrast, image can be made visually to produce stronger impression, and then reach the object of promoting image definition.The function of common gray scale histogram (Histogram) its GTG (Gray) level is the number of times that in reaction image, pixel occurs often kind of gray scale levels.Raw video as is shown in fig. 1, Fig. 2 is the gray scale histogram statistics of the raw video of Fig. 1, wherein the horizontal ordinate of this gray scale histogram is gray scale levels, ordinate is the number of pixels that this gray scale levels occurs, pixel major part from the gray scale histogram of Fig. 2 in known Fig. 1 is seated between gray scale levels 20 to 200, therefore, the marker method of this gray scale histogram is the most basic statistical nature of image.And the principle that image contrast strengthens, being the contrast by expanding between its histogram of neighbor, making human eye visually can perceive more image detail texture and edge part.

The existing image contrast enhancing technology being Contrast enhanced process basis with the histogram of image is mainly divided into two classifications, be the image contrast enhancing technology that Contrast enhanced method (Histogram Stretching Based Methods) and histogram equalization Contrast enhanced method (Histogram Equalization Based Methods) are stretched in histogram exhibition respectively, its technology contents is respectively described below.

Histogram exhibition is stretched Contrast enhanced method (Histogram Stretching Based Methods) and is utilized raw video histogram with its universe grey-scale range, again stretch and distribute its new image histogram distribution, wherein more famous method has the technology such as Histogram Stretching (HS) and Improved Histogram Stretching method (IHS), the advantage of this type of technology is only just to make former histogrammic gray-scale distribution to stretch and can not change the brightness distribution curve of original image, thus algorithm is comparatively simple and can not change the luminance difference relation of each pixel, the situation of image fault can not be produced, but within the scope of intensity gray scale value, seldom have one whole section of luminance area all without the situation of pixel distribution because of the histogram of most image, thus therefore this method often can not find and cannot be able to tell on for the brightness range stretched.

Aforementioned HS method utilizes the grey decision-making of raw video to do histogram analysis, afterwards its histogram is stretched and extend to full dynamic area, as shown in Fig. 3 to 5, Figure 3 shows that original histogram, the histogram of Fig. 3 is deducted minimum value (0 to 130) by Fig. 4, Fig. 5 be histogram after deleting of Fig. 4 by after HS method, produce the histogram of expansion.This HS method is applicable to the image using histogram to concentrate, and utilizes equation (1) to obtain new output pixel value, and wherein this numerical value 255 is the maximum brightness value inputting image at 8-bit.

[several 1]

$\mathrm{NewPixel}=\frac{\mathrm{Inputpixel}-\mathrm{Lowpixel}}{\mathrm{High}-\mathrm{lowpixel}}\×255---\left(1\right)$

Based on HS method, Hee-Chul Kim proposes IHS method, can be applicable to the method realizing image processing on hardware, is represented by equation (2):

[several 2]

New pixel＝(Inputpixel-Low pixel)×(M+US) (2)

This IHS method is the pixel value deciding to expand histogrammic intensity via weight parameter M (Multiple) and US (User Select) and obtain image output, the value of weight parameter M determines main growth intensity, weight parameter US is user's setup parameter, mainly in the intensity of this weight parameter M of fine setting, the original HS of this IHS method improvement has the generation of division formula, and its advantage realizes on hardware for easier.

Histogram equalization Contrast enhanced method (Histogram Equalization Based Methods) mainly carries out nonlinear stretching to raw video histogram, and redistribute image pixel value, make the histogram of raw video be balanced histogram distribution.HE method, owing to can produce average Luminance Distribution, makes image phenomenologically seem comparatively natural, and therefore current enhancing in image contrast is the technology be generally employed, and this method is commonly used to increase the universe of many images to specific strength.First, defining a raw video is X, and the total pixel number of this raw video is N, and image gray scale range set is [0, L-1], and L represents the maximum grey decision-making of image pixel.Its probability density function (Probability Density Function, PDF) definition is as equation (3):

[several 3] $P\left(X_k\right)=\frac{n_k}{N},\mathrm{fork}=\mathrm{0,1},...,L-1\left(3\right)$

What above-mentioned k represented is gray scale levels, n _krepresent the number of pixels that this gray scale levels occurs.Fig. 6 is the statistics after being calculated by PDF.And cumulative density function (Cumulative Density Function, CDF) is then the calculating carrying out progression according to PDF, as shown in equation (4):

[several 4] $C\left(X_k\right)=\underset{i=0}{\overset{k}{\mathrm{\Σ}}}P\left(X_k\right),\mathrm{fork}=\mathrm{0,1},...,L-1---\left(4\right)$

And the result of CDF statistics, as shown in Figure 7, the maximal value of its progression result will equal 1; Finally by the transfer function of image, as shown in equation (5),

[several 5]

f(X _k)＝X ₀+(X _L-l-X ₀)C(X _k) (5)

Carry out histogram equalization conversion process to obtain the image after a Contrast enhanced.As shown in Figure 8, after the process of HE method, corresponding new histogram distribution be in 0 ~ L-1 between the expansion of universe equilibrium.

