KR101814806B1 - A method of enhancing contrast using bezier curve - Google Patents

Abstract

A histogram distribution is calculated from an image photographed or read into a storage medium, the image is classified according to a contrast state, the control point movement direction of a three-dimensional Bezier curve is determined for each type, and the histogram distribution And calculates the movement distance of the control point of the 3-dimensional Bezier curve based on the number of luminance values existing in the 3-dimensional Bezier curve. Based on the determined control point movement direction and the control point movement distance, two control points of the three-dimensional Bezier curve are moved, and two Bezier curves are generated using two moved control points and two fixed control points. Then, the image is improved by applying the generated Bezier curve.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contrast enhancement method using a Bezier curve,

The present invention relates to a method for improving the contrast of an image. More specifically, the present invention calculates a histogram distribution from an image photographed or read into a storage medium, classifies the image according to the contrast state, determines a control point movement direction of the three-dimensional Bezier curve for each type, , The control point shift distance of the 3-dimensional Bezier curve is calculated based on the number of luminance values in the histogram distribution. Based on the determined control point movement direction and the control point movement distance, two control points of the three-dimensional Bezier curve are moved, and two Bezier curves are generated using two moved control points and two fixed control points. Then, the image is improved by applying the generated Bezier curve.

Contrast has a significant impact on image quality in human visual perception. Improving image contrast is basically to improve the mutual translability or perception of image information for the human eye and to provide "better" input for other automated image processing techniques. The main purpose of image enhancement is to change the properties of an image to make it more suitable for a given task and a particular observer. In this process, one or more attributes of the image are changed.

Typically, the contrast of an image is defined as the range of intensity intensities available in a given image. Mathematically, the image contrast can be expressed as: " (1) "

Contrast = (Lmax - Lmin) / (Lmax + Lmin) (1)

Here, Lmax and Lmin are the maximum luminance value and the minimum luminance value of a given image, respectively.

Contrast enhancement is a process that makes the image features more distinct by using the maximum range of intensity provided by the number of bits used to quantify the intensity.

Among these contrast improvement methods, there is a need for a method of improving the contrast so as to be more optimized for human eyes.

It is an object of the present invention to provide a method of improving contrast in a spatial domain that adaptively improves contrast according to the type of an image.

According to an embodiment of the present invention, a method of improving contrast using a Bezier curve is provided. The method includes: reading an image; Calculating a luminance value distribution of the image; Calculating a pixel count for each of the low luminance, the medium luminance and the high luminance section of the luminance value; Determining a type of the image based on the interval pixel count; Determining a control point movement direction based on the type of the image; Calculating a movement distance of the control point based on a luminance value distribution of the image; Moving a control point based on the control point moving direction and the control point moving distance and generating a three-dimensional Bezier curve based on the moved control point; And performing contrast enhancement using the generated three-dimensional Bezier curve.

According to the present invention, it is possible to improve the contrast so that it can be more optimized in the human eye with a small amount of calculation.

Figure 1 shows the types of images according to the luminance value histogram distribution of an image.
2 shows a histogram distribution of an image according to an embodiment of the present invention.
FIG. 3 illustrates a method of setting control points of a three-dimensional Bezier curve according to an embodiment of the present invention.
FIG. 4 illustrates a method of moving a control point for each image type according to an embodiment of the present invention.
5 is a diagram for explaining the number N of luminance values present in a luminance value distribution according to an embodiment of the present invention.
Figure 6 illustrates a method of moving a control point by a calculated travel distance, in accordance with an embodiment of the present invention.
7 is a reference curve using the finally obtained Bezier curve according to an embodiment of the present invention.
8 is a flow chart illustrating an adaptive contrast enhancement method, in accordance with an embodiment of the present invention.

Embodiments of the present invention will now be described in more detail with reference to the drawings.

Figure 1 shows the types of images according to the luminance value histogram distribution of an image. As shown in Fig. 1, (a) is a dark image, (b) is a bright image, (c) is an image with backlight, (d) is a low contrast image, .

The histogram distribution of the image can be divided into three regions as shown in FIG. 2, namely, a low luminance (L), a medium luminance (M), and a high luminance (H) region. Preferably, the section lengths of the low luminance section, the middle luminance section, and the high luminance section are set to be substantially the same. That is, the luminance values included in the low luminance section, the middle luminance section, and the high luminance section may be 85, 85, and 86 out of 256 discrete luminance values in total.

