CN101882305B - Method for enhancing image - Google Patents

Abstract

The embodiment of the invention discloses a method for enhancing an image, which comprises the following steps of: performing bi-dimensional empirical mode decomposition (BEMD) on an original image to obtain an intrinsic mode function (IMF) of the image; performing wavelet transformation on a high-frequency IMF in the IMF; performing 5 layer wavelet transformation on the IMF and respectively obtaining a wavelet coefficient of each IMF component; performing wavelet adaptive de-noising on the high-frequency IMF after the wavelet transformation, a first layer using dual threshold enhancement, a second and a third layers using self-adaption enhancement, a fourth and a fifth layer using single threshold enhancement; performing wavelet inverse transformation on the high-frequency IMF after the wavelet adaptive de-noising; and synthesizing a low-frequency IMF in the IMF and the high-frequency IMF after the wavelet inverse transformation to obtain an enhanced image. By implementing the method for enhancing the image, the image can be enhanced while removing image noises, and useful information is highlighted.

Description

A kind of method of image enhancement processing

Technical field

The present invention relates to technical field of image processing, be specifically related to a kind of method of image enhancement processing.

Background technology

Figure image intensifying is the importance that image is processed, it mainly adopts specific Enhancement Method to give prominence to some information in image according to the ambiguity of concrete application scenarios and image, weaken simultaneously or eliminate irrelevant information, to reach, emphasizing the integral body of image or the object of local feature.Along with the widespread use of image, figure image intensifying has become one of urgent problem of key.

Algorithm for image enhancement is more at present, and conventional image enhancement processing mode comprises greyscale transformation, histogram modification, image sharpening, noise remove, geometric distortion correction, frequency domain filtering, wavelet transformation etc.These algorithms are divided into algorithm for image enhancement and the algorithm for image enhancement based on frequency domain based on time domain according to the difference of processing space.Time domain refers to image itself, and the figure image intensifying based on time domain is to be directly treated to basis to the pixel of image; Algorithm for image enhancement based on frequency domain take revise image Fourier transform as basis, the conversion coefficient of image (frequency components) is directly carried out to computing, then by Fourier inverse transformation to obtain the enhancing effect of image.The concrete technical scheme of these methods is as follows:

(1) greyscale transformation.Any r value in intensity profile region changes: wherein, T strengthens operation to s=T (r), and T (r) changes the gray-level of image, and method has image to negate, strengthen contrast and dynamic range compression etc.Conventional compress mode is the conversion T that adopts logarithmic form, s=clog (1+|r|).The method can increase the whole contrast of image, but the denoising that cannot remove image.

(2) histogram modification (histogram equalization).Histogram equalization (histogram equalization) thus be a kind ofly by histogram transformation, to realize the method that grey scale mapping reaches figure image intensifying object.Histogrammic horizontal ordinate is gray level, and ordinate is the frequency that this gray level occurs; Histogram is the most basic statistical nature of image, and the intensity profile situation of original image is provided, and the integral body that has also provided piece image gray-scale value is described.Basic thought: the histogram transformation of original graph is become to equally distributed form, like this, just increased the dynamic range of grey scale pixel value, thereby reach the effect that strengthens integral image contrast.But the method is not obvious to the removal of noise really.

(3) frequency domain filtering.HFS in the corresponding Fourier conversion of the edge of image and noise, thus low-pass filter can smoothed image, remove out noise.The part of gradation of image generation fusion is corresponding with the high fdrequency component of frequency spectrum, so adopt high frequency filter decay or suppress low frequency component, can carry out sharpening to image.But, because the edge of image is also high fdrequency component, so adopt low-pass filter will weaken the marginal information of image.

In the research and practice process to the method, the present inventor finds: image enchancing method mainly carries out in time domain or frequency domain, but it is uncommon with these two class methods, to be combined into basic enhancing technology, and the method for using is all unilateral to the enhancing effect of image, has also lost more detailed information in filtering noise.For this problem, need a kind of details that retains as much as possible image in denoising.