Because HE method has directly simple characteristic, Contrast enhanced improved method based on HE method is delivering the technology of many improvement in recent years successively, with BBHE method be wherein one class technology representative illustrate: first BBHE utilizes the histogram of original input image to be divided into two sub-histogram (Sub-Histogram) regions according to average brightness value (Xm), histogram equalization process is done respectively afterwards, as shown in Figure 9 for this two region.Similar method has DSIHE and SSTHE, is equally also to be cut histogram by mean flow rate or intermediate value brightness, does indivedual histogram equalization process afterwards to each sub-histogram.

Another kind of method takes the relevance between pixel position and its pixel gray scale levels into consideration, to reach the effect of Contrast enhanced, for Contrast limited adaptive histogram equalization method (CLAHE), its step realized is as follows:

(1) blockette (Block): will input image and divide equal-sized not overlapping sub-block, the total pixel number that each sub-block contains is set as M.The scope of sub-block setting is larger, and the effect of its Contrast enhanced is more obvious, but but easily produces the situation that image detail easily loses.

(2) compute histograms: X _jjrepresent the histogram of sub-block, k represents gray scale levels, and the setting range of its gray scale levels is 0 ~ L-1, and wherein L expresses possibility the GTG progression occurred.

(3) limit grade β is calculated: utilize equation (6) to calculate limit grade (Restrained Level), S in formula _maxbe called maximum slope, pass through S _maxsetting can determine the amplitude of Contrast enhanced, and the scope of its value setting is the integer between 1 ~ 4; α is called limiter stage coefficient, and the scope of its value setting is between 0 ~ 100.When α equals 0, β gets minimum value when α equals 100, get maximal value now corresponding contrast drawing effect is the most obvious.

[several 6] $\mathrm{\β}=\frac{M}{N}(1+\frac{\mathrm{\α}}{100}(S_{\max }-1))---\left(6\right)$

(4) pixel is redistributed: according to previous step, to each sub-block, use corresponding β limit grade parameter to carry out restriction to the histogram of sub-block to shear, next reassigned in the histogrammic gray scale levels of sub-block by the number of pixels sheared, the pixel be sheared until all is assigned with complete.

(5) histogram equalization process: the histogram of each sub-block that pixel is assigned, does individually the process of histogram equalization according to HE method.

(6) pixel gray level level is rebuild and is calculated: the average brightness calculating each sub-block according to previous step, it is set as reference point, adopts the account form of bilinear interpolation method (Bilinear Interpolation) to obtain exporting its new gray scale levels of each pixel in each sub-block.

A class technology is also had to be that the mode of the PDF utilizing restriction or amendment raw video is to obtain the image of new Contrast enhanced.For Adaptively increasing the value of histogram (AIVHE): as shown in equation (7) and (8), AIVHE crosses enhancing for prevention image and adds a parameter P _hto limit the upper limit of PDF, in addition according to basic parameter P _bascollocation adaptability adjustment parameter alpha (k), by adjustment β and γ user adjust parameter to original image PDF do modify external form (Reshape) action to obtain new PDF, rear by HE method to reach image enhaucament contrast effectively.The feature of AIVHE method is as follows: 1.AIVHE utilizes the limiting parameter P preset _havoid exporting the visual effect having and strengthen; 2. make it adapt to reach the effect of Contrast enhanced on various different images by adjustment parameter alpha (k) can be adapted to; 3. can determine to contrast the scope expanded by the adjustment of γ; 4., by adjustment β, make the image of its Contrast enhanced more level off to original image brightness.The method reaching image comparison enhancing equally by revising PDF has the methods such as BUBO, AMHE and WTHE, wherein WTHE has the normalized action of increase by brightness for the ability maintaining original image mean flow rate, so can reach the factitious display result of unlikely generation when broadcasting dynamic image.