On the other hand, Bezier curves are frequently used parameter curves in computer graphics and related fields. Bezier curves are widely used in computer graphics to model smooth curves. In the present invention, a three-dimensional Bezier curve is used to improve the contrast.

A three-dimensional Bezier curve can be defined by the following equation.

... (Equation 2)

Here, P0, P1, P2, and P3 are the control points of the 3-D Bezier curve.

A three-dimensional Bezier curve can be used to model the problem of contrast enhancement for different scenarios dynamically based on scene or image content. By changing the position of the control points, all necessary curve shapes necessary for improving the contrast can be obtained. Depending on the position of the control points P0, P1, P2, and P3, the shape of the Bezier curve to be generated is different.

Hereinafter, how to set the position of the control point to obtain a curve to be used for contrast enhancement will be described according to an embodiment of the present invention. FIG. 3 illustrates a method of setting control points of a three-dimensional Bezier curve according to an embodiment of the present invention.

First, among the four control points P0, P1, P2, and P3, the control points P0 and P3 are fixed in the diagonal direction facing each other. That is, in Fig. 3, P0 is fixed to (0,0) and P3 is fixed to (1,1). P1 and P2 are located at the same positions as P0 and P3 respectively and move in the direction away from P0 and P3 by a distance determined by the three regions of the luminance value histogram distribution, that is, the pixel counts of low luminance, medium luminance and high luminance section do. The direction in which P1 and P2 move away from P0 and P3 may vary depending on the type of image.

The types of images can be classified into the five types shown in Fig. 1, and the type of image belongs to the pixel count belonging to each section of the luminance value histogram.

N (L) is a pixel count belonging to a low luminance section, N (M) is a pixel count belonging to a middle luminance section, and N (H) is a pixel count belonging to a high luminance section.

If (N (L)> N (M)) && ((N (L)> N (H))) then dark image;

Else If (N (H)> N (M)) && ((N (H)> N (L))) then bright image;

Else If (N (L)> N (M)) && (N (H)> N (M))) backlight image;

Else If (N (M)> N (L)) && ((N (M)> N (H))) then a low contrast image;

Else high contrast image.

Preferably, when the image is classified into other types except for the high contrast image, if the size of N (L), N (M), and N (H) is large, such as 10% or more of the total number of pixels of the image . There is a difference in the values of N (L), N (M), and N (H), so that it can be determined to be a type other than high contrast in terms of arithmetic. However, Is preferable. For example, if the total number of pixels in the image is 1780, then N (L) = 600, N (M) = 590, and N (H) = 590 are classified as dark images in the above classification, Because it is not big, you can just classify it as a high contrast image. When the pixel count difference per section is 1780 * 10% = 178 or more total pixels, it can be classified into other types except for the high contrast image. The above 10% criterion is a criterion for judging whether the pixel count difference is large for each section, and the value may be changed to 20% or 30% depending on the embodiment.

Referring to FIG. 4, if the image is classified as (a) dark image, P2 is fixed and is in the same position as P3, and P1 moves up. As a result, Bezier curves can be generated by P0, P1, P2, and P3.

If the image is classified as (b) bright image, P1 is fixed and is at the same position as P0, and P2 moves downward.

If the image is categorized as a backlit image, P1 moves up and P2 moves down.

If the image is classified as (d) low contrast image, P1 moves to the right and P2 moves to the left.

If the image is classified as a (e) high-contrast image, the control points do not move but a straight Bezier curve is generated. When contrast enhancement is performed using a straight-line Bezier curve, the input value and the improved value become equal to each other, so that there is substantially no change in the image.

The distance traveled by P1 and P2 is determined by the tonal contrast ratio of (Equation 3).

Tone contrast ratio (TC) = TC = N / (Lmax - Lmin + 1)

N is the number of luminance values present in the luminance value distribution, Lmax is the maximum luminance value of the luminance value distribution, and Lmin is the minimum luminance value of the luminance value distribution.

The number N of luminance values existing in the luminance value distribution will be described with reference to FIG. Assuming that the histogram distribution as shown in FIG. 5 is calculated, there is no pixel count for all luminance values from 0 to 255, but a pixel count may exist only for specific luminance values. In Fig. 5, pixel counts are concentrated at nine luminance values. In this case, N = 9. For a particular image, N may range from 1 to 256.

The moving distance (D) of the control points P1 and P2 is determined by (Expression 4).

D = (2? N-1) * TC (4)

For example, from the histogram distribution of an image, if the type of image is determined as a backlight image, N = 9, Lmax = 240, Lmin = 5, then the tone contrast ratio TC is:

TC = 9 / (240-5 + 1) = 0.038135

Since the image type is a backlit image, P1 will move up and P2 will move down, and the travel distance is determined as follows.