Summary of the invention

The invention provides a kind of method of image enhancement processing, the method for traditional wavelet frequency domain filtering can be fused to intrinsic mode function territory, can in denoising, retain as much as possible the details of image.

A kind of method that the invention provides image enhancement processing, comprising:

Original image is obtained to the intrinsic mode function IMF of image by Bidimensional Empirical Mode Decomposition BEMD;

High-frequency I MF in IMF is carried out to wavelet transformation, IMF is carried out to 5 layers of wavelet decomposition, obtain respectively the wavelet coefficient of each IMF component;

To carrying out the high-frequency I MF of wavelet transformation, carry out the denoising of small echo self-adaptation, the 1st layer is used dual threshold to strengthen, and the 2nd, 3 layers are used self-adaptation to strengthen, and the 4th, 5 layers are used single threshold to strengthen;

To carrying out the high-frequency I MF of small echo self-adaptation denoising, carry out wavelet inverse transformation;

Low frequency IMF in IMF and the high-frequency I MF that carries out wavelet inverse transformation are synthesized to processing, the image after being enhanced.

The described intrinsic mode function IMF that original image is obtained to image by Bidimensional Empirical Mode Decomposition BEMD comprises:

A, r _i-1(x, y)=f (x, y) (i=1) wherein f (x, y) be original image;

C, initialization: h _j-1(x, y)=r _i-1(x, y) j=1;

The method of c, the mathematical morphology reconstruct of employing based on geodesy operator is determined h _j-1the maximum value of (x, y) and minimum point, adopt cubic spline interpolation algorithm to obtain extreme value coenvelope up _j-1(x, y) and lower envelope down _j-1(x, y), and average envelope $m_{j-1}(x,y)=\frac{{\mathrm{up}}_{j-1}(x,y)+{\mathrm{down}}_{j-1}(x,y)}{2};$

D, from received image signal, deduct average envelope surface:

H _j(x, y)=h _j-i(x, y)-m _j-1(x, y), makes j=j+1;

E, calculating: $\mathrm{SD}=\underset{x,y}{\mathrm{\Σ}}\left[\frac{{|h_j(x,y)-h_{j-1}(x,y)|}^2}{h_{j-1}^2(x,y)}\right]$

If SD < is sd (sd generally gets 0.2～0.3), make IMF _i(x, y)=h _j(x, y), otherwise repeating step d～e;

f、r _i(x，y)＝r _i-1(x，y)-IMF _i(x，y)；

G, i=i+1, repeating step b～f, until the IMF quantity (N) that obtains requiring finishes.Now can represent the image as: $f(x,y)=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{IMF}}_i(x,y)+R_N(x,y)$

Wherein, IMF _i(x, y) is two-dimentional IMF, R _n(x, y) is corresponding to the residual error part after decomposing.

Because the embodiment of the present invention adopts Bidimensional Empirical Mode Decomposition and adaptive wavelet method, it is based on data time domain local feature, adaptive time frequency analyzing tool for Bidimensional Empirical Mode Decomposition (Bidimensional Empirical Mode Decomposition, DEMD).Its local feature based on signal decomposes, can by " sieve " process, the data acquisition of any complexity be decomposed into a series of intrinsic mode function limited, local arrowband (Intrinsic Mode Function adaptively, IMF), it processes frequency information (distance between extreme point) in time domain.By carrying out the high-frequency I MF that Bidimensional Empirical Mode Decomposition obtains, decompose again, high-frequency I MF composition can be decomposed in the middle of wavelet field like this, can be fully noise information be focused in the middle of radio-frequency component, to retain as much as possible image detail, and adopt improved wavelet thresholding method to carry out denoising to its radio-frequency component, therefore this invention can make the noise of removing image simultaneously by figure image intensifying, and Useful Information is outstanding.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, to the accompanying drawing of required use in embodiment or description of the Prior Art be briefly described below, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skills, do not paying under the prerequisite of creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the process flow diagram of the image enhancement processing method in the embodiment of the present invention;