[several 7] $P_{\mathrm{AIVHE}}\left(k\right)=\left\{\begin{array}{cc}{P}_h& ,\mathrm{ifP}\left(k\right)\&GreaterEqual;P_h\\ P\left(k\right)-\mathrm{\&alpha;}\left(k\right)(P\left(k\right)-P_{\mathrm{bas}})\&times;\mathrm{\&beta;}& ,\mathrm{if}{P}_{\mathrm{bas}}<P\left(k\right)<P_h\\ P\left(k\right)+\mathrm{\&alpha;}\left(k\right)(P_{\mathrm{bas}}-P\left(k\right))\&times;\mathrm{\&beta;}& ,\mathrm{ifP}\left(k\right)\&le;P_{\mathrm{bas}}\end{array}\right.---\left(7\right)$

As previously mentioned, current various image contrast enhancing technology is all conceived to whole image be done code reassignment by the intensity gray scale value of each pixel, to obtain the effect of Contrast enhanced; But consider the special image features such as such as medical contrastographic picture, the common trait of this type of image is: the intensity gray scale value of pixel is obviously distributed unbalance, as whole medical image has very large some to be close to without the black dull block (non-medical examination district) of brightness, and the key area of institute's wish observation makes its pixel intersperse among in one piece of very little grey-scale range because being identical tissue.And the black dull picture that video camera was shot with video-corder without obvious light source at night, then make its pixel gray level of exhausted most region of whole image all fall within close to the GTG part without brightness, the brighter pixel of minority often or the noise formed because of resolution deficiency or line-hit.Because this type of image is by histogrammic distribution, pixel distribution may be had but really needs the pixel of observing closely to concentrate on really in the shade of gray of certain section by all shade of gray, if thus with traditional contrast enhancement technique process, namely because these technology are all designed by all pixels being conceived to whole pictures are considered, cause last enhancing effect very limited and very easily cause the result of content distortion.

Summary of the invention

As as described in front take off, existing image contrast strengthens technology and is all conceived to whole image to do code reassignment by the intensity gray scale value of each pixel, when intensity gray scale value for the pixel of image is obviously distributed unbalance, for its intensity gray scale value of local enhancement, if easily cause it to strengthen limited efficiency with existing contrast enhancement technique process, and cause whole picture material distortion problem, therefore fundamental purpose of the present invention is providing the method and apparatus of a histogram local image Contrast enhanced, by selecting the mode of brightness range, compress the GTG space of non-emphasis luminance area, the pixel of non-emphasis luminance area is made to be expressed among a small circle, and the pixel of emphasis luminance area can obtain larger intensity gray scale value equalization space, again by technology such as specific HS or HE, by the expansion of the grey decision-making scope of image emphasis block, and obtain the effect of more obvious Contrast enhanced and image definition, solve whole image for its intensity gray scale value of local enhancement, the intensity gray scale value of the pixel in other regions of image distribute unbalance and cause whole picture material distortion problem.

The technical way taked for reaching aforementioned object makes aforementioned histogram local image Contrast enhanced device, includes:

One input video conversion module, it is in order to input an image and to be converted to a YCbCr signal;

One regional luminance strengthens module, and it be electrically connected with input video conversion module, and this regional luminance strengthens module cluster according to the individual control signals of majority, changes and YCbCr signal after producing an adjustment;

One image output modular converter, itself and regional luminance strengthen module and are electrically connected, and this input video conversion module exports and strengthens the YCbCr signal after module adjustment through regional luminance;

One image bright territory inputting interface, itself and regional luminance strengthen module and are electrically connected, this image bright territory inputting interface in order to selected more than one the shade of gray scope (H, L) for enhancing, or inputs more than one pixel coordinate (X, Y) and grey-scale range threshold (Z) thereof;

One coordinates regional luminance analyzing unit, it is electrically connected with image bright territory inputting interface, and pixel coordinate (X, Y) and the grey-scale range threshold (Z) of this coordinates regional luminance analyzing unit foundation image bright territory inputting interface produce an easily extensible value range (Hnew, Lnew);

One bright territory generator, it strengthens module, coordinates regional luminance analyzing unit and image bright territory inputting interface be electrically connected with input video conversion module, regional luminance respectively, this bright territory generator is according to the lightness (Y) in YCbCr signal, set up the probability density function of this signal of video signal, the shade of gray scope (H, L) that image bright territory inputting interface or coordinates regional luminance analyzing unit produce, provides regional luminance to strengthen module according to this extendible scope (Hnew, Lnew) and carries out Contrast enhanced process.

