KR100933556B1 - Color image processing apparatus and method for extending the dynamic range - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing apparatus having a wide dynamic range, comprising: a long exposure / short exposure image acquisition circuit for acquiring long and short exposure video signals from an image pickup device, and an output of the long exposure / short exposure image acquisition circuit. A luminance / color difference signal conversion circuit that separates a video signal into luminance and color difference signals, and a luminance signal combining circuit which receives luminance signals of long and short exposures from the luminance / color difference signal conversion circuit and compresses the luminance by a weighted average technique; And a chrominance signal post-processing circuit for correcting a signal attenuated in the luminance compression process, and a chrominance signal using a weighted average method according to the intensity of saturation by receiving chrominance signals of long and short exposures from the luminance / color difference signal conversion circuit. A color difference signal combining circuit to combine, and an output luminance signal of the luminance signal post-processing circuit and a color difference signal of the color difference signal combining circuit. Relates to a color image processing apparatus and method that extends the dynamic range consisting of RGB inverse transformation circuit.

Image processing device, dynamic range, color difference signal coupling, luminance signal coupling

Description

Color image processing apparatus and method for expanding dynamic range

The present invention relates to an image signal processing apparatus having a wide dynamic range. The present invention relates to a long-exposure image and a short-exposure image by dividing the color image into a color difference and luminance to maintain the clarity of saturation and to correct the signal attenuated during the luminance compression process. An extended color image processing apparatus and method are disclosed.

In image signal processing, studies are being actively conducted to reproduce images captured by a camera so as to be closer to human visual characteristics. Indeed, the displacement, or dynamic range, that a human can see from bright to dark is about 120 dB. However, in general, the dynamic range of an image processed in an image processing system using a solid-state image pickup device (CCD) is only about 60 dB. As a result, since it is difficult to obtain sensitivity characteristics such as human vision from an image having dark and bright luminance levels, it is necessary to enlarge the dynamic range of the image to overcome this problem.

As a method of enlarging the dynamic range of the above-described image, there are largely a method of improving an analog circuit at a sensor level and a method through digital image processing. Among them, the present invention relates to a digital image processing method, which combines images acquired by multiple exposure to increase dynamic range.

In this case, an important factor in combining the long exposure image and the short exposure image by multiple exposure is a question of how to combine the saturation region and the low luminance region of the short exposure region into appropriate exposure image information.

Conventionally, a method of combining two images of a long exposure image and a short exposure image by simply a weighted average is used. A simple average combination of two images has the advantage of expressing the dynamic range without loss most with respect to luminance characteristics. However, in terms of color, it does not properly reflect saturation, and thus has a disadvantage in that a clear image cannot be realized.

In the prior art, ① a patented video signal processing device having a wide dynamic range and a registered patent 10-0672856 and a 10-0363827 'TV signal processing device for generating a video signal having a wide dynamic range and a signal processing device are described. Branches' television cameras and television signal processing methods.

1 is a block diagram of a conventional camera video signal processing apparatus 1, and FIG. 2 is a block diagram of a conventional television camera and a television signal processing method ②.

First, referring to FIG. 1, a configuration of a conventional camera image signal processing apparatus 1 includes: a long exposure / short exposure acquisition circuit for acquiring long and short exposure video signals from an image pickup device; A difference circuit for calculating a difference value between the two video signals; A multiplication coefficient calculating circuit configured to generate a multiplication coefficient by inputting a multiplication ratio coefficient for producing a multiplication coefficient which varies according to the difference image value and the difference image obtained from the difference circuit; First and second multiplication circuits for multiplying each image by a multiplication coefficient of the two image signals; And a synthesis circuit for synthesizing two images obtained by the multiplication.

In the above-described prior art ①, the weighted coefficient is determined by a difference between the luminance values of the long exposure and the short exposure and a specific parameter Rls, and the two images are combined to allow the result of the synthesized image to naturally respond to the change in the amount of light.