Fig. 2 is the process flow diagram of Bidimensional Empirical Mode Decomposition (BEMD);

Fig. 3 is the Lena image B EMD decomposing schematic representation based on BEMD;

Fig. 4 is the small echo mapping function figure that single threshold is processed;

Fig. 5 is the small echo mapping function figure that dual threshold is processed;

Fig. 6 is the small echo mapping function figure of adaptive algorithm;

Fig. 7 is that the self-adaptation of small echo strengthens process flow diagram.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is only the present invention's part embodiment, rather than whole embodiment.Embodiment based in the present invention, those of ordinary skills, not making all other embodiment that obtain under creative work prerequisite, belong to the scope of protection of the invention.

Below in conjunction with accompanying drawing, the invention will be further described.

The embodiment of the present invention provides a kind of method of the figure image intensifying based on Bidimensional Empirical Mode Decomposition and adaptive wavelet, the method of traditional wavelet frequency domain filtering can be fused to intrinsic mode function territory, can in denoising, retain as much as possible the details of image.Below be elaborated respectively.

Process flow diagram of the present invention as shown in Figure 1, comprises step: (1) obtains the intrinsic mode function (IMF) of image by Bidimensional Empirical Mode Decomposition BEMD; (2) high frequency intrinsic mode function is carried out to wavelet transformation; (3) realize adaptive denoising; (4) wavelet inverse transformation; (5) image after being enhanced.

Each step is specific as follows:

Step (1): the intrinsic mode function (IMF) that obtains image by BEMD

In existing technology, propose a complete definition of intrinsic mode function, and should meet following two conditions:

(1), in whole data sequence, the number of extreme point and the number of zero crossing must equate, or differ at most one.?

(N _z-1)≤N _e≤(N _z+1)

N wherein _erepresent extreme point number, comprise maximum point and minimum point; N _zrepresent zero crossing number.

(2), on a time point in office, the mean value of the lower envelope that the coenvelope being formed by the Local modulus maxima of signal and local minizing point form is zero.?

$\frac{[f_{\max }\left(t_i\right)+f_{\min }\left(t_i\right)]}{2}=0,$ t _i∈[t _a，t _b]

Wherein, f _max(t) coenvelope that representative is determined by maximum point; f _min(t) lower envelope that representative is determined by minimum point.

When two-dimensional image data is carried out to BEMD decomposition, we must first make the following assumptions: 2-D data plane at least comprises a maximum point and a minimum point, or whole two dimensional surface does not have extreme point but after carrying out single order or multistage derivative operation, can occur a maximum point and a minimum point; Distance definition between extreme point for characteristic dimension.

As shown in Figure 2, the basic thought of Bidimensional Empirical Mode Decomposition (BEMD) method is as follows:

A, r _i-1(x, y)=f (x, y) (i=1) wherein f (x, y) be original image;

B, initialization:

h _j-1(x，y)＝r _i-1(x，y)，j＝1；

The method of c, the mathematical morphology amount structure of employing based on geodesy operator is determined h _j-1the maximum value of (x, y) and minimum point, adopt cubic spline interpolation algorithm to obtain extreme value coenvelope up _j-1(x, y) and lower envelope down _j-1(x, y), and average envelope $m_{j-1}(x,y)=\frac{{\mathrm{up}}_{j-1}(x,y)+{\mathrm{down}}_{j-1}(x,y)}{2};$

D, from received image signal, deduct average envelope surface:

H _j(x, y)=h _j-i(x, y)-m _j-1(x, y), makes j=j+1;

E, calculating:

$\mathrm{SD}=\underset{x,y}{\mathrm{\&Sigma;}}\left[\frac{{|h_j(x,y)-h_{j-1}(x,y)|}^2}{h_{j-1}^2(x,y)}\right]$

