CN101706953A - Histogram equalization based image enhancement method and device - Google Patents

Abstract

The invention discloses a histogram equalization based image enhancement method and device. In the invention, the whole-frame input images are subjected to histogram equalization by sectors; therefore, by carrying out histogram equalization on the sector singly, the layering in the sector can be enhanced and the expressive forces of the edges which are rich in changes of gray values and the detailed information can be improved. Hereafter, interpolation processing is carried out on the pixel gray values of the edges of each sector and the adjacent sectors around the sector so that the sectors are related to avoid obvious boundaries among the sectors as each sector is respectively subjected to histogram equalization, thus improving the quality of the enhanced images. In addition, the detailed information of the edges is further enhanced by adjusting the local contrast of the enhanced input images so as to improve the contrast of the images after equalization, thus further improving the quality of the enhanced images.

Description

Image enchancing method and device based on histogram equalization

Technical field

The present invention relates to image enhancement technique, particularly a kind of image enchancing method and a kind of image intensifier device based on histogram equalization based on histogram equalization.

Background technology

The figure image intensifying is the basic means of Flame Image Process, and it is toward the preprocessing process that firmly is various graphical analyses when handling.The method of figure image intensifying generally is divided into spatial domain and transform domain two big classes, and histogram equalization is one of the most frequently used, most important algorithm in the spatial domain figure image intensifying.

Histogram equalization is done the basis with probability theory, to have the whole two field picture that different gray-value pixel point constitutes, be converted to the histogram of the pixel statistics of different gray-scale values in the whole two field picture, and then to revising the grey scale pixel value of putting in order in the two field picture by histogrammic equalization is handled, make the pixel quantity that drops between different gray areas in the whole two field picture as far as possible on average and between each gray area pixel be arranged all, thereby by making whole two field picture have the purpose that stronger stereovision reaches the figure image intensifying.

Yet for whole two field picture, it includes fringe region, the details area of enriching gray-value variation, might lose owing to the gray-scale value equilibrium to whole two field picture, thereby make the picture quality after the enhancing not high.

Summary of the invention

In view of this, the invention provides a kind of image enchancing method and a kind of image intensifier device, can improve the picture quality after the enhancing based on histogram equalization based on histogram equalization.

A kind of figure provided by the invention comprises based on the image intensifying method of histogram equalization:

A, respectively histogram equalization is carried out in each zone in the input picture and handle;

B, respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

Described step a comprises:

A1, add up the pixel quantity of each gray-scale value in each zone respectively, obtain in each zone the histogram sequence of each gray-value pixel quantity in this zone of expression;

A2, to the operation that adds up of each regional histogram sequence, obtain the histogram accumulated sequence of each gray-value pixel cumulative distribution of expression in each zone;

A3, calculate the corresponding following equalization mapping table in each zone according to each regional histogram accumulated sequence:

$\left\{\mathrm{Mapping}\right[\mathrm{Value}{]}_n\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}\right[\mathrm{Value}{]}_n\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray value interval among Value_Low～Value_High, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone,

Hist[t] _nBe the pixel quantity of t gray-scale value in n the zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value, and n is greater than 1 and smaller or equal to the zone sum;

A4, utilize the corresponding equalization mapping table in each zone respectively, the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone.

Value_Low gets 0, and Value_High gets 255.

Described step b comprises: be the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different is carried out interpolation processing.

Respectively will be divided in each zone four sub-pieces equating of ranks number, and four adjacent sub-blocks that are positioned at diagonal angle splicing place in the adjacent area to the splicing of per four diagonal angles are carried out interpolation processing.

Each grey scale pixel value in each sub-piece at place, splicing angle, carry out interpolation processing in the following manner:

L′ _i，j＝rowRevW×[colRevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]

+rowW×[colRevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent the grey scale pixel value after the capable j row of the i interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, and colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) _UL, Mapping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the grey scale pixel value of the capable j row of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

Input picture is divided into 8 * 8 totally 64 zones.