The technical way taked for reaching aforementioned object makes aforementioned histogram local image Contrast enhanced method, includes:

Obtain an image;

Add up the histogram distribution of this image, each for image pixel is divided into key area and non-key area by grey decision-making;

The non-key area of this histogram distribution is compressed, obtains enough GTG spaces to provide this key area to carry out the Contrast enhanced of ductility to specific gray-scale areas;

This region gray scale levels degree widened and improves luminance difference, and strengthening the contrast of its area image, to improve details and the sharpening of image.

Above-mentioned histogram local image Contrast enhanced method, wherein compresses the non-key area at histogram two ends, passes through formula with

key area is made to extend expansion to two ends.

Above-mentioned histogram local image Contrast enhanced method, wherein compress the GTG space of the non-emphasis brightness range of non-key area, the pixel of non-key area is made to be expressed among a small circle, and the emphasis brightness range in a large amount of set of pixels obtains larger intensity gray scale value equalization space, and the stretching of grey decision-making is done by HS technology

Pass through formula

Be converted to the GTG of larger section, make grey decision-making expanded range.

Above-mentioned histogram local image Contrast enhanced method, after wherein each for image pixel being divided into the non-key area in key area and two ends by grey decision-making, user adopts the GTG of assigned picture ad-hoc location to be benchmark GTG by the selection of operator scheme, stored by a queue (Queue) and determined the coordinate with reference point same block/object, whether pixel more adjacent is one by one subordinate to same block/object, and the adjacent coordinates meeting same block/object is restored in this queue (Queue), and repeatedly process until determine all pixel coordinates being subordinate to same block/object.

Above-mentioned histogram local image Contrast enhanced method, whether wherein more adjacent pixel is subordinate to same block/object, first judge whether the grey decision-making of neighbor and the difference of adjacent reference point grey decision-making are less than a luminance difference threshold, if do not meet, the grey decision-making of this pixel and the grey decision-making of neighboring pixel are done on average, environmentally to close its grey decision-making of adjustment, if the difference of this grey decision-making and adjacent reference point grey decision-making is less than luminance difference threshold, then also be subordinate to same block/object at last, and according to its original luminance value adjustment emphasis grey-scale range, this adjustment emphasis grey-scale range refers to according to the pixel in judged block/object, minimum gray value is wherein selected to be L parameter, maximum gray value is H parameter.

Compared with prior art, the histogram local image Contrast enhanced device utilizing aforementioned components to form, contrast enhancement processing is carried out for specific gray-scale areas, and the contrast of non-reinforcing whole image, therefore the intensity gray scale region of this image is first found out, recompress the histogram block in other region, provide key area to carry out the Contrast enhanced of ductility to produce enough GTG spaces, make the sharpening effect that the emphasis main body in image reaches better.Because the image feature that the present invention is directed to special-purpose does the Contrast enhanced process of emphasis formula, user can utilize image bright territory inputting interface to select different operational methods and different Contrast enhanced modes, and by confirming emphasis gray-scale areas and correlation parameter, reach multi-functional application purpose.When solving prior art for carrying out local enhancement intensity gray scale value to whole image, the intensity gray scale value of the pixel in these other regions of image can because of distribute unbalance and cause whole picture material distortion problem.

Accompanying drawing explanation

Fig. 1 is existing raw video figure;

Fig. 2 is the gray scale histogram of corresponding diagram 1 presentation content;

Fig. 3 is the original histogram of another image existing;

Fig. 4 is histogram Fig. 3 being deducted minimum value;

Fig. 5 is that the histogram after being deleted by Fig. 4 opens up the histogram stretched Contrast enhanced method (HS) and expand by histogram;

Fig. 6 is the histogram calculated through probability density function (PDF);

Fig. 7 is the statistical graph calculated through cumulative density function (CDF);

Fig. 8 is by histogram equalization Contrast enhanced method (HE) schematic diagram that universe equilibrium is expanded between 0 ~ L-1;

Fig. 9 is divided into two sub-histograms by BBHE and the schematic diagram of difference equalization processing;

Figure 10 is the circuit block diagram of present pre-ferred embodiments.