If SD < is sd (sd generally gets 0.2～0.3), make IMF _i(x, y)=h _j(x, y), otherwise repeating step d～e;

f、r _i(x，y)＝r _i-1(x，y)-IMF _i(x，y)；

G, i=i+1, repeating step b～f, until the IMF quantity (N) that obtains requiring finishes.Now can represent the image as:

$f(x,y)=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{IMF}}_i(x,y)+R_N(x,y)$

Wherein, IMF _i(x, y) is two-dimentional IMF, R _n(x, y) is corresponding to the residual error part after decomposing.

Shown in Fig. 2 is 4 IMF components and the residual error remainder that image obtains after BEMD decomposes.

Step (2): high frequency intrinsic mode function is carried out to wavelet transformation

By 4 IMF components that obtain in step (1), are high fdrequency components, residual error remainder is low frequency component.IMF is carried out to 5 layers of wavelet decomposition, obtain respectively the wavelet coefficient of each IMF component.

Shown in Fig. 3 is the schematic diagram of IMF after wavelet transformation.

Step (3): adaptive denoising

Through step (2), afterwards wavelet coefficient is amplified to processing.Problem is, if original image contains noise, so many wavelet coefficients, particularly the wavelet coefficient on high-resolution yardstick is caused by noise, when carrying out inverse wavelet transform (normally final step), any amplification to these wavelet coefficients, all can be reflected in the enhancing of noise.

Single threshold strengthen to adopt is a kind of simple but mapping function of effectively processing wavelet coefficient, as shown in Figure 3.By known this mapping function of analysis chart 2, be expressed as:

This single G is gain, W _inthe wavelet coefficient values of input, W _outit is the wavelet coefficient values of output.Use this mapping function can amplify noise.Noise in the middle of image shows as less wavelet coefficient in wavelet transformed domain, and this single mapping function is to absolute value, the wavelet coefficient lower than T all amplifies by maximum gain, has therefore inevitably caused the amplification of noise.

It is that what to adopt is a kind of mapping function that uses the processing wavelet coefficient of dual threshold that dual threshold strengthens, as shown in Figure 4.In this function, all absolute values fall (being namely set to 0) lower than the wavelet coefficient of less threshold value T1 is all suppressed; Absolute value is greater than T1 but the wavelet coefficient that is less than higher thresholds T2 has identical gain G, and for absolute value, is greater than the wavelet coefficient of T2, and its gain reduces along with the increase of wavelet coefficient values.The mapping function of this form can be defined as follows:

$W_{\mathrm{out}}=\left\{\begin{array}{c}{W}_{\mathrm{in}}+T_2^{*}(G-1)-T_1^{*}G,T_2<W_{\mathrm{in}};\\ G^{*}(W_{\mathrm{in}}-T_1),T_1<W_{\mathrm{in}}\&le;T_2;\\ 0,-T_1\&le;W_{\mathrm{in}}\&le;T_1;\\ G^{*}(W_{\mathrm{in}}+T_1),-T_2\&le;W_{\mathrm{in}}<-T_1;\\ W_{\mathrm{in}}-T_2^{*}(G-1)+T_1^{*}G,W_{\mathrm{in}}<T_2.\end{array}\right.$

Analysis chart 3 is known, if suppress the wavelet coefficient of very little value, not only can eliminate noise, simultaneously also can fuzzy or damage detail signal.The minor detail of using the mapping function meeting failure pattern picture of this form, this just disagrees with the original intention that strengthens processing.

In order to improve arbitrarily wavelet coefficient values again when suppressing noise, we need to a kind ofly can distinguish in wavelet coefficient, which coefficient is by the caused method of noise, self-adapting enhancement method, and its mapping function is as shown in Figure 5.This method needn't only rely on the size of wavelet coefficient values and process, and will value following wavelet coefficient values in all Fujian all from day ground, not suppress.We can utilize this feature of correlativity of wavelet coefficients between two neighboring scales on the same space position.The correlativity of the wavelet coefficient that in fact, the same position place between adjacent yardstick is caused by noise is conventionally all very little.Therefore, we just can utilize the correlativity of wavelet coefficient under different scale carry out differentiate between images in the middle of which wavelet coefficient by noise, caused, which is caused by signal itself again.