After the described step b, this method further comprises: c, all grey scale pixel values in the input picture are carried out the local contrast adjustment.

Described step c carries out the local contrast adjustment in the following manner each local adjustment in the window:

x′ _p，q＝Avr+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}Be that the part adjusts the gray-scale value of the capable q row of p pixel after interpolation processing in the window, Avr be the local balanced average gray of all pixels after interpolation processing in the window of adjusting, and α is the adjustment coefficient preset and α 〉=1.

Local adjustment window is 3 * 3 window.

A kind of image intensifier device based on histogram equalization provided by the invention comprises:

The region histogram balance module is used for respectively histogram equalization being carried out in each zone of input picture and handles;

Neighbouring region interpolation processing module, be used for respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

Described region histogram balance module comprises:

Sequence is added up submodule, is used for adding up respectively the pixel quantity of each each gray-scale value of zone, obtains representing in each zone the histogram sequence of each gray-value pixel quantity in this zone;

Sequence accumulation submodule is used for the operation that adds up of each regional histogram sequence is obtained representing in each zone the histogram accumulated sequence of each gray-value pixel cumulative distribution;

Submodule is set up in mapping, is used for calculating the corresponding following equalization mapping table in each zone according to each regional histogram accumulated sequence:

$\left\{\mathrm{Mapping}\right[\mathrm{Value}{]}_n\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}\right[\mathrm{Value}{]}_n\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray value interval among Value_Low～Value_High, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone, Hist[t] _nBe the pixel quantity of t gray-scale value in n the zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value, and n is greater than 1 and smaller or equal to the zone sum;

Balanced mapping submodule is used for utilizing respectively each regional corresponding equalization mapping table, and the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone.

Value_Low gets 0, and Value_High gets 255.

Described neighbouring region interpolation processing module is the plurality of sub piece with each area dividing respectively, and is the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different is carried out interpolation processing.

Described neighbouring region interpolation processing module respectively will be divided in each zone four sub-pieces equating of ranks number, and four adjacent sub-blocks that are positioned at diagonal angle splicing place in the adjacent area to the splicing of per four diagonal angles are carried out interpolation processing.

Described neighbouring region interpolation processing module is carried out interpolation processing in the following manner for each grey scale pixel value in each sub-piece at place, splicing angle:

L′ _i，j＝rowRevW×[colRevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]+rowW×[colRevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent the grey scale pixel value after the capable j row of the i interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, and colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) _UL, Mapping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the grey scale pixel value of the capable j row of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

Input picture is divided into 8 * 8 totally 64 zones.

Further comprise: the local contrast adjusting module is used for all grey scale pixel values of input picture are carried out the local contrast adjustment.

Described local contrast adjusting module carries out the local contrast adjustment in the following manner each local adjustment in the window:

x′ _p，q＝Avr+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}For adjusting the gray-scale value of the capable q row of p pixel after interpolation processing in the window in the part, Avr is the local balanced average gray of all pixels after interpolation processing in the window of adjusting, and α is default adjustment coefficient and α 〉=1.

Local adjustment window is 3 * 3 window.

As seen from the above technical solution, the present invention carries out histogram equalization rather than whole two field picture is carried out histogram equalization whole frame input picture subregion, if the pixel in the zone is too concentrated between some gray area, then the stereovision of image meeting expressive force relatively poor, the edge of image detailed information is not high, therefore, will make stereovision in this zone strengthen and can improve to have the edge that enriches gray-value variation and the expressive force of detailed information by separately histogram equalization being carried out in this zone; After this, more respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing, can make each interregionally have correlativity, avoid because handle and on the obvious border of interregional generation through histogram equalization respectively in each zone.Thus, the input picture of input picture after the expressive force of the edge details information behind the histogram equalization can be enhanced and strengthen has smoothness preferably, improved the quality that strengthens image than existing mode of losing edge details information.

