CN103440630A - Large-dynamic-range infrared image display and detail enhancement method based on guiding filter - Google Patents

Abstract

The invention discloses a large-dynamic-range infrared image display and detail enhancement method based on a guiding filter. According to the method, firstly, the guiding filter is used for carrying out edge reserved smoothing filtering processing on an original large-dynamic-range infrared image to obtain base layer image information with blurry details, the difference between a base layer image and the original image is obtained to obtain image detail layer information, self-adaption enhancement processing is carried out on a detail layer image, dynamic range compressing is carried out on the base layer image, finally, the processed detail layer image and the processed base layer image are combined, an overflow value is removed, and a low-dynamic-range image finally used for being displayed is obtained. According to the method, image data can be compressed, the image contrast can be effectively improved at the same time, image details are enhanced, and the visual effect of the images is improved. Meanwhile, the method is less in calculated amount and facilitates practical application.

Description

Great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method

Technical field

The inventive method belongs to the infrared image processing technical field, and particularly a kind of great dynamic range infrared image of processing based on the guiding filter filtering shows and detail enhancing method.

Background technology

The modern high performance thermal infrared imager can obtain the raw image data that dynamic range is very large, and bit wide is generally in the 12-14 position, but the dynamic range bit wide of display device is generally 8.Generally, after the original image of acquisition great dynamic range, must be remapped to original image, by its dynamic range compression so that the demonstration of display device.This process need to reach two conditions usually: the first, compression raw image data can be complementary with the dynamic range of display device the dynamic range of high-performance thermal infrared imager output image; The second, simultaneously, retain as far as possible the details existed in original image in compression process, make the observer finally on display device, can observe the image with better visual effect, and can be easier to distinguish and be hidden in the weak target in background.

Dynamic range compression algorithm commonly used, have automatic gain to control and histogram equalization, is most popular image display technology in infrared imaging system.Automatic gain is controlled and is at first rejected the extremum in raw image data, and then by whole dynamic range linear mapping to 8, but this method there will be, contrast is low, the obvious problem of loss in detail.The histogram equalization output image there will be the problems such as enhancing, the amplification of homogeneous area noise, bleaching effect.

Detail enhancement algorithms commonly used, as extensively visible (referring to document one: K.Zuiderveld in many documents as contrast enhancement algorithms, " Contrast limited adaptive histogram equalizaiton; " in Graphics Gems IV, pp.474-485, Academic Press Professional, Inc., San Diego1994.).But contrast enhancement algorithms is inapplicable to the great dynamic range original image mostly for the low-dynamic range infrared image; And only considered the enhancing problem of rest image, do not considered the application in real-time system.

Consider that automatic gain is controlled and the deficiency of histogram equalizing method, many more complicated methods are suggested, as Retinex, but these methods are mainly for visible images, there is good effect for visible images, and to the infrared image poor effect, the problem that exists noise to amplify; And for example based on layering, process the algorithm of framework (referring to document two: Chao Zuo, Qian Chen, Ning Liu et al.Display and detail enhancement for high-dynamic-range infrared images[J] .Optical Engineering, 2011,50 (12): 127401 (9) .), its main thought is that image is divided into to levels of detail and basic layer, then processed respectively, but the method operand is large, cause real-time poor, and easily produce the gradient flop phenomenon.

Summary of the invention

The present invention proposes a kind of great dynamic range infrared image based on the guiding wave filter and shows and detail enhancing method, picture contrast and image detail after making to process all been significantly enhanced, and operand is little, real-time, good without gradient flop phenomenon, noise suppression effect.

In order to solve the problems of the technologies described above, the present invention proposes a kind of great dynamic range infrared image based on the guiding wave filter and shows and detail enhancing method, it is characterized in that, comprises the following steps:

Step 1: use the guiding wave filter to carry out the disposal of gentle filter to original infrared picture data, obtain the base layer data of original infrared image, and base layer data and original infrared picture data are done poor, acquisition levels of detail data;

Step 2: use the guiding filtering core to carry out self-adaptation to the levels of detail data and strengthen calculating, obtain the levels of detail data after self-adaptation strengthens;

Step 3: use the histogram projection technology base layer data is carried out to dynamic range compression and comparison is strengthened and processes, the base layer data after the acquisition dynamic range compression;

Step 4: levels of detail data and the base layer data after the described dynamic range compression of step 3 after the described self-adaptation of step 2 is strengthened merge, and reject overflow value, obtain final output image data.