However, the conventional camera image signal processing apparatus ① does not separate and process RGB components into luminance and chrominance components, thereby combining luminance on each domain of RGB.

In addition, when the Rls parameter is lowered to increase the combined reflectance of the short exposure image, an area having a low gray level of the short exposure image is reflected as an awkward image (reverse phenomenon) in the combined image.

On the other hand, referring to Fig. 2, the structure of the conventional television camera and the television signal processing method ② will be described. The peak value detection circuit 52 for detecting the peak value P in one field period of the high luminance video signal V2, for example, will be described. )Wow; A multiplication coefficient calculating circuit 52 for calculating a multiplication coefficient L of the standard luminance video signal V1 and a multiplication coefficient S of the high luminance video signal V2 from the addition ratio R and the peak value P; A multiplication circuit 53 for multiplying the standard luminance video signal V1 by the multiplication coefficient L to output the table luminance video signal O1; A multiplication circuit 54 for multiplying the high luminance video signal V2 by the multiplication coefficient S to output a high luminance calculation signal O2; And a synthesizing circuit 55 for adding the standard luminance video signal O1 and the high luminance video signal O2 to output a wide dynamic range video signal W.

The overall image acquisition system of the prior art ② is the same as in the prior art ①, except that the histogram and various elements required to obtain the coupling coefficient in one frame are acquired and applied to the next frame, and the peak value is detected from the short exposure image. The process of stretching the exposure image is further included.

However, in conventional TV cameras and TV signal processing methods ②, when the peak value is high, the reflection ratio of the high illuminance image is generally lowered, and the overall synthesized image is darkened. There is a problem losing.

When the black area is taken and the saturated area (fluorescent lamp or white species) is taken, or the saturated area is taken while the black area is taken or the saturated pixels are irregularly detected in the dark image. In this case, there is a problem in that the image itself to be synthesized by a sudden peak value change also causes a sudden stretching.

In addition, there is a problem in that the brightness of the screen itself changes fluidly according to the height of the peak value detection value.

The present invention is to solve the above problems, a color image that extends the dynamic range to maintain the sharpness of the saturation in the dynamic range expansion process through digital image processing and to secure the signal attenuated in the luminance combining process It is an object to provide a treatment apparatus and method.

The above object of the present invention is to provide a video signal processing apparatus comprising: a long exposure / short exposure image acquisition circuit for acquiring long and short exposure video signals from an image pickup device; A luminance / color difference signal conversion circuit for separating the output image signal of the long exposure / short exposure image acquisition circuit into a luminance and a color difference signal; A luminance signal combining circuit for receiving long and short exposure luminance signals from the luminance / color difference signal conversion circuit and compressing the luminance by a weighted average technique; A luminance signal post-processing circuit for correcting a signal attenuated in the luminance compression process; A color difference signal combining circuit which receives color difference signals of long exposure and short exposure from the luminance / color difference signal conversion circuit and combines the color difference signals by a weighted average method according to the intensity of saturation; And an RGB inverse conversion circuit for synthesizing the output luminance signal of the luminance signal post-processing circuit and the color difference signal of the chrominance signal combining circuit.

The luminance signal post-processing circuit of the present invention includes histogram stretching means for stretching a combined luminance signal of two images; Gamma adjusting means for performing gamma conversion only on a region having a low gradation value as a result of the histogram stretching; And edge emphasis processing means for enhancing the weakened edge of the gamma-adjusted luminance signal.

In the method for realizing a wide dynamic range, the long exposure image I _LE photographed with a long exposure time using the imaging devices CCD and CIS and the short exposure image I _SE photographed with a short exposure time Combining R-1 to obtain an image having an extended dynamic range; An R-2 step of converting the acquired two images into a luminance Y signal and a color difference signal (RY, GY, BY); R-3 step of combining the color difference signals by weighted averaging the two images for each color difference signal (RY, GY, BY) pixel by pixel (Pixel by Pixel); An R-4 step of compressing the luminance (Y) signal by combining the luminance of two images by weight averaging; R-5 stretching the histogram to correct the luminance attenuated during the luminance compression process; R-6 performing gamma conversion to compensate for insufficient luminance compression due to the weighted average combination and to increase visibility of dark areas; An edge enhancement process R-7 for compensating for the sharpness of the image which is blurred due to the luminance process; And an R-8 step of synthesizing the color difference signal combined image and the edge-enhanced luminance signal to provide an RGB image having an extended dynamic range.