Further, though the subregion histogram equalization can be lost edge details information as few as possible, but because the subregion histogram equalization has also significantly improved the vision dynamic range of whole frame input picture, thereby can make edge details outstanding inadequately to a certain extent, therefore, the present invention also can by the input picture after strengthening is carried out the local contrast adjustment, to improve the contrast of image after the equalization, make the part that strengthens image have bigger dynamic range, make part such as edge details information more outstanding, thereby can further improve the quality that strengthens image.

And the present invention realizes simply, and hardware costs is little, and can embed in the chip.

Description of drawings

Fig. 1 is the exemplary process diagram of the embodiment of the invention based on the image enchancing method of histogram equalization;

Fig. 2 is in the embodiment of the invention being divided into adjacent area the synoptic diagram of sub-piece;

Fig. 3 is the exemplary block diagram of the embodiment of the invention based on the image intensifier device of histogram equalization.

Embodiment

For making purpose of the present invention, technical scheme and advantage clearer, below with reference to the accompanying drawing embodiment that develops simultaneously, the present invention is described in more detail.

In the embodiment of the invention, whole frame input picture subregion is carried out histogram equalization rather than whole two field picture is carried out histogram equalization, if the pixel in the zone is too concentrated between some gray area, then the stereovision of image meeting expressive force relatively poor, the edge of image detailed information is not high, like this, will make stereovision in this zone strengthen and can improve to have the edge that enriches gray-value variation and the expressive force of detailed information by separately histogram equalization being carried out in this zone.

Certainly, because the grey scale pixel value in the one's respective area is only considered in each zone when carrying out histogram equalization respectively, thereby each zone is respectively after histogram equalization is handled, easily on the tangible border of interregional generation, for make each interregionally have correlativity, to avoid since each zone handle and on the obvious border of interregional generation through histogram equalization respectively, the embodiment of the invention also can be respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

Thus, the input picture of input picture after can not losing edge details information and enhancing behind the histogram equalization has smoothness preferably, improved the quality that strengthens image than existing mode of losing edge details information.

In addition, above-mentioned subregion histogram equalization, and the interpolation processing of adjacent area edge, though can lose edge details information as few as possible, promptly improve the quality that strengthens image by many preserving edges detailed information, but because the subregion histogram equalization also can significantly improve the vision dynamic range of whole frame input picture, thereby can make edge details outstanding inadequately to a certain extent, thus, the present invention also can be by carrying out the local contrast adjustment to the input picture after strengthening, contrast with image after the raising equalization, make the part that strengthens image have bigger dynamic range, make part such as edge details information more outstanding, thereby can further improve the quality that strengthens image.

Below, at first the image enchancing method based on histogram equalization in the embodiment of the invention is elaborated.

Fig. 1 is the exemplary process diagram of the embodiment of the invention based on the image enchancing method of histogram equalization.As shown in Figure 1, the image enchancing method based on histogram equalization comprises the steps: in the present embodiment

Step 101 is carried out histogram equalization to each zone in the input picture respectively and is handled.

Before this step, can also carry out pre-service such as denoising to input picture.

In this step, at first input picture can be divided into M * N zone, then statistics with histogram be carried out in each zone, preferably select M=N=8 for use, can carry out histogram equalization then in the following manner and handle:

1) add up the pixel quantity Hist[Value of each gray-scale value, respectively], wherein, Value represents gray-scale value.Correspondingly, each gray-value pixel quantity Hist[Value in each zone] can constitute the histogram sequence that constitutes in proper order by each gray-value pixel quantity in n the zone Hist[Value] _n, n is greater than 1 and smaller or equal to regional sum M * N.

2), respectively to n the zone histogram sequence Hist[Value] _nThe operation that adds up, obtain representing each gray-value pixel cumulative distribution the histogram accumulated sequence that increases progressively HistSum[Value] _n, each element in this sequence

Hist[t] _nIt is the pixel quantity of t gray-scale value in n the zone.

For example, for a histogram sequence Hist[Value] _n}={ 0,1,2,4,5,7,6 ..., the histogram accumulated sequence that increases progressively that it obtains HistSum[Value] _n}={ 1,3,7,12,19,25,26 ....

