KR101877808B1 - Image contrast enhancement method and apparatus using local histogram equalization based on multiple layters overlapped block - Google Patents

Abstract

The present invention discloses a method and apparatus for enhancing image contrast using local histogram smoothing based on a multi-layer overlap block. According to another aspect of the present invention, there is provided a method for enhancing an image local contrast, comprising: converting brightness values of an input image according to multiple layers and generating multi-layer images having brightness values; Reducing noise of multi-layer images; And generating an enhanced local contrast image using the noise-reduced multi-layer images. According to the present invention, an image having improved image quality can be obtained as compared with the conventional algorithm.

Description

[0001] The present invention relates to a method and apparatus for enhancing image contrast using local histogram smoothing based on a multi-layer overlap block,

The present invention relates to a method and apparatus for enhancing image contrast, and more particularly, to a method and apparatus for enhancing image contrast using local histogram smoothing.

The contrast enhancement of the image is intended to reveal details that are not visible in the image and to improve the quality of the image. Local histogram equalization is used to emphasize the local information of the image and is used to improve the contrast of the image.

Previous studies related to local histogram smoothing include Strickland Robin N, "Image Processing Techniques for Tumor Detection, Image Processing, Marcel Dekkler, Inc, 2002", Yu TH, Histogram-Shape Preserving Algorithm for Image Enhancement, IEEE ISCAS, Vol. 1, pp. 407-410, 1993, Nicholas Sia Pik Kong, "Multiple Layers Block Overlapped Histogram Equalization for Local Content Emphasis, Computers and Electrical Engineering Vol.37, p.631-643, 2010 ".

An object of the present invention is to extend the existing local histogram smoothing technique to improve the contrast of an image, thereby improving the contrast of the image more efficiently and quickly and improving the quality of a resultant image.

According to an aspect of the present invention, there is provided a method of enhancing image contrast, comprising: converting brightness values of an input image according to multi-layers and generating multi-layer images having brightness values converted; ; And generating an enhanced local contrast image using the noise-reduced multi-layer images.

In the present invention, it is preferable that the noise of the multi-layer images is reduced by applying a bilateral filter to each of the multi-layer images.

In the present invention, the multi-layer may include a first layer for redistributing brightness values of the input image, a second layer for improving the contrast of a small object included in the input image, and a third layer for improving the sharpness of the input image .

In the present invention, the step of generating an enhanced local contrast image may include generating a multi-layer integrated image in which noise-reduced multi-layer images are summed according to predetermined weights; And generating an enhanced local contrast image using the multi-layer integrated image and the input image.

In the present invention, the brightness values of the input image are converted according to the multi-layers, and the brightness values of the input image are converted through the histogram smoothing that applies windows of different sizes defining the multiple layers to the input image , And the histogram smoothing is preferably performed by a method of limiting the contrast of the local histogram. Here, restricting the contrast may be configured to limit the contrast by distributing a histogram value exceeding a predetermined reference value in all the histogram regions in the local histogram. The predetermined weight may be determined in consideration of the entropy of the multi-layer images. In particular, it is preferable that the average entropy of the multi-layer images and the ratio of the relative difference between the entropy of each of the multi-layer images are determined.

According to another aspect of the present invention, there is provided a computer-readable recording medium having recorded therein source code programmed to perform the above-described method for enhancing an image contrast.

According to another aspect of the present invention, there is provided an apparatus for enhancing an image local contrast, comprising: a multi-layer image converting unit for converting a brightness value of an input image according to multi- Generating unit; A noise reduction unit for reducing noise of the multi-layer images; And an integrated image generating unit for generating an enhanced local contrast image using the noise-reduced multi-layer images.

According to the present invention, it is possible to efficiently remove noise from a resultant image, and to save an object or an edge component, which is an important component of an image, thereby improving the contrast of the image more accurately and efficiently.