Compared with prior art, its remarkable advantage is in the present invention:

Use guiding filter smoothing image, by adjusting its filtering core correlation parameter, thereby can effectively distinguish image detail zone and background area, when levels of detail is carried out to the self-adaptation enhancing, effectively suppressed ground unrest, and calculated amount is little, real-time, not there will be the gradient flop phenomenon.When basic layer is processed, the inventive method increases a bias factor by the view data to histogram output, to successfully manage various scenes, the output image contrast is improved, and the noise that not there will be excessive stretching to cause amplifies.

The accompanying drawing explanation

Fig. 1 is the inventive method process flow diagram.

Fig. 2 is the levels of detail view data row figure extracted while guiding wave filter to get different windows size and ε value, row figure laterally from left to right every leu time mean that window size is at 3 * 3,5 * 5 and 7 * 7 o'clock, row figure vertically from top to bottom every row mean that successively the ε value is 100,500,1000 and at 2000 o'clock, obtained levels of detail view data row figure.

Fig. 3 is used respectively the output image effect contrast figure that self-adaptation strengthens and non-self-adapting strengthens acquisition, wherein, Fig. 3 (a) is original image, Fig. 3 (b) is used non-self-adapting to strengthen the output image obtained, Fig. 3 (c) is that the self-adaptation obtained in using this method processing original image process strengthens the gain coefficient mapping graph, and Fig. 3 (d) is used the middle self-adaptation of Fig. 3 (c) to strengthen gain coefficient to carry out the output image obtained after the details enhancing.

Fig. 4 is used respectively the inventive method and the contained effect contrast figure who processes the frame algorithm acquisition based on layering of list of references two, wherein, Fig. 4 (a) is original infrared image, Fig. 4 (b) is the processing image that uses the inventive method to obtain, Fig. 4 (c) is the processing image that uses two support methods of list of references to obtain, Fig. 4 (d) be in Fig. 4 (b) partial result 1. with Fig. 4 (c) in partial result amplification effect comparison diagram 2..

Embodiment

In conjunction with Fig. 1, the inventive method comprises the following steps:

Step 1: use the guiding wave filter to carry out the disposal of gentle filter to original infrared picture data, obtain the base layer data of original infrared image, and base layer data and original infrared picture data are done poor, acquisition levels of detail data.

As a kind of preferred version of step 1, use the computing method of guiding wave filter calculating base layer data in described step 1 as the formula (1),

$I_B(i,j)=\frac{1}{\left|w\right|}\underset{(i^{\′},j^{\′})\∈w_{i,j}}{\mathrm{\Σ}}(a(i^{\′},j^{\′})I_{\mathrm{in}}(i^{\′},j^{\′})+b(i^{\′},j^{\′}))---\left(1\right)$

In formula (1), I _infor the original infrared picture data of input, I _bfor base layer data, (i, j) is the pixel coordinate, w _{i, j}be the window of pixel centered by pixel (i, j), w is window w _{i, j}the number of middle pixel, a _{(i', j')}and b _{(i', j')}the linear coefficient in the window of pixel centered by (i', j'), a _{(i', j')}computing method as the formula (2), b _{(i', j')}computing method as the formula (3),

$a(i,j)=\frac{\frac{1}{\left|w\right|}\underset{(i^{\′},j^{\′})\∈w_{i,j}}{\mathrm{\Σ}}{I}_{\mathrm{in}}(i^{\′},j^{\′})G(i^{\′},j^{\′})-{\mathrm{\μ}}_{i,j}\stackrel{\‾}{I_{\mathrm{in}}(i^{\′},j^{\′})}}{{\mathrm{\σ}}_{i,j}+\mathrm{\ϵ}}.---\left(2\right)$

