CN101488218B - Image self-adapting airspace homographic filtering method - Google Patents

Abstract

The invention discloses a self-adapting empty domain homomorphic filtering method for an image, which carries out low-pass filtering on a luminance and grey scale image to obtain low-pass information so as to obtain high-pass information by taking the logarithm of the a luminance value of a pixel point and then subtracting a low-pass signal, and then obtain the luminance value after empty domain homomorphic filtering through exponent arithmetic. Two parameters alpha and beta in extraction of self-adapting high-pass information are changed along with self-adapting of the low-pass information, therefore, the method utilizes the low-pass information of the image to self-adaptively extract the high-pass information on an empty domain to avoid error of the current empty domain homomorphic filtering caused by linear processing.

Description

A kind of image self-adapting airspace homographic filtering method

Technical field

The present invention relates to technical field of image processing, specifically, relate to a kind of image self-adapting airspace homographic filtering method.

Background technology

The figure image intensifying is the important means that improves visual effect in the image pre-service, comprises removing noise, expanded contrast etc.Homomorphic filtering is a kind of algorithm that arrives commonly used in image enhancement processing, and it is commonly used to handle the image of illumination unevenness.

For piece image, its gray-scale value can be thought to be made up of incident light component and reflected light component two parts product, and wherein incident light depends on and light source, it is more even, less with spatial position change, so incident light occupies the low frequency part of frequency field, the correspondence image background.And reflected light depends on and the character of object itself, and promptly the brightness of scenery depends primarily on the object reflected light.Because object character is different with design feature, catoptrical power is also inequality, and is more violent with spatial position change, so reflected light occupies the HFS of frequency field, and the correspondence image details.

The uneven image that throws light on is carried out enhancement process suppress low frequency component exactly, reduce the influence of penetrating light component as far as possible, the enlarged image high fdrequency component strengthens the catoptrical component of object simultaneously.Strengthen the contrast of image when can suppress its dynamic range, reach the purpose of the uneven Flame Image Process of light and shade by such processing.

In the prior art, adopt the mode of homomorphic filtering that video image is carried out enhancement process usually:

One, classical frequency domain homomorphic filtering

This method is carried out on frequency domain, utilize earlier FFT with image transformation on frequency field, and then low frequency part and HFS are applied different influences with the Gaussian high-pass filtering function of revising, last IFFT recovers original image.

Classical frequency domain homomorphic filtering has following defective: FFT handles entire image, so homomorphic filtering is based upon on the basis of the information that obtains entire image and carries out, and dirigibility is little and complexity is big.

Two, airspace homographic filtering

This method is carried out on the spatial domain, at first to a gradation of image coefficient on duty to keep low-pass information, use the low pass template to carry out low-pass filtering, original image deducts low-pass information and obtains high pass information.

Airspace homographic filtering has following defective: though avoid using FFT that image is handled, but space low pass template can not rationally be chosen the template size of low-pass filtering and not only can effectively reflect low-pass information but also can reduce computation complexity, and to image high pass information is linear, because it is inaccurate that low-pass information is extracted, and causes image fault at last.

Summary of the invention

The objective of the invention is to overcome the deficiency of existing airspace homographic filtering, provide a kind of image fault little image self-adapting airspace homographic filtering method.

To achieve the above object of the invention, image self-adapting airspace homographic filtering method of the present invention is characterized in that, may further comprise the steps:

(1), the brightness value of the pixel of calculating input image, obtain the brightness image of present image, wherein the brightness value of pixel be f (x, y), f (x, y) denotation coordination is x, the brightness value of the pixel of y;

(2), the brightness image is carried out low-pass filtering, obtain low-pass information LPF (x, y);

(3), according to low-pass information LPF (x, y) calculate self-adaptation high pass information g (x, the y) parameter alpha in extracting, β:

α＝γ×log _φLPF(x，y)

β＝log _φ(LPF(x，y)+φ)-ε

Wherein, γ, φ, ε are constant

(4), after obtaining two parameter alpha, β, carry out extracted in self-adaptive high pass information g (x, y):

g(x，y)＝α×logf(x，y)-β×LPF(x，y)

(5), with the high pass information g that obtains (x y) gets exponent arithmetic, obtain behind the airspace homographic filtering brightness value s (x, y):

s(x，y)＝expg(x，y)

The object of the present invention is achieved like this: airspace homographic filtering method of the present invention is based on that following irradiation-reflection model makes up,

f(x，y)＝i(x，y)×r(x，y)

In the model, and f (x, y) denotation coordination is x, the brightness value of the pixel of y, (x y) is the illumination component to i, and (x y) is reflected light component to r.Intensity of illumination generally has consistance, the character that spatially has slow variation usually show as the low frequency component after Fourier changes, yet different materials or reflected by objects rate is widely different, thereby make the brightness value of image change the HFS of correspondence image.

By the computing of taking the logarithm the irradiates light of formula (1) is separated with reflected light, that is:

logf(x，y)＝logi(x，y)+logr(x，y)

Like this known illumination component i (x, y) and brightness value f (x, under situation y), can calculate reflected light component r (x, y).In the present invention, carry out low-pass filtering, obtain low-pass information LPF (x by the brightness image, y) this low-pass information LPF (x, is described as background technology, y) with illumination component i (x, y) corresponding, and high pass information g (x, y) with r (x, y) be reflected light component, deduct the high pass information that low-pass signal just obtains image after taking the logarithm by the brightness value of pixel like this, then by get brightness value s after exponent arithmetic obtains airspace homographic filtering (x, y).