FIG. 1 is a reference diagram for explaining the concept of a histogram smoothing technique (NOBHE) using non-overlapping blocks.
FIG. 2 is a reference diagram for explaining the concept of a histogram smoothing technique (BOHE) using overlapping blocks.
3 is a block diagram showing an image contrast enhancement apparatus according to an embodiment of the present invention.
4 is a reference view showing an output image of each component of the image contrast enhancement apparatus of FIG.
FIG. 5 shows an output image of the multi-layer image generating unit of FIG. 3, FIG. 6 shows an output image of the noise reducing unit of FIG. 3, FIG. 7 shows an output image of the first integrated image generating unit, And an output image of the image generating unit.
9 is a flowchart illustrating an image contrast enhancement method according to an exemplary embodiment of the present invention.
10 and 11 are reference views for comparing a resultant image according to an embodiment of the present invention with a resultant image according to an existing method.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. Those skilled in the art will be able to implement the present invention even further considering the contents described in the following description or drawings and the known technology. It is to be understood that the scope of the present invention is not limited to the following embodiments and drawings, since all the conditional terms and embodiments listed in the present specification are intended to be understood in principle as the concept of the present invention.

It is also to be understood that all statements relating to the present invention are intended to cover structural, conceptual, and functional equivalents. It is also to be understood that such equivalents may include all elements contemplated to perform the same function irrespective of the currently known equivalents as well as the equivalents to be developed in the future. It should also be understood that the block diagrams and flow charts herein represent exemplary conceptual aspects embodying the principles of the invention.

Also, the time-series concepts included herein may be implemented in computer-readable source code by those skilled in the art, and such source code may reside in a computer-readable storage medium, Lt; RTI ID = 0.0 > and / or < / RTI > The explicit use of terms such as processor, control, or similar concepts should not be interpreted exclusively as hardware capable of executing the software, but may include without limitation digital signal processor (DSP) hardware, ROM), random access memory (RAM), and non-volatile memory. Other hardware may also be included.

The components represented as means for performing the functions of the present invention include, for example, all types of software including circuitry, or combinations thereof, and firmware / microcode, etc., performing the functions described herein , And may be implemented in combination with appropriate circuitry for executing the software to perform the functions described herein.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings, in which: will be. In the following description, a detailed description of known technologies related to the present invention will be omitted when it is determined that the gist of the present invention may be unnecessarily blurred.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the concept of image contrast enhancement according to the present invention will be described, and a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

The present invention relates to a method of enhancing the contrast of an image in order to improve the quality of an image, and more particularly, to a method of enhancing contrast of an image based on a multi-layer block overlap histogram smoothing technique which extends the existing local histogram smoothing technique Thereby improving the local contrast of the image and eliminating the noise of the resultant image, thereby improving the quality of the image. The most representative and frequently used method to improve the image quality is image enhancement. The main purpose of enhancing image contrast is to reveal details that are not visible in the image and improve the quality of the image. Therefore, the contrast enhancement technique can be used as a post-processing technique in an automatic processing system such as a television, a mobile, or a camera.

The global histogram smoothing technique has the advantage of low computational complexity and simplicity, but it does not over-enhance the bright or dark pixels in the resulting image. Over-equalization problems or washed-out effects.

Local histogram smoothing is a method to solve the problem of global histogram smoothing. The local histogram smoothing technique defines a Contexual Region (CR) for a middle pixel of a window using a small window and calculates a cumulative density function (CDF) for a block of pixels away therefrom to determine a cumulative density function To improve the contrast. The conversion function for each CR for local histogram smoothing is given by Equation 1 below.

[Equation 1]

Where x is the gray level, L is the intensity range of the image, C _CR is the CDF at the current CR and can be defined as Equation 2 below.

&Quot; (2) "

는 CR의 누적 히스토그램이며, N_CR은 CR 내의 픽셀들의 전체 수 이다. here,

Is the cumulative histogram of the _CR , and N _CR is the total number of pixels in the CR.

Local histogram smoothing techniques include histogram smoothing (NOBHE), block overlap histogram smoothing (BOHE), partially overlapping subblock histogram smoothing (POSHE), and adaptive histogram smoothing (CLAHE), which do not overlap the blocks. In the case of the histogram smoothing (NOBHE) method in which blocks are not overlapped, a block which is not overlapped is used to smooth the histogram of the image. The blocks are treated as one individual image, and the histogram smoothing .