$b(i,j)=\stackrel{\‾}{I_{\mathrm{in}}(i^{\′},j^{\′})}-a(i,j){\mathrm{\μ}}_{i,j}.---\left(3\right)$

In formula (2) and formula (3), μ _{i', j'}the average that means each pixel in window; σ _{i', j'}the variance that means each pixel in window; ε is linear regression coeffficient, for determining the level and smooth degree of this wave filter;

mean given window w _{i, j}middle I _inaverage; G (i', j') for the guiding wave filter for the treatment of the navigational figure data, G (i', j')=I _in(i, j), raw image data itself is as the navigational figure data;

Guide the size of window of wave filter and the value of ε by adjusting, can obtain detail extraction ability in various degree, the window size selection range of the inventive method guiding wave filter is that 3 * 3 to 9 * 9, ε values are 100 to 2500.Accompanying drawing 2 has provided under the value parameter combinations of several groups of different window sizes and ε, the levels of detail view data effect of extraction.

As the another kind of preferred version of step 1, described base layer data and original infrared picture data are done the poor computing method that obtain the levels of detail data as the formula (4),

I _D(i，j)＝I _in(i，j)-I _B(i，j) （4）

In formula (4), I _dpresentation video levels of detail data.

Step 2: use the guiding filtering core to carry out self-adaptation to the levels of detail data and strengthen calculating, obtain the levels of detail data after self-adaptation strengthens.

As a kind of preferred version of step 2, described self-adaptation strengthens computing method as the formula (5),

I _DP＝I _D*(W(i，j)*a+b) （5）

In formula (5), I _dPfor self-adaptation strengthens the levels of detail data that obtain after rear calculating, I _dfor the image detail layer data, a and b are the linear coefficients of adjusting, and the span that the span of a is 1.5 to 3, b is 0 to 0.5; W (i, j) is guiding filter kernel coefficient, the computing method of W (i, j) as the formula (6),

$W(i,j)=\frac{1}{\left|w\right|}\underset{(i^{\′},j^{\′})\∈w_{i,j}}{\mathrm{\Σ}}\left(\frac{{(I(i,j)-{\mathrm{\μ}}_{i^{\′},j^{\′}})}^2}{{{\mathrm{\σ}}_{i^{\′},j^{\′}}}^2+\mathrm{\δ}}\right)---\left(6\right)$

In formula (6), μ _{i', j'}and σ _{i', j'}respectively average and the variance of each pixel in the window of pixel centered by (i', j'); δ is the details decision threshold; W (i, j), for the mask as differentiating details, carries out the self-adaptation enhancing to levels of detail.

Step 3: use the histogram projection technology base layer data is carried out to dynamic range compression and comparison is strengthened and processes, the base layer data after the acquisition dynamic range compression.

As a kind of preferred version of step 3, obtain the computing method of the base layer data after dynamic range compression as the formula (7),

$I_{\mathrm{BP}}=\frac{(D-R)}{P}+D\left(I_B\right)*R---\left(7\right)$

In formula (7), I _bPfor the base layer data after dynamic range compression; P is for for controlling the regulatory factor of output image overall brightness, and the brightness regulation factor can guarantee when output gray level is less, and for guaranteeing that output image can not be compressed to very low brightness, the span of P can be further defined to 2 to 6; Effective indication range that D is display, the D span is 0 to 255; D(I _b) be I _bthe accumulation histogram data; The gray level maximal value that R is output image;

D(I _b) account form as the formula (8), $D\left(I_B\right)=D\left(x\right)=\left\{\begin{array}{c}0,x=0\\ \frac{\underset{y=0}{\overset{x-1}{\mathrm{\Σ}}}H\left(y\right)}{n_{\mathrm{valid}}},\mathrm{other}\end{array}\right.---\left(8\right)$

In formula (8), n _validmean the valid gray level after projection process, H (y) is the statistic histogram data, H (y) value as the formula (9),

$H\left(y\right)=\left\{\begin{array}{c}0,n_y<T\\ 1,n_y>T\end{array}\right.---\left(9\right)$

In formula (9), n _ymean the number of pixels that gray level y has; T is the projection threshold value, and the span of T can be defined as 10～35 through a step;

The value mode of R as the formula (10),

R＝min(n _valid，D) （10）。

Step 4: levels of detail data and the base layer data after the described dynamic range compression of step 3 after the described self-adaptation of step 2 is strengthened merge, and reject overflow value, obtain final output image data.