For this problem, the present invention judges that by the confidence level of calculating noise wavelet coefficient is that noise causes or signal causes.If oneself knows that yardstick is that n and adjacent yardstick are the wavelet coefficient on n+1, we only need to multiply each other two wavelet coefficient values of each pixel seriatim so, just can on each location of pixels, obtain correlation:

Corr _n＝W _n ^*W _n+1

Calculate the general power of n layer correlation and the general power of wavelet coefficient:

P _Corr＝∑Corr _n ^*Corr _n

P _W＝∑W _n ^*W _n

And then can obtain the correlation of following form:

${\mathrm{Corr}2}_n={{\mathrm{Corr}}_n}^{*}\sqrt{P_W/P_{\mathrm{Corr}}}$

Can be gone up now and minimum absolute value in correlation:

Corr2 _min＝min{|Corr2 _n|}

This value can be used for mark noise: all Corr2 _n, absolute value equals Corr2 _minposition we think that wavelet coefficient is caused by noise, its noise reliability E is set to 1; To Corr2 _nabsolute value compares W _nwe think that wavelet coefficient is caused by signal the large position of absolute value, and it is 0 that E is set in these positions; And for Corr2 _nabsolute value compares W _nabsolute value is little but compare Corr2 _minlarge position, we arrange E is the value between 0 to 1, this value is that W produces, absolute value and Corr2 _ndifference and the W of absolute value _nabsolute value and Corr2 _minthe ratio of difference, that is:

$E=\frac{\mathrm{abs}\left(W_n\right)-\mathrm{abs}\left({\mathrm{Corr}2}_n\right)}{\mathrm{abs}\left(W_n\right)-{\mathrm{Corr}2}_{\min }}$

When E shows wavelet coefficient, by noise, caused, should adopt so in Fig. 5 that defined mapping function thinner in two solid lines.On the other hand, if E shows wavelet coefficient, be not caused by noise, so at this moment should select the corresponding mapping function of that solid line thicker in Fig. 5.At E, can not judge that wavelet coefficient is cause by noise or by signal in the situation that on earth, gain gets 0 to G _maxa middle value.Particularly: if E is greater than the upper limit 0.9 (experiment value), gain is made as 0; If E is less than lower limit 0.75 (experiment value), gain is made as maximal value G _maxif E is between 0.9 and 0.75, gain linearity changes.

As shown in Fig. 6 process flow diagram, in the present invention, will carry out 5 layers of wavelet decomposition to each IMF component, the 1st layer is used dual threshold to strengthen, and the 2nd, 3 layers are used self-adaptation to strengthen, and the 4th, 5 layers are used single threshold to strengthen.After processing respectively, obtain new wavelet coefficient.

Step (4): carry out wavelet inverse transformation

IMF after step (3) is processed is used respectively wavelet inverse transformation to obtain new IMF component IMF.

Step (5): the image after being enhanced

Through step (4) IMF ' and BEMD R afterwards afterwards _nthe image of (x, y) summation after being enhanced be:

Because the embodiment of the present invention adopts Bidimensional Empirical Mode Decomposition and adaptive wavelet method, it is based on data time domain local feature, adaptive time frequency analyzing tool for Bidimensional Empirical Mode Decomposition (Bidimensional Empirical Mode Decomposition, DEMD).Its local feature based on signal decomposes, can by " sieve " process, the data acquisition of any complexity be decomposed into a series of intrinsic mode function limited, local arrowband (Intrinsic Mode Function adaptively, IMF), it processes frequency information (distance between extreme point) in time domain.By carrying out the high-frequency I MF that Bidimensional Empirical Mode Decomposition obtains, decompose again, high-frequency I MF composition can be decomposed in the middle of wavelet field like this, can be fully noise information be focused in the middle of radio-frequency component, to retain as much as possible image detail, and adopt improved wavelet thresholding method to carry out denoising to its radio-frequency component, therefore this invention can make the noise of removing image simultaneously by figure image intensifying, and Useful Information is outstanding.