In chip design, can utilize the little characteristics of variation in two two field pictures, the subregion statistics with histogram of the current input image that obtains in this step is handled for next frame used.

3), calculate corresponding equalization mapping table according to each regional histogram accumulated sequence:

$\left\{\mathrm{Mapping}\right[\mathrm{Value}{]}_n\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}\right[\mathrm{Value}{]}_n\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray-scale value among Value_Low～Value_High, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value.Preferably, all have maximum dynamic range in order to make each zone after equalization is handled, Value_Low can get 0, and Value_High can get 255.

Suppose, gray-scale value 0～128 interval interior pixel quantity proportion is excessive, so, behind equalization mapping table correction grey scale pixel value, gray-scale value 0～128 interval interior grey scale pixel value near 128 can surpass 128, thereby it is to greater than in other interval of 128, balanced thereby the pixel quantity in making between each gray area is tending towards by balanced.

4) utilize the corresponding equalization mapping table in each zone, respectively, the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone, so that obtaining equalization, handles the grey scale pixel value in should the zone, thereby make the pixel quantity that drops in each zone between different gray areas as far as possible on average and between each gray area pixel be arranged all, thereby by making the edge details information that may have in each zone have the purpose that stronger stereovision reaches the figure image intensifying.

Certainly, though in the prior art not the subregion carry out histogram equalization and handle, how to realize the histogram equalization in this step, can also adopt other algorithm arbitrarily by those skilled in the art, enumerate no longer one by one at this.

Step 102, grey scale pixel value regional and adjacent area edge around it carries out interpolation processing to each respectively.

In this step, can be the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different is carried out interpolation processing.More specifically, as shown in Figure 2, can be respectively with 2 * 2 sub-pieces of totally 4 as dotted line division that are divided in each zone shown in solid line that the pixel column columns equates, and in the adjacent area to the splicing of per 4 diagonal angles, 4 adjacent sub-blocks that are arranged in diagonal angle splicing place of being drawn as Fig. 2 arrow are carried out interpolation processing, promptly the adjacent upper left prescription piece UL of per 4 diagonal angle splicings place among Fig. 2, upper right prescription piece UR, lower-left prescription piece BL, bottom right prescription piece BR are carried out interpolation processing.Need to prove,, can not handle or adopt any suitable processing mode to handle in the edge for the sub-piece that does not indicate UL, UR, BL, BR in the fringe region.

And, carry out interpolation processing in the following manner for each grey scale pixel value of each sub-piece among adjacent UL, the UR of 4 diagonal angle splicings place, BL, the BR:

L′ _i，j＝rowRevW×[colRevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]+rowW×[colRevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent that the capable j row of i are earlier after the grey scale pixel value after histogram equalization and the interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) _UL, Mapping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the capable j of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece and be listed as grey scale pixel value through histogram equalization;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

Utilize the row interpolation matrix of coefficients rowW of above-mentioned both forward and reverse directions and row interpolation coefficient matrix colW, the colRevW of rowRevW and both forward and reverse directions, can make that the picture element interpolation coefficient the closer to splicing top, angle is big more in each sub-piece, the picture element interpolation coefficient away from splicing top, angle is more little more.

And, because the pixel in each sub-piece is when interpolation arithmetic, use with needs, and only use once the rest of pixels of self place row and column, thereby among the row interpolation matrix of coefficients rowW and rowRevW of both forward and reverse directions, can only there be the interpolation coefficient of a matrix to choose 1～rows, another then can only choose 0～rows-1, the row interpolation coefficient matrix colW of both forward and reverse directions, among the colRevW, can only there be the interpolation coefficient of a matrix to choose 1～cols, another then can only choose cols-1, thereby can avoid reusing when interpolation processing the rest of pixels gray-scale value of each pixel place row and column.

Step 103 is carried out the local contrast adjustment to all grey scale pixel values in the input picture.