The histogram smoothing (NOBHE) technique in which blocks do not overlap can be understood with reference to Fig. That is, the histogram smoothing method based on NOBHE performs histogram smoothing by moving subblocks so that the subblocks do not overlap each other. The smoothing result of each block is reassembled into one result image by rearranging to the original position in the original image. This method results in blocking artifacts at the block boundary in the resulting image, although each pixel in the image contributes to one CDF, which is faster than the relatively different local histogram smoothing techniques.

The Block Overlap Histogram Smoothing (BOHE) technique can be understood with reference to FIG. This technique can solve the problem of the NOBHE technique mentioned above, but uses overlapping blocks instead of non-overlapping blocks. One window is used to move it one pixel at a time, and only the center pixel in the CR is updated with the new CDF value. Thus, the blocking artifacts can be reduced because the blocks overlap each other. However, it has a disadvantage that the calculation time is considerably long and the speckle noise is amplified in a relatively flat region.

The partial overlapped subblock histogram smoothing (POSHE) technique is effective in reducing the blocking artifact while reducing the computational complexity of the BOHE technique mentioned above. This technique can be thought of as a mixture of NOBHE and BOHE. As with other LHE techniques, it is not easy to improve the unnatural contrast and determine the appropriate window size in the resulting image.

The histogram smoothing (CLAHE) technique, which adaptively limits the contrast, is effective in reducing the amplification of the noise level, which is a problem associated with the BOHE technique. This technique can reduce the amplification of noise in the resulting image by limiting the contrast in the local histogram. In the case of the CLAHE technique, the histogram is clipped to a preset value before calculating the cumulative density function for histogram smoothing. In this case, instead of discarding the histogram value exceeding the limited clipping value, the excess histogram value is redistributed equally to all the histogram regions. However, if the clipping value is not appropriately selected, the quality of the image may deteriorate considerably. Therefore, proper selection of the clipping value is important.

The problems common to the above-described local histogram smoothing (LHE) techniques are as follows. First, most LHE techniques cause unnatural contrast enhancement and tend to enhance speckle noise. Also, the shape of the histogram of the input image must be maintained in order to maintain the information contained in the input image. In addition, the average intensity brightness of the input image and the output image must be maintained to avoid the problem of intensity saturation artifacts occurring in the output image.

The present invention relates to an image contrast enhancement apparatus and method for mitigating the problems of the above-described image contrast enhancement technique, particularly LHE techniques, and for enhancing the quality of an output image. Hereinafter, The image contrast enhancing device will be described in more detail.

3 includes an input unit 110, a multi-layer image generating unit 120, a noise reducing unit 130, and an integrated image generating unit 140. Basically, the image contrast enhancement apparatus 100 according to the present embodiment basically performs a smoothing operation based on a multi-layer block overlap histogram smoothing technique, and applies an adaptive histogram smoothing technique for limiting the contrast with respect to brightness value change And a new by-product filter is used in the use of the filter.

The input unit 110 receives an input image, which is an object to improve image contrast. In this embodiment, the input unit is represented as a component, but it is also possible to configure the multi-layer image generator 120 to receive the input image without the input unit.

The multi-layer image generating unit 120 converts brightness values of the input image according to preset multi-layers, and generates multi-layer images in which the brightness values are converted. Here, the multi-layer means a plurality of image layers obtained by differentiating the brightness value of the image.

The multi-layer image generation unit 120 may change the brightness value of the input image through the histogram smoothing technique. Histogram smoothing can change the brightness value by using block overlap based histogram smoothing (BOHE), multilayer block overlap based histogram smoothing (MLBOHE), and adaptive histogram smoothing (CLAHE).

In particular, the multi-layer image generation unit 120 of the present embodiment modifies the brightness value of the input image based on the multi-layer block overlap-based histogram smoothing technique. In the histogram smoothing method based on the multi - layer block overlap, the brightness value of the input image is modified by further considering the adaptive histogram smoothing technique which limits the contrast while reducing the excessive amplification of the noise.