It should be noted that, the contents such as the information interaction between said apparatus and intrasystem each unit, implementation, due to the inventive method embodiment based on same design, particular content can, referring to the narration in the inventive method embodiment, repeat no more herein.

One of ordinary skill in the art will appreciate that all or part of step in the whole bag of tricks of above-described embodiment is to come the hardware that instruction is relevant to complete by program, this program can be stored in a computer-readable recording medium, storage medium can comprise: ROM (read-only memory) (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), disk or CD etc.

The embodiment above embodiment of the present invention being provided is described in detail, applied specific case herein principle of the present invention and embodiment are set forth, the explanation of above embodiment is just for helping to understand method of the present invention and core concept thereof; , for one of ordinary skill in the art, according to thought of the present invention, all will change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention meanwhile.

Claims (2)

1. a method for image enhancement processing, is characterized in that, comprising:

Original image is obtained to the intrinsic mode function IMF of image by Bidimensional Empirical Mode Decomposition BEMD;

High-frequency I MF in IMF is carried out to wavelet transformation, IMF is carried out to 5 layers of wavelet decomposition, obtain respectively the wavelet coefficient of each IMF component;

To carrying out the high-frequency I MF of wavelet transformation, carry out the denoising of small echo self-adaptation, the 1st layer is used dual threshold to strengthen, and the 2nd, 3 layers are used self-adaptation to strengthen, and the 4th, 5 layers are used single threshold to strengthen;

To carrying out the high-frequency I MF of small echo self-adaptation denoising, carry out wavelet inverse transformation;

Low frequency IMF in IMF and the high-frequency I MF that carries out wavelet inverse transformation are synthesized to processing, the image after being enhanced.

2. the method for claim 1, is characterized in that, the described intrinsic mode function IMF that original image is obtained to image by Bidimensional Empirical Mode Decomposition BEMD comprises:

A, r _i-1(x, y)=f (x, y) (i=1) wherein f (x, y) be original image;

B, initialization: h _j-1(x, y)=r _i-1(x, y), j=1;

The method of c, the mathematical morphology reconstruct of employing based on geodesy operator is determined h _j-1the maximum value of (x, y) and minimum point, adopt cubic spline interpolation algorithm to obtain extreme value coenvelope up _j-1(x, y) and lower envelope down _j-1(x, y), and average envelope $m_{j-1}(x,y)=\frac{{\mathrm{up}}_{j-1}(x,y)+{\mathrm{down}}_{j-1}(x,y)}{2};$

D, from received image signal, deduct average envelope surface:

H _j(x, y)=h _j-i(x, y)-m _j-1(x, y), makes j=j+1;

E, calculating: $\mathrm{SD}=\underset{x,y}{\mathrm{\&Sigma;}}\left[\frac{{\left|(h_j(x,y)-h_{j-1}(x,y))\right|}^2}{h_{j-1}^2(x,y)}\right]$

If SD < is sd, make IMF _i(x, y)=h _j(x, y), wherein sd gets 0.2～0.3, otherwise repeating step d～e;

f、r _i(x，y)＝r _i-1(x，y)-IMF _i(x，y)；

G, i=i+1, repeating step b～f, until the IMF quantity N that obtains requiring finishes, now represents the image as: $f(x,y)=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{IMF}}_i(x,y)+R_N(x,y)$

Wherein, IMF _i(x, y) is two-dimentional IMF, R _n(x, y) is corresponding to the residual error part after decomposing.

Images