D = (2 9 -1) * 0.038135 = 9.72457

Therefore, P1 and P2 are shifted by 9.72457 units, that is, 9.72457 luminance values, respectively. At this time, in order to reduce the calculation amount, the movement amounts of P1 and P2 can be rounded off and determined as an integer value. For example, 9.72457 can be moved by 10 units rounded.

Referring to FIG. 6, there is shown a case where P1 and P2 move by 10 units respectively up and down. In FIG. 6, the x-axis and the y-axis are scaled by [0, 255]. If the x and y axes are scaled to [0,1], then the computed travel distance will be shifted by the distance divided by 256.

When a three-dimensional Bezier curve is generated with four control points obtained through such a process, as shown in FIG. The x-axis is the input luminance value and the y-axis is the output luminance value. The contrast can be improved by mapping a given image to a new luminance value using the reference curve as shown in Fig.

8 is a flow chart illustrating an adaptive contrast enhancement method, in accordance with an embodiment of the present invention.

In step S11, the image is read. The image can be photographed with a digital camera or image data stored in a storage medium.

In step S12, the RGB value of the read image is converted into YCbCr, and the Y value as the luminance component is extracted for each pixel.

In step S13, pixel counts for low luminance, medium luminance, and high luminance sections are calculated based on the extracted Y for each pixel.

In step S14, the image type is determined based on the pixel count for each section. The image type may be one of the five types of FIG. 4 as described above.

In step S15, the direction of movement of the control point is determined based on the determined image type. As described above, P0 and P3 are fixed to (0,0) and (1,1), respectively, and P1 and P2 are moved up, down, right, or left, respectively.

In step S16, the movement distance of the control points P1 and P2 is calculated based on the histogram distribution of the luminance values. At this time, as described above, the moving distance can be calculated in the above-described manner by extracting the number N of luminance values in which the pixel count exists among the luminance values of 0 to 255 in the histogram distribution of the luminance values.

In step S17, a three-dimensional Bezier curve is generated based on the moved control points P1 and P2 and the fixed control points P0 and P3.

The contrast enhancement is performed using the Bezier curve generated in step S18 or the reference curve scaled from 0 to 255. [

Claims (5)

A method for improving contrast using a Bezier curve,
Reading an image;
Calculating a luminance value distribution of the image;
Calculating a pixel count for each of the low luminance, the medium luminance and the high luminance section of the luminance value;
Determining a type of the image based on the interval pixel count;
Determining a control point movement direction based on the type of the image;
Calculating a movement distance of the control point based on a luminance value distribution of the image;
Moving a control point based on the control point moving direction and the control point moving distance and generating a three-dimensional Bezier curve based on the moved control point; And
And performing contrast enhancement using the generated three-dimensional Bezier curve,
The step of calculating the movement distance of the control point based on the brightness value distribution may include:
Extracting a number N of luminance values existing in the luminance value distribution;
Calculating a tone contrast ratio (TC) based on the number of luminance values;
And calculating a travel distance based on the tone contrast ratio (TC)
The tone contrast ratio (TC)
TC = N / (Lmax - Lmin + 1)
The moving distance (D)
D = (2 " N-1) * TC,
Wherein Lmax is a maximum luminance value of the luminance value distribution, and Lmin is a minimum luminance value of the luminance value distribution. The method according to claim 1,
Wherein the image type is one of a dark image, a bright image, a backlight image, a low contrast image, or a high contrast image. The method according to claim 1,
Wherein determining the type of the image based on the interval pixel count comprises:
Determining a dark image if the pixel count of the low luminance section is greater than the pixel count of the middle luminance section and the pixel count of the low luminance section is greater than the pixel count of the high luminance section,
Determining a bright image if the pixel count of the high luminance section is greater than the pixel count of the middle luminance section and greater than the pixel count of the high luminance section,
Determining a backlight image if a pixel count of the low luminance section is greater than a pixel count of the middle luminance section and a pixel count of the high luminance section is greater than a pixel count of the middle luminance section,
Determining a low contrast image if the pixel count of the medium luminance section is greater than the pixel count of the low luminance section and the pixel count of the medium luminance section is greater than the pixel count of the high luminance section,
And determining the other case as a high contrast image. delete The method according to claim 1,
Wherein the control point is shifted by an integer rounded by the calculated movement distance (D).

Images