Figure 11 is the original histogram of present pre-ferred embodiments;

Figure 12 is the histogram of the expansion of present pre-ferred embodiments;

Figure 13 is the process flow diagram of the region GTG expansion enhancing method of present pre-ferred embodiments;

Figure 14 A, 14B are the process flow diagrams of the coordinates regional brightness analysis method of present pre-ferred embodiments;

Figure 15 is the contrast figure () before and after the image contrast of brain tomoscan (CT) image strengthens;

Figure 16 is the contrast figure (two) before and after the image contrast of brain tomoscan (CT) image strengthens;

Figure 17 is the contrast figure before and after the image contrast of magnetic resonance imaging (MRI) strengthens;

Wherein, Reference numeral:

10 input video conversion module 20 regional luminances strengthen module

30 image output modular converter 40 image bright territory inputting interfaces

50 coordinates regional luminance analyzing unit 60 bright territory generators

Embodiment

The present invention is setting up a histogram local image Contrast enhanced device, for the image that medical imaging contrast apparatus and general monitor are shot with video-corder under black dull environment, first carry out local image Contrast enhanced by hardware by before this image output to display, make the picture being finally output in display be clear picture through emphasis local image Contrast enhanced.User can select two kinds of work patterns complete the selected of specific grey-scale region or directly specify the picture block that will strengthen in the mode clicking picture ad-hoc location by this device, the compression/extension ratio that this technology will be set up by user, gray-scale areas selected by user is done to the image definition process of GTG expansion: the histogram distribution of first adding up this image, again the histogram block in two extreme other regions in distribution is done and compress, vacateing enough GTG spaces provides key area to do the Contrast enhanced of ductility, this region gray scale levels degree is widened namely luminance difference mutation large, the contrast of its area image can be strengthened immediately, make the better accurate recognition of details to reach better sharpening effect.Because this device directly makes with the electronic circuit of Hardware and logical circuit, there is the advantage that processing speed is fast, be applicable to allowing user observe image effect, while adjustment parameter or switching operation modes.

About preferred embodiment of the present invention, refer to shown in Figure 10, include input video conversion module 10, regional luminance and strengthen module 20, image output modular converter 30, image bright territory inputting interface 40, coordinates regional luminance analyzing unit 50 and a bright territory generator 60, wherein

This input video conversion module 10 is in order to input different image formats (NTSC, PAL, RGB or DVI) and to be converted to a YCbCr signal (YCbCr).

It is be electrically connected with input video conversion module 10 that this regional luminance strengthens module 20, this regional luminance strengthens module 20 groups and is provided with multiple control signal (threshold value (H), threshold value (L), threshold value (Hnew), threshold value (Lnew) and grey-scale range threshold (Z)), and it is changed aforementioned YCbCr signal according to these control signals and produces the YCbCr signal after an adjustment (New YCbCr).

This image output modular converter 30 strengthens module 20 with regional luminance to be electrically connected, and this image output modular converter 30 exports and strengthens the YCbCr signal after module 20 adjustment through regional luminance.

This image bright territory inputting interface 40 strengthens module 20 with regional luminance to be electrically connected, this image bright territory inputting interface 40 has most adjustment parameters (threshold value (H), threshold value (L), pixel coordinate (X, Y) and grey-scale range threshold (Z)), this image bright territory inputting interface 40 in order to selected shade of gray scope threshold value (H, L) for strengthening, input more than one pixel coordinate (X, Y) and grey-scale range threshold (Z) thereof; In this preferred embodiment, this shade of gray scope threshold value (H, L) adjusts its numerical value by a knob or a switch.

This coordinates regional luminance analyzing unit 50 is that territory bright with image inputting interface 40 is electrically connected, and this coordinates regional luminance analyzing unit 50 produces an easily extensible value range (Hnew, Lnew) according to the pixel coordinate (X, Y) of image bright territory inputting interface 40 and grey-scale range threshold (Z).

This bright territory generator 60 be respectively with input video conversion module 10, regional luminance strengthens module 20, image bright territory inputting interface 40 and coordinates regional luminance analyzing unit 50 are electrically connected, this bright territory generator 60 is according to the lightness (Y) in the YCbCr signal of input video conversion module 10, set up the probability density function (PDF) of this signal of video signal, the shade of gray scope threshold value (H that image bright territory inputting interface 40 or coordinates regional luminance analyzing unit 50 produce, L), according to this extendible scope (Hnew, Lnew) provide regional luminance to strengthen module 20 and carry out Contrast enhanced process.

Histogram local image Contrast enhanced device of the present invention, when it receives the dynamic image of medical contrast apparatus or photography monitor, first the anaiog signals such as NTSC are decoded into digital signal, the digital signal process such as release of an interleave are done again for video standard, signal is organized into brightness and color-difference formats (YCbCr), wherein brightness is Y, blue color difference is Cb, red color is Cr, regional luminance strengthens module 20 and can process for brightness Y, the Y processed again with Cb, Cr is combined into the YCbCr signal after adjustment (New YCbCr), this NewYCbCr is transferred to NTSC again by image output modular converter 30, PAL, the output based version such as DVI or RGB, deliver to corresponding display device (screen) again.