The images of different layers generated by the multi-layer image generating unit 120 are used to unhide details hidden in the image and uniformly distribute the brightness of the image, and are integrated after the noise removal process. However, when the block overlap histogram smoothing (BOHE) technique is used, speckle noise, which is a common problem in existing LHE techniques, can not be completely eliminated. Therefore, it is desirable to modify the brightness value based on the adaptive histogram smoothing (CLAHE) technique in order to limit the contrast while reducing the over-amplification of the noise appearing in the resultant image.

This adaptive histogram smoothing technique is effective to reduce the noise amplification in the resultant image by limiting the contrast in the local histogram. The amplification of the spark noise can be limited by clipping the histogram to a preset value before calculating the cumulative density function for histogram smoothing. That is, the multi-layer image generating unit 120 of the present embodiment uniformly redistributes the brightness values by uniformly redistributing the histograms to all the histogram regions instead of discarding the histogram values exceeding a predetermined clipping value, Amplification can be reduced. In addition, the redistribution of brightness values based on the CLAHE technique has the advantage of faster computation than conventional local histogram smoothing techniques. This brightness value modification can be used to obtain good performance images in general standard image and medical image.

3, the multi-layer image generation unit 120 of the present embodiment includes a first image layer generation unit 122, a second image layer generation unit 124, and a third image layer generation unit 126 . The generated three images are images of different layers from the input image, which is an original image, with improved contrast through block overlap histogram smoothing based on CLAHE.

FIG. 4 shows an example of an output image for each configuration of the image contrast enhancement apparatus shown in FIG. In the present embodiment, when the size of an input image as an original image is MxN, a first image layer generating unit 122, a second image layer generating unit 124, and a third image layer generating unit 122, which are used for block overlap histogram smoothing, The size of the window used in the unit 126 may be (M / 2) x (N / 2), (M / 4) x have.

4, an image 222 is an output image of the first image layer generating unit, an image 224 is an output image of a second image layer generating unit, and an image 226 is an output image of a third image layer generating unit. The multi-layer of this embodiment includes a first image layer, a second image layer, and a third image layer, wherein a first image layer according to a window of (M / 2) x (N / 2) (M / 4) x (N / 4) size window is intended to enhance a small object of the image, and the second image layer according to the size of (M / The third image layer according to the window of FIG.

The noise reduction unit 130 receives the images 222, 224, and 226 generated by the multi-layer image generation unit 120 and reduces the noise of the input images. In the case of the histogram smoothed images generated by the multi-layer image generating unit, the noise level of the speckle noise is enhanced. The noise reducing unit 130 removes It functions to mitigate.

The noise reduction unit 130 generates images 232, 234, and 236 with noise removed or reduced through noise removal filtering on the images 222, 224, and 226 whose brightness values are changed. In this case, the noise filtering is preferably median filtering, and the bilateral filtering is more preferable. In particular, bilateral filtering is more preferable.

First, the median filtering method has a higher noise level in the resultant image than in the case of histogram smoothing with a large window size, so that h ₁ , h ₂ , h ₃ It is possible to perform median filtering on windows of (3x3), (5x5), and (7x7) sizes, respectively. However, in the case of the median filter, spark noise, which is a common problem in local histogram smoothing techniques, can not be effectively removed. Since the median filter removes only a part of the speckle noise and mainly removes random noise, sometimes the object or the edge component of the image is not preserved and the entire image blurs, There is a limit in that image degradation phenomenon appears.

The noise reduction unit 130 of the present embodiment is configured to reduce noise by using a bilateral filter which is a filter that efficiently smoothing an image and reducing noise while preserving edge components of the image.

The bilateral filter includes two simple Gaussian filters, and has the advantage that the computation complexity is not high. First, when I (x) and I (y) are pixel values of the brightness-converted image, the filtered result can be expressed by Equation (3) below.

&Quot; (3) "

Where c (), s () represent the pixel location associativity and similarity between pixels, respectively, and? Represents the neighboring pixel locations that include the current location. The normalization term k () can be expressed by Equation (4) below.

&Quot; (4) "

Further, in order to give a higher weight to the pixels spatially adjacent to the middle pixel, c () can be expressed by the following equation (5).