This step is optional step (illustrating with frame of broken lines in Fig. 1), and can carry out the local contrast adjustment in the window in the following manner each local adjustment in this step:

x′ _p，q＝Avr+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}For adjusting in the window the capable q row of p pixel earlier after the gray-scale value after histogram equalization and the interpolation processing in the part, can be with the L ' that obtains in the step 102 _{I, j}As x _{P, q}, Avr is the local balanced average gray of all pixels after interpolation processing in the window of adjusting, α is default adjustment coefficient and α 〉=1.

As above as seen, when α＞1, if Avr＞x _{P, q}, x ' then _{P, q}＞x _{P, q}If Avr＜x _{P, q}, x ' then _{P, q}＜x _{P, q}Thereby realized the details enhancing, mainly played the local local contrast humidification of adjusting in the window in other words, but do not regulate the dynamic range of view picture input picture substantially. but just so, the subregion histogram equalization can be remedied and the outstanding inadequately defective of edge details can be caused to a certain extent owing to significantly improve the vision dynamic range of whole frame input picture, thereby make present embodiment can not only regulate dynamic range and preserving edge detailed information, but also can strengthen local contrast, thereby make that the image effect that strengthens is better.

Preferably, in order to make the local contrast adjustment more careful, it is 3 * 3 window that the local window of adjusting can be set.

So far, above-mentioned flow process finishes.

Below, again the image intensifier device based on histogram equalization in the present embodiment is elaborated.

Fig. 3 is the exemplary block diagram of the embodiment of the invention based on the image intensifier device of histogram equalization.As shown in Figure 3, the image intensifier device based on histogram equalization comprises in the present embodiment: region histogram balance module, neighbouring region interpolation processing module, local contrast adjusting module.

The region histogram balance module is used for respectively histogram equalization being carried out in each zone of input picture and handles.

Specifically, the region histogram balance module can comprise sequence statistics submodule, sequence accumulation submodule, shine upon and set up submodule and balanced mapping submodule (not shown among Fig. 3), wherein:

Sequence statistics submodule, be used for adding up respectively the pixel quantity Hist[Value of each gray-scale value], Value represents gray-scale value, correspondingly, each gray-value pixel quantity Hist[Value in each zone] can constitute the histogram sequence that constitutes in proper order by each gray-value pixel quantity in n the zone Hist[Value] _n, n is greater than 1 and smaller or equal to regional sum M * N;

Sequence accumulation submodule, be used for respectively to n regional histogram sequence Hist[Value] _nThe operation that adds up, obtain representing each gray-value pixel cumulative distribution the histogram accumulated sequence that increases progressively HistSum[Value] _n, each element in this sequence Hist[t] _nIt is the pixel quantity of t gray-scale value in n the zone;

Submodule is set up in mapping, is used for calculating corresponding equalization mapping table according to the histogram accumulated sequence that increases progressively in each zone:

$\left\{\mathrm{Mapping}\right[\mathrm{Value}{]}_n\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}\right[\mathrm{Value}{]}_n\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray-scale value in 0～255, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value, preferably, in order to make each zone after equalization is handled all have maximum dynamic range, Value_Low can get 0, and Value_High can get 255;

Balanced mapping subelement is used for utilizing respectively each regional corresponding equalization mapping table, and the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone.

Neighbouring region interpolation processing module, be used for respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

In the practical application, neighbouring region interpolation processing module can be the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different carried out interpolation processing. more specifically, as shown in Figure 2, can be respectively with 2 * 2 sub-pieces of totally 4 as dotted line division that are divided in each zone shown in solid line that the pixel column columns equates, and in the adjacent area to the splicing of per 4 diagonal angles, 4 adjacent sub-blocks that are arranged in diagonal angle splicing place of being drawn as Fig. 2 arrow are carried out interpolation processing, promptly to the adjacent upper left prescription piece UL of per 4 diagonal angle splicings place among Fig. 2, upper right prescription piece UR, lower-left prescription piece BL, bottom right prescription piece BR carries out interpolation processing. need to prove, for not indicating UL in the fringe region, UR, BL, the sub-piece of BR can not handled, or adopt any suitable processing mode to handle in the edge.