User carries out following operation by image bright territory inputting interface 40:

1. select its using forestland and compressibility (Z value) by knob or switch.

If 2. select dynamic conditioning pattern, then can according to display result with knob adjustment brightness primary part observation region (L, H value).

If 3. object mode is specified in choosing, then by the input media such as Touch Screen or slide-mouse, directly specify a wherein point coordinate (X, Y) of screen-picture, make the object of specifying in picture, produce the effect of Contrast enhanced.

A frame buffer (Frame Buffer) (not shown) is provided with in the input video conversion module 10 of histogram local image Contrast enhanced device, its Y value matrix storing image specifies L and the H value of object analyze and produce extending range (Lnew for coordinates regional luminance analyzing unit 50 and bright territory generator 60, Hnew), finally done the stretching/compressing of last image histogram distribution specific region again according to L, H, Lnew, Hnew and Z value by regional luminance enhancing module 20.

In the generation of Lnew and Hnew, compress the GTG space of the non-emphasis brightness range of non-key area, the pixel of non-key area is made to be expressed among a small circle, and therefore the emphasis brightness range in a large amount of set of pixels can obtain larger intensity gray scale value equalization space, done the stretching of grey decision-making again by HS technology, by originally because intensity gray scale value excessive close to and the image emphasis GTG section of the fuzzy a slice that seems; For the histogram of Figure 11, by the gray scale levels space of Le to L in figure and H to He part, reposition is placed into after compression 1/Z times, reposition is the space of 0 to Lnew and Hnew to 255 shown in Figure 12, the minimum gray scale angle value that in figure, Le and He represents whole raw video and most high gray angle value, it is less than gray scale levels degree is caused in Le and the region being greater than He waste because not using, therefore move in the lump in the space that this slatterns by the present invention can stretch zones.According to the characteristic of medical imaging gray-scale distribution, user (doctor) concentrates between histogram H and L for the position of diagnosis usually, therefore the region of shade of gray between H and L is ordered as wish Contrast enhanced region by this method, the region being less than L and being greater than H is inessential compressible region, after compressible region is compressed as shown in figure 12, add the gray scale levels space do not used end to end being originally less than Le and being greater than He, the scope between Hnew to the Lnew produced is the gray scale levels space vacateed.

About histogram distribution specific region stretching/compressing technology as described below: by judge for process pixel (K) be positioned at which region histogrammic, select corresponding mode process, be less than the gray scale levels part of L value and be greater than the gray scale levels part of H value, these two parts can do the action of space folding compression according to the folding ratio Z of input, be positioned at centre and be greater than L value to the part being less than H value, then do the action that histogram equalization stretches, the value of Hnew equation (9) and Lnew equation (10) need be used in this step.

[several 9]

[several 10]

In formula (9), the maximum gray value of Maxgray chart picture, a certain pixel in K representative graph, if a figure is 512 × 512, then K just has 262144, pixel (K) represents the gray scale levels degree of this pixel of K, and NewPixel (K) represents the gray scale levels degree that this pixel of K is new.Receive pixel (K), it is interval if judge, its value is positioned at histogrammic L < pixel (K) < H, this interval histogram will stretch and expand its scope to Lnew to Hnew to reach image contrast enhancing effect, its mode is as shown in equation (11), Pixel (K)-L represents that the GTG of this shade of gray and L is differential, then is multiplied by old scope and the stretch ratio of new range and get smallest positive integral, it is poor to calculate at the gray scale levels degree of new range and Lnew, finally adds that Lnew just can draw the new gray scale levels Pixel (K) of K.

[several 11]

Figure 13 is that region GTG HS expands and strengthens the process flow diagram of algorithm, and in this flow process, institute is assumed to [0..255] for the grey decision-making scope processed, therefore Maxgray is 255, but this method is also applicable to other different GTG codomain.First this flow process is read in the figure for process and is put into a matrix, input H, L, Z value: the wherein grey-scale range of H and L table for strengthening, the pixel energy making GTG fall within [L:H] scope borrows this method to produce the effect extending and strengthen, Z value is then in the gray-scale distribution of original image, and the part of non-emphasis scope (be less than L or be greater than H) is to the multiplying power of two pole compression.Then analyze the histogram of this picture and find out this figure minimum luminance value Le and maximum brightness value He, then calculating all number of pixels of this figure, to put into allelement stand-by, in process flow diagram, k represents certain pixel, pixel value pixel (k) of the pixel that following use for loop makes picture all sequentially enters formulae discovery (k is by 1 to allelement), judges that the in-scope of pixel value Pixel (k) does the calculating of the new grey decision-making of this point in alignment processing mode.K=k+1 represents this pixel and has calculated, and changes next pixel, and all pixels have all calculated its respective newpixel (k), are deposited in another matrix by new figure, have been the image of Contrast enhanced.