In the R-3 step of combining the color difference signal of the present invention, the weight used in the weight averaging formula is a scalar value obtained by the simplified standard deviation of the RGB 3 component.

In the R-6 step of gamma-converting the luminance signal of the present invention, the gamma conversion is characterized in that the conversion is performed only for a desired section.

As described above, according to the present invention, color saturation signals are classified in two images having different exposure times and weighted and averaged by the magnitude of saturation, thereby maintaining the saturation of the image with the extended dynamic range.

In addition, by classifying the luminance signals in two images having different exposure times, combining the luminance signals, and stretching the histogram, it is possible to correct the attenuated signal during the luminance combining process to improve the weakening of the contrast.

In addition, by gamma-converting only a specific region having a low gradation value as a result of histogram stretching, an image having high visibility and an appropriate contrast effect for the entire image may be provided.

In addition, by performing edge enhancement after gamma conversion, the edge component reduced in the luminance process may be restored to increase the sharpness of the image.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram of a color image processing apparatus according to the present invention.

The color image processing apparatus of the present invention includes a long exposure / short exposure image acquisition circuit 11 for acquiring long and short exposure image signals from an image pickup device; A luminance / color difference signal conversion circuit (12) for separating the output image signal of the long exposure / short exposure image acquisition circuit (11) into luminance and color difference signals; A luminance signal combining circuit 13 which receives the luminance signals of long exposure and short exposure from the luminance / color difference signal conversion circuit 12 and compresses the luminance by a weighted average technique; A luminance signal post-processing circuit (14) for correcting the signal attenuated in the luminance compression process; A color difference signal combining circuit 15 which receives the color difference signals of long exposure and short exposure from the luminance / color difference signal conversion circuit 12 and combines the saturation by a weighted average method according to the intensity of the color difference signal; And an RGB inverse conversion circuit 16 for combining the output luminance signal of the luminance signal post-processing circuit and the color difference signal of the color difference signal combining circuit.

The luminance signal post-processing circuit 14 includes histogram stretching means for stretching the combined luminance signal of two images; Gamma adjusting means for performing gamma conversion only on a region having a low gradation value as a result of the histogram stretching; And edge enhancement processing means for enhancing the weakened edge of the gamma-adjusted luminance signal.

In the chrominance signal combining circuit 15, the weight value is expressed as a scalar value of saturation, but preferably obtained as a simplified standard deviation of the RGB 3 component.

As described above, in contrast to simply combining two images captured in the related art, the configuration of the present invention classifies and corrects the luminance signal and the chroma signal of the two images separately, and then synthesizes the two signals to maintain the sharpness of saturation well. By compensating for the attenuated signal in the luminance compression process, an image having a wide dynamic range with excellent image quality can be provided.

In addition, the luminance signal post-processing circuit stretches the combined luminance signal of the two images and thus gamma-converts only the region having a low gradation value, thereby increasing the visibility of the region having the low signal value and at the same time expecting an appropriate contrast effect for the entire image. Finally, by applying edge emphasis to the luminance signal, a strong sharpening effect can be expected by canceling side effects of edge weakening, which is inevitably involved in the image combining process.

Next, how the above-described configuration of the present invention operates organically will be described in detail with reference to FIG. 4.

4 is an image processing flowchart of a color image processing apparatus according to the present invention.