&Quot; (5) "

는 c() 의 L₂-norm 과 표준편차를 나타낸다.here,

Represents the L ₂ -norm and standard deviation of c ().

The s () indicating the similarity between x and y pixels is determined by the similarity of the pixel values so that the pixel values I (x) and I (y) can be used as parameters of the Gaussian filter instead of the pixel positions x and y, Weights are given. Therefore, s () can be expressed by Equation (6) below.

&Quot; (6) "

Here, σ _s represents the standard deviation of s ().

Since the adaptive combination of c () and s () allows the Gaussian filter to be processed along the edge direction, the bilateral filter effectively separates the textual information and structural information of the image compared to the conventional Gaussian filter .

Adaptive histogram by applying a smoothing techniques improve the local contrast in which the resultant image is h _1, h _2, for each of the _{h 3 (3x3), (5x5} ), by lateral filter with a window size of (7x7) to limit the contrast The resulting images are denoted h _1s , h _2s , and h _3s , respectively.

FIG. 6 shows an image resulting from bilateral filtering. In the case of the bilateral filtering, the edge component of the image is well preserved, and the noise component is efficiently removed. As a result, it can be confirmed that the filter performs better than the median filtering.

The integrated image generating unit 140 generates an integrated image having improved local contrast by using multi-layer images with reduced noise. The integrated image generating unit 140 of the present embodiment includes a first integrated image generating unit 142 and a second integrated image generating unit 144.

The first integrated image generation unit 142 receives the noise-filtered images 232, 234, and 236 from the first through third noise reduction units, and integrates the filtered images. The second integrated image generating unit 144 re-integrates the combined image in the first integrated image generating unit 142 and the input image of the input unit 110 to finally obtain an enhanced contrast image.

The integrated image generator 140 does not significantly change the mean intensity brightness of the input image and the output image in order to solve intensity saturation in the output image, which is one of the problems in LHE techniques. That is, the three result layer images h _1s , h _2s , h _3s Are combined into one image and then combined with the input image. First, the process of merging the three layer images h _1s , h _2s , and h _3s into one image P is shown in Equation (7).

&Quot; (7) "

Here, w ₁ , w ₂ , and w ₃ are positive weights. In order to make the weights dependent on the image, it is desirable to set the ratio of these values to be proportional to the entropy value according to Equation (8) below. According to the following equation, the first integrated image generator generates a combined image P by weighting the ratio between the entropy of the images and the input image.

&Quot; (8) "

Here, E _h1s , E _h2s , and E _h3s represent the entropy values of the noise-reduced images h _1s , h _2s , and h _3s , respectively, and the entropy can be obtained by the following equation (9).

&Quot; (9) "

That is, according to the present embodiment, the layer having the highest entropy difference value has the highest weight.

The second integrated image generation unit 144 may obtain the final output image Y according to Equation (10) by adding the integrated image P and the input image X generated by the first integrated image generation unit 142. [

&Quot; (10) "

Where α and β are positive weighting coefficients and σ + β = 1.

As the weight of? Increases, the noise level increases. When the weight of? Increases, the noise level decreases but the contrast does not improve properly. For the optimum result image, the values obtained from the experiment are preferably in the range of 0.6 to 0.8 and in the range of 0.2 to 0.4, particularly preferably in the range of σ = 0.7 and β = 0.3.

FIG. 7 shows an example of obtaining an image P (242) integrated in the first integrated image generation unit 142 and FIG. 8 shows an example of obtaining an integrated image Y (244) integrated in the second integrated image generation unit 144. FIG. . The resultant image Y obtained by the second integration can obtain the result that the saturation artifact is removed and the average luminance of the original image is maintained.

The apparatus and method for enhancing image contrast of the present invention described in the above embodiments of the present invention reveal details that are not visible in the image and improve the quality of the image. Therefore, the contrast enhancement technique is suitable to be used as a post-processing technique in an automatic processing system such as television, mobile, and camera.

9 is a flowchart illustrating an image contrast enhancement method according to an exemplary embodiment of the present invention. The method shown in FIG. 9 is performed in a time-series manner in the image contrast enhancement apparatus 100 shown in FIG. 3 and includes the following steps. The foregoing and common points are excluded from the following description.