The each pixel of aforesaid image divides into the non-key area in key area and two ends by grey decision-making, user adopts the GTG of assigned picture ad-hoc location to be benchmark GTG by the selection of operator scheme, utilize queue (Queue) to store and determine the coordinate with reference point same block/object, whether pixel more adjacent is one by one subordinate to same block/object, and the adjacent coordinates meeting same block/object is restored in queue (Queue), process is until determine all pixel coordinates being subordinate to same block/object so repeatedly.If whether more adjacent pixel is subordinate to same block/object, first judge whether the grey decision-making of neighbor and the difference of adjacent reference point grey decision-making are less than a luminance difference threshold, if do not meet, the grey decision-making of this pixel and the grey decision-making of neighboring pixel are done average environmentally to close to adjust its grey decision-making, if the difference of this grey decision-making and adjacent reference point grey decision-making is less than luminance difference threshold, then also be subordinate to same block/object at last, and according to its original luminance value adjustment emphasis grey-scale range.This adjustment emphasis grey-scale range, refers to according to the pixel in judged block/object, and select minimum gray value to be wherein L parameter, maximum gray value is H parameter.

Figure 14 A and Figure 14 B is that the coordinate inputted according to user does regional luminance analysis, to produce the brightness range L of specified region/object, the calculation process flow diagram of H value, point coordinate (the x that its function is utilizing user to input, y) starting point is done, branch is utilized to limit (Branch-&-Bound) logic to four directions expansion to find the brightness range values (L of the region/object clicked, H), in this flow process for process image gray-scale level value scope be also assumed to [0..255], but following this method is also applicable to other different GTG codomain.And g (x, y) function is then as shown in equation (12), a in formula, b parameter is the interval term of reference of obfuscation of default, w [] matrix is then the interval range weighted ratio weight values of default, this equation is to point (x, y) the Fuzzy processing formula of adjacent block is done, make judging point (x, y) with adjoint point (x+i, the judgement of adaptability can be done when y+j) whether belonging to one block/object according to the luminance difference of surrounding, and tool this produce comparatively meet reality L, H value.

[several 12] $g(x,y)=\frac{{\mathrm{\&Sigma;}}_{s=-a}^a{\mathrm{\&Sigma;}}_{t=-b}^{b}w(s,t)f(x+s,y+t)}{{\mathrm{\&Sigma;}}_{s=-a}^a{\mathrm{\&Sigma;}}_{t=-b}^{b}w(s,t)}---\left(12\right)$

Relevant the present invention is applied to the conversion effect of the image that medical contrast apparatus obtains, refer to Figure 15 to 17, the image produced due to medical radiography equipment is mostly gray-scale figure, but because being limited to instrument performance, the image that medical treatment radiography equipment produces is often clear not, though details is taken but often made us being difficult to identification with visual because brightness (shade of gray) is too close, obviously can learn that the image after conversion is comparatively clear and be easy to identification by histogram local image Contrast enhanced device.

The image of Figure 15 is brain tomoscan (CT) shooting, obviously can find out that the image brain grey matter after Contrast enhanced (choosing is enclosed on right side) is clearly more demarcated with white matter (choosing place is enclosed in top) by figure below circle favored area, and the border of tumour (left side circle choosing place) and tumour inclusions more clear, this figure diagnoses emphasis to be to judge tumor boundaries, and inspect in tumour the blood vessel whether having calcification, fiberization or hyperplasia, after Contrast enhanced, tumor boundaries and content are clearly more demarcated, and focus all can be helped to judge.

The image of Figure 16 is also brain tomoscan (CT) shooting, this figure cardinal symptom is doubtfully have hemorrhage situation, the right temporal lobe of upper figure is doubtful have hemorrhage, figure below after Contrast enhanced, obviously can find out that right temporal lobe has obviously different (choosing place is enclosed on right side) from left temporal lobe lines color and luster shape, can conclude there is hemorrhage situation.