And for each grey scale pixel value of each sub-piece among adjacent UL, the UR of 4 diagonal angle splicings place, BL, the BR, neighbouring region interpolation processing module can be carried out interpolation processing in the following manner:

L′ _i，j＝rowRevW×[colRevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]

+rowW×[colRevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent that the capable j row of i are earlier after the grey scale pixel value after histogram equalization and the interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) _UL, Mapping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the capable j of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece and be listed as grey scale pixel value through histogram equalization;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

Utilize the row interpolation matrix of coefficients rowW of above-mentioned both forward and reverse directions and row interpolation coefficient matrix colW, the colRevW of rowRevW and both forward and reverse directions, can make that the picture element interpolation coefficient the closer to splicing top, angle is big more in each sub-piece, the picture element interpolation coefficient away from splicing top, angle is more little more.

And, because the pixel in each sub-piece is when interpolation arithmetic, use with needs, and only use once the rest of pixels of self place row and column, thereby among the row interpolation matrix of coefficients rowW and rowRevW of both forward and reverse directions, can only there be the interpolation coefficient of a matrix to choose 1～rows, another then can only choose 0～rows-1, the row interpolation coefficient matrix colW of both forward and reverse directions, among the colRevW, can only there be the interpolation coefficient of a matrix to choose 1～cols, another then can only choose cols-1, thereby can avoid reusing when interpolation processing the rest of pixels gray-scale value of each pixel place row and column.

The local contrast adjusting module for optional functional module (illustrating with frame of broken lines in Fig. 3), is used for all grey scale pixel values of input picture are carried out the local contrast adjustment.

In the practical application, the local contrast adjusting module can carry out the local contrast adjustment in the following manner each local adjustment in the window:

x′ _p，q＝Avr+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}For adjusting in the window the capable q row of p pixel earlier after the gray-scale value after histogram equalization and the interpolation processing, the L ' that neighbouring region interpolation processing module can be obtained in the part _{I, j}As x _{P, q}, Avr is the local balanced average gray of all pixels after interpolation processing in the window of adjusting, α is default adjustment coefficient and α 〉=1.

As above as seen, when α＞1, if Avr＞x _{P, q}, x ' then _{P, q}＞x _{P, q}If Avr＜x _{P, q}, x ' then _{P, q}＜x _{P, q}Thereby, realize the details enhancing, mainly played the local local contrast humidification of adjusting in the window in other words, but the dynamic range of not regulating the view picture input picture substantially.But just so, the subregion histogram equalization can be remedied and the outstanding inadequately defective of edge details can be caused to a certain extent owing to significantly improve the vision dynamic range of whole frame input picture, thereby make present embodiment can not only regulate dynamic range and preserving edge detailed information, but also can strengthen local contrast, thereby make that the image effect that strengthens is better.

Preferably, in order to make the local contrast adjustment more careful, it is 3 * 3 window that the local window of adjusting can be set.

The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to and replace and improvement etc., all should be included within protection scope of the present invention.

Claims (20)

1. the image enchancing method based on histogram equalization is characterized in that, comprises the steps:

A, respectively histogram equalization is carried out in each zone in the input picture and handle;

B, respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

2. image enchancing method as claimed in claim 1 is characterized in that, described step a comprises:

A1, add up the pixel quantity of each gray-scale value in each zone respectively, obtain in each zone the histogram sequence of each gray-value pixel quantity in this zone of expression;

A2, to the operation that adds up of each regional histogram sequence, obtain the histogram accumulated sequence of each gray-value pixel cumulative distribution of expression in each zone;

A3, calculate the corresponding following equalization mapping table in each zone according to each regional histogram accumulated sequence:

$\left\{\mathrm{Mapping}{\left[\mathrm{Value}\right]}_n\right\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}{\left[\mathrm{Value}\right]}_n\right\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray value interval among Value_Low～Value_High, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone,

Hist[t] _nBe the pixel quantity of t gray-scale value in n the zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value, and n is greater than 1 and smaller or equal to the zone sum;

A4, utilize the corresponding equalization mapping table in each zone respectively, the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone.