The computing method of the described final output image data of step 4 as the formula (11), (whether being necessary to be formulated again)

In formula (11), operational symbol

mean to reject the overflow value operation.

The beneficial effect of the inventive method can further illustrate by following experimental result:

Can find out from accompanying drawing 3, use self-adaptation to strengthen the output image obtained, to the ground unrest inhibition, use non-self-adapting to strengthen the output image obtained effective.

From accompanying drawing 4, can find out, left side is tiltedly gone up directional arrow and is shown that Fig. 4 (b) Background suppression noise ratio is effective; Right side tiltedly descends directional arrow to show that Fig. 4 (b) can retain original image information near strong edge, can not produce gradient upset effect.Therefore be compared to two support methods of list of references, the inventive method can effectively strengthen the original image overall contrast, strengthen image detail, the while is the Background suppression noise effectively, there is no the gradient flop phenomenon.

Table 1 is to use two support methods of list of references and the inventive method to process the time contrast form of image, and test environment is Matlab2010b, and the test pattern data are identical.As can be seen from Table 1, when obtaining preferably treatment effect, the inventive method calculated amount is less, and the processing time is short, and treatment effeciency is high, and for the occasions such as hardware realization of relatively valuing algorithm complex, this algorithm is with the obvious advantage.

This algorithm of table 1 and comparison algorithm contrast service time (unit: second)

。

Claims (9)

1. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method, it is characterized in that, comprises the following steps:

Step 1: use the guiding wave filter to carry out the disposal of gentle filter to original infrared picture data, obtain the base layer data of original infrared image, and base layer data and original infrared picture data are done poor, acquisition levels of detail data;

Step 2: use the guiding filtering core to carry out self-adaptation to the levels of detail data and strengthen calculating, obtain the levels of detail data after self-adaptation strengthens;

Step 3: use the histogram projection technology base layer data is carried out to dynamic range compression and comparison is strengthened and processes, the base layer data after the acquisition dynamic range compression;

Step 4: levels of detail data and the base layer data after the described dynamic range compression of step 3 after the described self-adaptation of step 2 is strengthened merge, and reject overflow value, obtain final output image data.

2. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 1, it is characterized in that,

Use the computing method of guiding wave filter calculating base layer data in described step 1 as the formula (1),

$I_B(i,j)=\frac{1}{\left|w\right|}\underset{(i^{\&prime;},j^{\&prime;})\&Element;w_{i,j}}{\mathrm{\&Sigma;}}(a(i^{\&prime;},j^{\&prime;})I_{\mathrm{in}}(i^{\&prime;},j^{\&prime;})+b(i^{\&prime;},j^{\&prime;}))---\left(1\right)$

In formula (1), I _infor the original infrared picture data of input, I _bfor base layer data, (i, j) is the pixel coordinate, w _{i, j}be the window of pixel centered by pixel (i, j), w is window w _{i, j}the number of middle pixel, a _{(i', j')}and b _{(i', j')}the linear coefficient in the window of pixel centered by (i', j'), a _{(i', j')}computing method as the formula (2), b _{(i', j')}computing method as the formula (3),

$a(i,j)=\frac{\frac{1}{\left|w\right|}\underset{(i^{\&prime;},j^{\&prime;})\&Element;w_{i,j}}{\mathrm{\&Sigma;}}{I}_{\mathrm{in}}(i^{\&prime;},j^{\&prime;})G(i^{\&prime;},j^{\&prime;})-{\mathrm{\&mu;}}_{i,j}\stackrel{\&OverBar;}{I_{\mathrm{in}}(i^{\&prime;},j^{\&prime;})}}{{\mathrm{\&sigma;}}_{i,j}+\mathrm{\&epsiv;}}.---\left(2\right)$