The image of Figure 17 is the human body lumbar vertebrae image of magnetic resonance imaging (MRI), upper figure image cannot be known and sees whether endocranium epicoele has foreign matter to oppress, utilize the image processing of PACS that script can be made to divide compared with highlights more excessively bright, and middle endocranium epicoele cannot be known on the contrary and affects interpretation, the image of the figure below after adjustment can be known and sees endocranium epicoele (central authorities' circle choosing place) and oppressed nerve root (arrow).

Claims (6)

1. a histogram local image Contrast enhanced method, is characterized in that, include:

Obtain an image;

Add up the histogram distribution of this image, each for image pixel is divided into key area and non-key area by grey decision-making;

The non-key area of this histogram distribution is compressed, obtains enough GTG spaces to provide this key area to carry out the Contrast enhanced of ductility to specific gray-scale areas; And

This region gray scale levels degree widened and improves luminance difference, and strengthening the contrast of its area image, to improve details and the sharpening of image.

2. histogram local image Contrast enhanced method as claimed in claim 1, is characterized in that, compress, pass through formula to the non-key area at histogram two ends with

key area is made to extend expansion to two ends.

3. histogram local image Contrast enhanced method as claimed in claim 2, it is characterized in that, compress the GTG space of the non-emphasis brightness range of non-key area, the pixel of non-key area is made to be expressed among a small circle, and the emphasis brightness range in a large amount of set of pixels obtains larger intensity gray scale value equalization space, and the stretching of grey decision-making is done by HS technology, pass through formula

Be converted to the GTG of larger section, make grey decision-making expanded range.

4. as the histogram local image Contrast enhanced method as described in arbitrary in claim 1-3, it is characterized in that, after each for image pixel is divided into the non-key area in key area and two ends by grey decision-making, user adopts the GTG of assigned picture ad-hoc location to be benchmark GTG by the selection of operator scheme, stored by a queue (Queue) and determined the coordinate with reference point same block/object, whether pixel more adjacent is one by one subordinate to same block/object, and the adjacent coordinates meeting same block/object is restored in this queue (Queue), and repeatedly process until determine all pixel coordinates being subordinate to same block/object.

5. histogram local image Contrast enhanced method as claimed in claim 4, it is characterized in that, whether more adjacent pixel is subordinate to same block/object, first judge whether the grey decision-making of neighbor and the difference of adjacent reference point grey decision-making are less than a luminance difference threshold, if do not meet, the grey decision-making of this pixel and the grey decision-making of neighboring pixel are done on average, environmentally to close its grey decision-making of adjustment, if the difference of this grey decision-making and adjacent reference point grey decision-making is less than luminance difference threshold, then also be subordinate to same block/object at last, and according to its original luminance value adjustment emphasis grey-scale range, this adjustment emphasis grey-scale range refers to according to the pixel in judged block/object, minimum gray value is wherein selected to be L parameter, maximum gray value is H parameter.

6. a histogram local image Contrast enhanced device, is characterized in that, include:

One input video conversion module, it is in order to input an image and to be converted to a YCbCr signal;

One regional luminance strengthens module, and it be electrically connected with input video conversion module, and this regional luminance strengthens module cluster according to the individual control signals of majority, changes and YCbCr signal after producing an adjustment;

One image output modular converter, itself and regional luminance strengthen module and are electrically connected, and this input video conversion module exports and strengthens the YCbCr signal after module adjustment through regional luminance;

One image bright territory inputting interface, itself and regional luminance strengthen module and are electrically connected, this image bright territory inputting interface in order to selected more than one the shade of gray scope (H, L) for enhancing, or inputs more than one pixel coordinate (X, Y) and grey-scale range threshold (Z) thereof;

One coordinates regional luminance analyzing unit, it is electrically connected with image bright territory inputting interface, and pixel coordinate (X, Y) and the grey-scale range threshold (Z) of this coordinates regional luminance analyzing unit foundation image bright territory inputting interface produce an easily extensible value range (Hnew, Lnew);

One bright territory generator, it strengthens module, coordinates regional luminance analyzing unit and image bright territory inputting interface be electrically connected with input video conversion module, regional luminance respectively, this bright territory generator is according to the lightness (Y) in YCbCr signal, set up the probability density function of this signal of video signal, the shade of gray scope (H, L) that image bright territory inputting interface or coordinates regional luminance analyzing unit produce, provides regional luminance to strengthen module according to this extendible scope (Hnew, Lnew) and carries out Contrast enhanced process.