First, a long exposure image I _LE photographed with a long exposure time using image pickup devices CCD and CIS is combined with a short exposure image I _SE photographed with a short exposure time to obtain an image with extended dynamic range. Acquisition (R-1)

Since image combining is a process of compressing a wideband signal into a narrow area signal, it involves non-linear processing, so when executed in the RGB area, color distortion results in different compression ratios. Therefore, first, the RGB color signal conversion operation for converting the luminance (Y) signal and the color difference signal (R-Y, G-Y, B-Y) is performed.

Here, for the color difference signal, the two images are pixel-by-pixel (Pixel by Pixel) weighted averaging for the color difference to combine the color difference signals (step R-3).

In addition, the luminance is compressed by combining the luminance of the two images through a weighted averaging operation (step R-4), and then performing histogram stretching (step R-5) to correct the attenuated luminance. Performs a gamma conversion (R-6) to compensate for the lack of luminance compression while increasing the visibility of dark areas, and an edge enhancement process (R-7) to compensate for the sharpness of the image that has been lost due to the previous luminance processing. Perform

Finally, the combined color difference signal and the luminance signal of the two images are synthesized to provide an RGB image having an extended dynamic range.

In the above process, the color difference signal conversion of the RGB signal (step R-2) is obtained using Equations 1 and 2 below.

RY _i = R _i Y _i (Equation 1)

GY _i = G _i -Y _i

BY _i = B _i -Y _i

Y _i = (R _i + (G _i << 1) + B _i ) >> 2 (Equation 2)

In Equations 1 and 2, i = 1, 2, 1 corresponds to I _{LE (x, y)} and 2 corresponds to I _{SE (x, y)} .

In fact, according to the BT.601 standard, the formula for obtaining luminance (Y) is Y = 0.299 · R + 0.587 · G + 0.114 · B, but simply use Y = (R + G + B) / 3 to obtain luminance in grayscale. Can be. However, from the hardware implementation point of view, the former requires an integer operation, three multipliers and two adders, and the latter is not two adders and 2 ⁿ divisions. The two methods use hardware logic unnecessarily even though they are simply operations. Therefore, based on Equation 2, only two adders and a shift operation can obtain a luminance value that is almost similar to the former formula.

On the other hand, the most ideal method for combining images to extend the dynamic range is to properly compensate for the areas without image information due to poor exposure in long / short exposure images. That is, an ideal combination method is to supplement the dark region in the short exposure image with the long exposure image and to fill the saturated region of the long exposure image with the short exposure image with the short exposure image. In fact, these areas without image information are almost achromatic and easily identifiable. Therefore, the coupling ratio for this area should be reduced as much as possible, and information of other saturated images (long exposure or short exposure) should be used. To this end, in the present invention, color difference signal combining (step R-3) is performed.

In the present invention, the color difference signal combination (step R-3) follows Equations 4, 5, and 6.

The high luminance image may be seen in a brightly saturated region, and the low luminance image may have little color in a dark region. Therefore, image combining using a simple average over this region is very likely to reduce color saturation. In order to compensate for this, if the color is emphasized over the entire area, it may cause a problem that the normally acquired color difference information is excessively emphasized.

In order to solve the above problem, the present invention performs weighted averaging on the color difference by pixel by pixel. Equation 4 represents the weighted average combination of the color difference signals obtained in Equation 2, and each weight alpha (α) is obtained by Equation 5.

RY _{F = RY 1 · α + RY 2} · (1-α) ( Equation 4)

GY _F = GY ₁ · α + GY ₂ · (1-α)

_{BY F = BY 1 · α +} BY 2 · (1-α)

α = S ₁ / (S ₁ + S ₂ ), if (S ₁ + S ₂ ) = 0 then S ₁ = S ₂ = 0.5 (Equation 5)

The weight in Equation 5 is obtained by the intensity of the co-saturation of the long exposure and the short exposure. That is, S ₁ becomes the color intensity of the long exposure pixel and S ₂ becomes the color intensity of the short exposure pixel at the same position.

Here, the color intensity is conventionally obtained by the formula (7).

if max (R, G, B) = 0 then S _sat = 0 (Equation 7)

However, when saturation (S: Saturation) of the HSI coordinate system is used as the intensity of the color as shown in Equation 7, there is a problem that high saturation can be expressed by very small RGB scalar values.