In the present invention, self-adaptation high pass information g (β is for x, the y) parameter alpha in extracting:

α＝γ×log _φLPF(x，y)

β＝log _φ(LPF(x，y)+φ)-ε

Wherein, γ, φ, ε are constant

After obtaining two two parameter alpha that change with low-pass information, β, just can according to low-pass information extracted in self-adaptive high pass information g (x, y):

g(x，y)＝α×logf(x，y)-β×LPF(x，y)

Then with the high pass information g that obtains (x y) gets exponent arithmetic, obtain behind the airspace homographic filtering brightness value s (x, y):

s(x，y)＝expg(x，y)

Because at high pass information g (x, y) in the leaching process, therefore two parameter alpha, β, utilize image low-pass information extracted in self-adaptive high pass information to avoid present airspace homographic filtering because the error that linear process causes on the spatial domain with the low-pass information adaptive change.

Description of drawings

Fig. 1 is a kind of embodiment process flow diagram of image self-adapting airspace homographic filtering method of the present invention.

Embodiment

For understanding the present invention better, the present invention is more described in detail below in conjunction with the drawings and specific embodiments.In the following description, when perhaps the detailed description of existing prior art can desalinate subject content of the present invention, these were described in here and will be left in the basket.

Fig. 1 is a kind of embodiment process flow diagram of image self-adapting airspace homographic filtering method of the present invention.In the present embodiment, image self-adapting airspace homographic filtering method of the present invention may further comprise the steps:

Step ST1: input picture is converted to brightness image F, and the brightness value of image slices vegetarian refreshments is:

Y＝0.27×R+0.67×G+0.06×B (1)

Step ST2: calculate the brightness related coefficient.The brightness related coefficient of input picture R, G, B passage is designated as coff respectively _R, coff _G, coff _B, by the related adjustment that reaches color of input picture R, G, B channel signal with brightness.Be calculated as follows:

${\mathrm{coff}}_{{\mathrm{cl}}_i}=\frac{{\mathrm{cl}}_i}{Y_{\max }}---\left(2\right)$

Cl _iThe corresponding R of remarked pixel point i, G, B passage, Y _MaxExpression input image lightness maximal value.

Step ST3: use 9 * 9 templates to image carry out low-pass filtering obtain low-pass signal LPF (x, y).

Step ST4: according to low-pass information LPF (x, y) calculate self-adaptation high pass information g (x, the y) parameter alpha in extracting, β:

α＝γ×log _φLPF(x，y)

β＝log _φ(LPF(x，y)+φ)-ε (3)

Wherein, γ, φ, ε are constant

Step ST5: after obtaining two two parameter alpha that change with low-pass information, β, just can according to low-pass information extracted in self-adaptive high pass information g (x, y):

g(x，y)＝α×logf(x，y)-β×LPF(x，y) (4)

Step ST6: then with the high pass information g that obtains (x y) gets exponent arithmetic, obtain behind the airspace homographic filtering brightness value s (x, y):

s(x，y)＝expg(x，y) (5)

Step ST7: recover the RGB image, obtain image I mg ' behind the airspace homographic filtering:

${\mathrm{cl}}_i={\mathrm{coff}}_{{\mathrm{cl}}_i}\×s(x,y)---\left(6\right)$

In the present embodiment, we as can be seen, the present invention utilizes image low-pass information extracted in self-adaptive high pass information to avoid present airspace homographic filtering because the error that causes of linear process on the spatial domain.In addition, use new color space conversion pattern, make homomorphic filtering only handle, thereby reduced working time a luminance channel.

Although above the illustrative embodiment of the present invention is described; but should be understood that; the invention is not restricted to the scope of embodiment; to those skilled in the art; as long as various variations appended claim limit and the spirit and scope of the present invention determined in; these variations are conspicuous, and all utilize innovation and creation that the present invention conceives all at the row of protection.

Claims (2)

1. an image self-adapting airspace homographic filtering method is characterized in that, may further comprise the steps:

(1), the brightness value of the pixel of calculating input image, obtain the brightness image of present image, wherein, f (x, y) denotation coordination is x, the brightness value of the pixel of y, f (x, y)=i (x, y) * and r (x, y), i (x, y) be the illumination component, (x y) is reflected light component to r, take the logarithm the illumination component is separated with reflected light component, that is, and logf (x, y)=and logi (x, y)+logr (x, y);

(2), the brightness image is carried out low-pass filtering, obtain low-pass information LPF (x, y), low-pass information LPF (x, y) with illumination component i (x, y) corresponding;

(3), according to low-pass information LPF (x, y) calculate self-adaptation high pass information g (x, y) parameter alpha, the β in extracting:

α＝γ×log _φLPF(x，y)，

β＝log _φ(LPF(x，y)+φ)-ε，

Wherein, γ, φ, ε are constant;

(4), after obtaining two parameter alpha, β, carry out extracted in self-adaptive high pass information g (x, y), high pass information g (x, y) with reflected light component r (x, y) corresponding:

g(x，y)＝α×logf(x，y)-β×LPF(x，y)；

(5), with the high pass information g that obtains (x y) gets exponent arithmetic, obtain behind the airspace homographic filtering brightness value s (x, y):

s(x，y)＝expg(x，y)。

2. image self-adapting airspace homographic filtering method according to claim 1 is characterized in that, in step (1) afterwards, also input picture is carried out the brightness related coefficient and calculates, and obtains the brightness related coefficient of input picture R, G, B passage

${\mathrm{coff}}_{x_i}=\frac{x_i}{Y_{\max }},$

x _iThe corresponding R of expression input image pixels point i, G, B passage, Y _MaxExpression input image lightness maximal value;

Brightness value s after step (5) obtains airspace homographic filtering (x, y) after, recover the RGB image, obtain image behind the airspace homographic filtering:

${\mathrm{cl}}_i={\mathrm{coff}}_{x_i}\×s(x,y),$

Cl _iThe corresponding R of remarked pixel point i, G, B passage.