$b(i,j)=\stackrel{\&OverBar;}{I_{\mathrm{in}}(i^{\&prime;},j^{\&prime;})}-a(i,j){\mathrm{\&mu;}}_{i,j}.---\left(3\right)$

In formula (2) and formula (3), μ _{i', j'}the average that means each pixel in window; σ _{i', j'}the variance that means each pixel in window; ε is linear regression coeffficient, for determining the level and smooth degree of this wave filter; mean given window w _{i, j}middle I _inaverage; G (i', j') for the guiding wave filter for the treatment of the navigational figure data, G (i', j')=I _in(i, j), raw image data itself is as the navigational figure data;

In described step 1, base layer data and original infrared picture data are done the poor computing method that obtain the levels of detail data as the formula (4),

I _D(i，j)＝I _in(i，j)-I _B(i，j) （4）

In formula (4), I _dpresentation video levels of detail data.

3. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 2, and it is characterized in that, the window size selection range of described guiding wave filter is that 3 * 3 to 9 * 9, ε spans are 100 to 2500.

4. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 1, and it is characterized in that, in described step 2, self-adaptation strengthens computing method as the formula (5),

I _DP＝I _D*(W(i，j)*a+b) （5）

In formula (5), I _dPthe levels of detail data that obtain after calculating after strengthening for self-adaptation, a and b are the linear coefficients of adjusting, W (i, j) is guiding filter kernel coefficient, the computing method of W (i, j) as the formula (6),

$W(i,j)=\frac{1}{\left|w\right|}\underset{(i^{\&prime;},j^{\&prime;})\&Element;w_{i,j}}{\mathrm{\&Sigma;}}\left(\frac{{(I(i,j)-{\mathrm{\&mu;}}_{i^{\&prime;},j^{\&prime;}})}^2}{{{\mathrm{\&sigma;}}_{i^{\&prime;},j^{\&prime;}}}^2+\mathrm{\&delta;}}\right)---\left(6\right)$

In formula (6), μ _{i', j'}and σ _{i', j'}be respectively average and the variance of each pixel in the window of pixel centered by (i', j'), δ is the details decision threshold.

5. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 4, and it is characterized in that, the span that the span of a is 1.5 to 3, b is 0 to 0.5.

6. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 1, it is characterized in that, obtains the computing method of the base layer data after dynamic range compression in described step 3 as the formula (7),

$I_{\mathrm{BP}}=\frac{(D-R)}{P}+D\left(I_B\right)*R---\left(7\right)$

In formula (7), I _bPfor the base layer data after dynamic range compression; P is for controlling the regulatory factor of output image overall brightness; Effective indication range that D is display, the D span is 0～255; D(I _b) be I _bthe accumulation histogram data; The gray level maximal value that R is output image;

D(I _b) account form as the formula (8), $D\left(I_B\right)=D\left(x\right)=\left\{\begin{array}{c}0,x=0\\ \frac{\underset{y=0}{\overset{x-1}{\mathrm{\&Sigma;}}}H\left(y\right)}{n_{\mathrm{valid}}},\mathrm{other}\end{array}\right.---\left(8\right)$

In formula (8), n _validmean the valid gray level after projection process, H (y) is the statistic histogram data, H (y) value as the formula (9),

$H\left(y\right)=\left\{\begin{array}{c}0,n_y<T\\ 1,n_y>T\end{array}\right.---\left(9\right)$

In formula (9), n _ymean the number of pixels that gray level y has, T is the projection threshold value;

The value mode of R as the formula (10),

R＝min(n _valid，D) （10）。

7. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 6, and it is characterized in that, the span of described brightness regulation factor P is 2 to 6.

8. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 6, it is characterized in that the span 10 to 35 of described projection threshold value T.

9. the great dynamic range infrared image based on the guiding wave filter shows and detail enhancing method as claimed in claim 1, it is characterized in that, the computing method of the described final output image data of step 4 as the formula (11),

In formula (11), operational symbol

mean to reject the overflow value operation.

Images