3. image enchancing method as claimed in claim 2 is characterized in that Value_Low gets 0, and Value_High gets 255.

4. image enchancing method as claimed in claim 2 is characterized in that, described step b comprises: be the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different is carried out interpolation processing.

5. image enchancing method as claimed in claim 4 is characterized in that, respectively will be divided in each zone four sub-pieces equating of ranks number, and four adjacent sub-blocks that are positioned at diagonal angle splicing place in the adjacent area to the splicing of per four diagonal angles are carried out interpolation processing.

6. image enchancing method as claimed in claim 5 is characterized in that, each grey scale pixel value in each sub-piece at place, splicing angle carries out interpolation processing in the following manner:

L′ _i，j＝row?RevW×[col?RevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]

+rowW×[col?RevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent the grey scale pixel value after the capable j row of the i interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, and colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) _UL, Mapping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the grey scale pixel value of the capable j row of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

7. as each described image enchancing method in the claim 1 to 6, it is characterized in that input picture is divided into 8 * 8 totally 64 zones.

8. as each described image enchancing method in the claim 1 to 6, it is characterized in that after the described step b, this method further comprises: c, all grey scale pixel values in the input picture are carried out the local contrast adjustment.

9. image enchancing method as claimed in claim 8 is characterized in that, described step c carries out the local contrast adjustment in the following manner each local adjustment in the window:

x′ _p，q＝Avr′+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}For adjusting the gray-scale value of the capable q row of p pixel after interpolation processing in the window in the part, Avr is the local balanced average gray of all pixels after interpolation processing in the window of adjusting, and α is default adjustment coefficient and α 〉=1.

10. image enchancing method as claimed in claim 9 is characterized in that, local adjustment window is 3 * 3 window.

11. the image intensifier device based on histogram equalization is characterized in that, comprising:

The region histogram balance module is used for respectively histogram equalization being carried out in each zone of input picture and handles;

Neighbouring region interpolation processing module, be used for respectively to each zone with its around the grey scale pixel value of adjacent area edge carry out interpolation processing.

12. image intensifier device as claimed in claim 11 is characterized in that, described region histogram balance module comprises:

Sequence is added up submodule, is used for adding up respectively the pixel quantity of each each gray-scale value of zone, obtains representing in each zone the histogram sequence of each gray-value pixel quantity in this zone;

Sequence accumulation submodule is used for the operation that adds up of each regional histogram sequence is obtained representing in each zone the histogram accumulated sequence of each gray-value pixel cumulative distribution;

Submodule is set up in mapping, is used for calculating the corresponding following equalization mapping table in each zone according to each regional histogram accumulated sequence:

$\left\{\mathrm{Mapping}{\left[\mathrm{Value}\right]}_n\right\}=\mathrm{Value}\_\mathrm{Low}+\frac{\left\{\mathrm{HistSum}{\left[\mathrm{Value}\right]}_n\right\}}{{\mathrm{AllPixels}}_n}(\mathrm{Value}\_\mathrm{High}-\mathrm{Value}\_\mathrm{Low})$

Wherein, Value is any gray value interval among Value_Low～Value_High, Mapping[Value] _nBe n the equalization mapping table that the zone is corresponding, HistSum[Value] _nBe the histogram accumulated sequence in n zone, Hist[t] _nBe the pixel quantity of t gray-scale value in n the zone, AllPixels _nBe the sum of all pixels in n the zone, Value_Low is default equalization minimum value, and Value_High is default even weighing apparatus maximal value, and n is greater than 1 and smaller or equal to the zone sum;

Balanced mapping submodule is used for utilizing respectively each regional corresponding equalization mapping table, and the grey scale pixel value in this zone is modified to Mapping (L _{X, y}), L _{X, y}Represent the grey scale pixel value after the capable y row of the x interpolation processing in each zone.