In operation 1100, the multi-layer image generator 120 converts brightness values of the input image through histogram smoothing using windows having different sizes. In particular, the multi-layer image generating unit performs histogram smoothing based on block overlap using a window of a different size, and histogram smoothing is performed by clipping the histogram to a preset value before calculating the CDF.

In step 1200, the noise reduction unit 130 reduces noise through noise filtering on images according to the multi-layer in which brightness values are converted. In particular, when noise is reduced through bilateral filtering, the noise reduction unit 130 can more effectively reduce edge components or spark noise.

In step 1300, the first integrated image generation unit generates a multi-layer integrated image P obtained by summing the noise-reduced multi-layer images according to weights.

In step 1400, the second integrated image generating unit obtains a multi-layer integrated image P and an image Y having an improved contrast finally by summing the input image according to a predetermined weight.

Meanwhile, the image contrast enhancement method of the present invention can be implemented by a computer-readable code on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored.

Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device and the like, and also a carrier wave (for example, transmission via the Internet) . The computer readable recording medium may also be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers skilled in the art to which the present invention belongs.

[Picture Quality Evaluation]

The performance evaluation of the algorithm was evaluated by subjective image quality evaluation, entropy, and AMBE for general standard and medical images. With respect to entropy, this can measure the degree of improvement obtained from the resulting image. That is, the higher the quality of the image, the higher the entropy value as compared with the input image.

&Quot; (11) "

Another evaluation is AMBE (average mean brightness error), which is a measure of how much the average brightness value of the input image is maintained in the output image, and therefore, the higher the quality of the image, the lower the AMBE value.

&Quot; (12) "

Here, x and y represent the average brightness values of the input image and the output image, respectively. In this performance evaluation, the clipping value of the CLAHE algorithm is set to 10% of the histogram maximum value, and the two parameters of the bilateral filter are set to σ _c = 1.8 and σ _s = 30.

Table 1 shows that both entropy and AMBE perform better in the bilateral filter than in the median filter. Especially, for the baboon, chest2, and teeth images, the entropy is lower than the original image when the median filter is used. In the case of using the bilateral filter, the image is better for both the original image and the median filter. can confirm.

video Entropy AMBE original Median By letter Median By letter couple 1.1913 2.0187 2.3274 16.3271 10.2543 house 1.8862 2.0898 2.3092 11.6973 7.5172 sailboat 2.1267 2.1314 2.4639 10.4276 3.4099 baboon 2.1762 2.1150 2.3758 6.4344 2.1241 chest1 2.1499 2.1858 2.4031 1.4638 0.2538 chest2 2.1242 2.0877 2.3382 18.0214 5.4979 mammo. 2.1923 2.2144 2.3351 11.3652 3.7488 teeth 2.1711 2.1639 2.3931 12.1849 8.0738

In Table 2 and 3, entropy and AMBE of the existing algorithms and the well known LHE techniques BOHE, CLAHE and proposed algorithm are compared. For the existing LHE techniques as well as the existing algorithms, the proposed algorithm shows better performance in entropy and AMBE cmrausa.

10 and 11 show subjective image quality evaluation with respect to existing algorithms for the sailboat, which is a standard standard image, and chest2, which is a medical image, respectively.

video Entropy original LHE CLAHE MLBOHE Proposed couple 1.1913 2.3491 2.2567 2.1218 2.3274 house 1.8862 2.3109 2.2264 2.1474 2.3092 sailboat 2.1267 2.3450 2.3298 2.2202 2.4639 baboon 2.1762 2.3563 2.3663 2.1982 2.3758 chest1 2.1499 2.3551 2.2894 2.2603 2.4031 chest2 2.1242 2.3342 2.1437 2.2519 2.3382 mammo. 2.1923 2.2867 2.2907 2.2801 2.3351 teeth 2.1711 2.3369 2.2922 2.2761 2.3931

video AMBE LHE CLAHE MLBOHE Proposed couple 127.6431 48.0835 35.5553 10.2543 house 92.9383 30.9292 26.4576 7.5172 sailboat 35.1214 12.7118 7.2509 3.4099 baboon 29.8809 5.9425 5.5733 2.1241 chest1 19.4944 16.4358 8.9594 0.2538 chest2 72.8794 20.2154 19.5250 5.4979 mammo. 36.8343 10.1805 7.9010 3.7488 teeth 42.6450 15.2910 9.4552 8.0738

In both of FIGS. 10 and 11, it can be seen that the resultant image improves unnaturally in the case of LHE (b). In the case of the proposed algorithm, the noise is removed from the sailboat image shown in FIG. 8, and the edge is preserved and the contrast is naturally improved And the contrast is greatly improved in the vicinity of the ribs and the spine of the Chest 2 image of FIG. 11, and it is confirmed that the image is good in the medical image.

In addition, by using the technique of smoothing the histogram adaptively by restricting the contrast rather than the normal block overlap histogram smoothing technique, the entropy and the AMBE in the result image, the subjective image quality, and the LHE technique, .

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims and the claims.

Claims (14)

The present invention relates to an image local contrast enhancement method,
Converting a brightness value of an input image according to a layer and generating multi-layer images having brightness values converted;
Reducing noise of the multi-layer images; And
And generating an image having enhanced local contrast using the noise-reduced multi-layer images,
Wherein the step of converting the brightness values of the input image according to the multi-layer comprises: converting brightness values of the input image through adaptive histogram smoothing that applies windows of different sizes defining the multiple layers to the input image; Characterized in that before the cumulative density function for histogram smoothing is calculated, a histogram value exceeding a predetermined clipping value in the local histogram is redistributed equally to all histogram regions to uniformly redistribute the brightness value to reduce excessive amplification of the spark noise. To improve image contrast. The method according to claim 1,
Reducing the noise of the multi-
And applying a bilateral filter having windows of different sizes to each of the multi-layer images to reduce speckle noise. The method of claim 1,
A first layer for redistributing brightness values of the input image,
A second layer for improving the contrast of a small object included in the input image, and
And a third layer for improving the sharpness of the input image. The method according to claim 1,
Wherein the step of generating the enhanced local contrast image comprises:
Generating a multi-layer integrated image in which noise-reduced multi-layer images are summed according to predetermined weights; And
Further comprising the step of generating an enhanced local contrast image using the multi-layer integrated image and the input image. delete delete 5. The method of claim 4,
And determining entropy of the multi-layer images. 5. The method of claim 4, wherein the predetermined weight is
Wherein a ratio of a difference between an average entropy of the multi-layer images and an entropy according to each of the multi-layer images is determined. A method for enhancing image contrast according to any one of claims 1, 2, 3, 4, 7, and 8, comprising the steps of: Recording medium. An apparatus for enhancing an image local contrast,
A multi-layer image generation unit for converting a brightness value of an input image according to multiple layers and generating multi-layer images in which brightness values are converted;
A noise reduction unit for reducing noise of the multi-layer images; And
And an integrated image generating unit for generating an enhanced local contrast image using the noise-reduced multi-layer images,
Wherein the multi-layer image generation unit transforms brightness values of the input image through adaptive histogram smoothing that applies windows of different sizes defining multiple layers to the input image, and calculates an accumulated density function for histogram smoothing Wherein the histogram value exceeding a predetermined clipping value in the local histogram is redistributed equally to all histogram regions to uniformly redistribute the brightness value to reduce excessive amplification of the spark noise. 11. The apparatus of claim 10, wherein the noise reduction unit reduces a spark noise by applying a bilateral filter having windows of different sizes to each of the multi-layer images. 11. The method of claim 10,
A first layer for redistributing brightness values of the input image,
A second layer for enhancing contrast of a small object contained in the image, and
And a third layer for improving the sharpness of the input image. The image processing apparatus according to claim 10,
Layer-based first integrated image generation unit for generating a multi-layer integrated image in which noise-reduced multi-layer images are summed according to predetermined weights,
Further comprising a second integrated image generator for generating an enhanced local contrast image using the multi-layer integrated image and the input image. delete

Images