13. image intensifier device as claimed in claim 12 is characterized in that Value_Low gets 0, Value_High gets 255.

14. image intensifier device as claimed in claim 12, it is characterized in that, described neighbouring region interpolation processing module is the plurality of sub piece with each area dividing respectively, and is the plurality of sub piece with each area dividing respectively, and the adjacent sub-blocks that is positioned at zones of different is carried out interpolation processing.

15. image intensifier device as claimed in claim 14, it is characterized in that, described neighbouring region interpolation processing module respectively will be divided in each zone four sub-pieces equating of ranks number, and four adjacent sub-blocks that are positioned at diagonal angle splicing place in the adjacent area to the splicing of per four diagonal angles are carried out interpolation processing.

16. image intensifier device as claimed in claim 15 is characterized in that, described neighbouring region interpolation processing module is carried out interpolation processing in the following manner for each grey scale pixel value in each sub-piece at place, splicing angle:

L′ _i，j＝row?RevW×[col?RevW×Mapping(L _i，j) _UL+colW×Mapping(L _i，j) _UR]

+rowW×[col?RevW×Mapping(L _i，j) _BL+colW×Mapping(L _i，j) _BR]

Wherein, L ' _{I, j}Represent the grey scale pixel value after the capable j row of the i interpolation processing in described each sub-piece, rowW, rowRevW are respectively the row interpolation matrix of coefficients of both forward and reverse directions, and colW, colRevW are respectively the row interpolation coefficient matrix of both forward and reverse directions, Mapping (L _{I, j}) U _{L, M}Apping (L _{I, j}) _UR, Mapping (L _{I, j}) _BL, Mapping (L _{I, j}) _BRBe respectively the grey scale pixel value of the capable j row of i in upper left side block, upper right prescription piece, lower-left prescription piece, the bottom right prescription piece;

$\mathrm{rowRevW}=\left[\begin{array}{cccc}\mathrm{Rows}& \mathrm{Rows}& ...& \mathrm{Rows}\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\\ ...& ...& ...& ...\\ 1& 1& ...& 1\end{array}\right],$

$\mathrm{rowW}=\left[\begin{array}{cccc}0& 0& ...& 0\\ 1& 1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{Rows}-1& \mathrm{Rows}-1& ...& \mathrm{Rows}-1\end{array}\right],$

$\mathrm{colRevW}=\left[\begin{array}{cccc}\mathrm{cols}& \mathrm{cols}-1& ...& 1\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\\ ...& ...& ...& ...\\ \mathrm{cols}& \mathrm{cols}-1& ...& 1\end{array}\right],$

$\mathrm{colW}=\left[\begin{array}{cccc}0& 1& ...& \mathrm{cols}-1\\ 0& 1& ...& \mathrm{cols}-1\\ ...& ...& ...& ...\\ 0& 1& ...& \mathrm{cols}-1\end{array}\right],$

Rows in above-mentioned each matrix is the line number of each sub-piece, and cols is the columns of each sub-piece.

17., it is characterized in that input picture is divided into 8 * 8 totally 64 zones as each described image intensifier device in the claim 11 to 16.

18. as each described image intensifier device in the claim 11 to 16, it is characterized in that, further comprise: the local contrast adjusting module is used for all grey scale pixel values of input picture are carried out the local contrast adjustment.

19. image intensifier device as claimed in claim 18 is characterized in that, described local contrast adjusting module carries out the local contrast adjustment in the following manner each local adjustment in the window:

x′ _p，q＝Avr+α(x _p，q-Avr)

Wherein, x ' _{P, q}For adjusting in the window the capable q row of p pixel through the adjusted gray-scale value of local contrast, x in the part _{P, q}For adjusting the gray-scale value of the capable q row of p pixel after interpolation processing in the window in the part, Avr is the local balanced average gray of all pixels after interpolation processing in the window of adjusting, and α is default adjustment coefficient and α 〉=1.

20. image intensifier device as claimed in claim 19 is characterized in that, local adjustment window is 3 * 3 window.

Images