For example, if R = 2, G = 0, B = 0, it is close to achromatic color with black color and almost no saturation. In this case, the intensity of color becomes 100% when applied to Equation 7. .

Therefore, in the present invention, to obtain the color intensity by applying the equation (6). This is a simplified standard deviation of the RGB 3 components, which makes it possible to obtain stable weights by expressing the magnitude of saturation as a scalar.

S ₁ = │Avg ₁ -R ₁ │ + │Avg ₁ -G ₁ │ + │Avg ₁ -B ₁ │, Avg _i = (R _i + G _i + B _i ) / 3 (Equation 6)

S ₂ = │Avg ₂ -R ₂ │ + │Avg ₂ -G ₂ │ + │Avg ₂ -B ₂ │

However, since each Avg in Equation 6 can be almost identical to the luminance value of one pixel composed of RGB, if Avg is replaced with Equation 2, the weight S can be expressed as the absolute sum of the color difference signals obtained in Equation 1. Therefore, the hardware implementation based on Equation 6-1 below has the advantage of not using a divider.

S ₁ = | R ₁ -Y | + | G ₁ -Y | + | B ₁ -Y | (Equation 6-1)

S ₂ = R ₂ -Y | + | G ₂ -Y | + | B ₂ -Y |

Comparing Equations 6-1 and 7 above, Equation 7 calculates saturation through the relative magnitudes of the three RGB values. In this case, relative means that the magnitude of saturation is normalized by the largest value among RGB. On the other hand, Equation 6-1 calculates achromatic RGB (2,0,0) with low saturation, which is almost black as in the previous example, by absolute comparison of RGB values.

That is, if the pixel value determined to be dark or achromatic with the human eye is used as a wrong result of high saturation by Equation 7, it causes a problem in color reproduction of the combined image. In HSI coordinate systems, in particular, the lower the I (intensity or lightness), the higher the probability that the normalized saturation value will produce the unwanted results described above. This is because the lower I is in the HSI coordinate system of FIG. 5, the easier the S is to 100%.

5 is a diagram illustrating a general HSI coordinate system.

In the HSI coordinate system, the cone is upside down and the I (Intensity or Lightness) increases from bottom to top, and the bottom of the cone shows the color (H: Hue) and varies with angle. In other words, as I moves away from the cone, the saturation (S) increases and as I decreases, S becomes 100% easily.

In the above process, luminance combining (step R-4) uses the long exposure and short exposure luminance values obtained using Equation 3 below and performs weighted average combining based on Equation 8.

Y _combining = Y ₁ · β + Y ₂ · (1-β), if 0≤β≤1 (Equation 8)

In general, assuming that an appropriately exposed image is acquired by an appropriate exposure time, luminance combination, that is, luminance compression, of a long / short exposure image may be uniformly expressed throughout the two images through an average combination. In this case, the weighted beta value is applied as 0.5 in Equation 8, but in this case, the average luminance combination can completely reproduce the contrast of the two images, but the contrast is weakened, so the visibility is reduced. In the present invention, to solve this problem, the histogram stretching (step R-5) is performed.

FIG. 6 is a photograph of a low light image (a) and a high light image (b) and an image (c) obtained by averaging two images in a conventional brightness combining step, and FIG. 7 is a histogram (a) and high tone of the low light image of FIG. FIG. 8 is a diagram illustrating a histogram (b) of an image and a histogram (c) of an average combined image of two images. FIG. 8 is a diagram illustrating a histogram after a histogram stretching process of a luminance-coupled image.

In the related art, since the average image is simply combined with each other, the histogram a of the low light image is shifted up as shown in FIG. That is, as well as a weakening of the contrast of the average combined image, the luminance value of the full range (0 ~ 255) can not be used.

To improve this contrast weakening is to perform histogram stretching (R-5 step).

In fact, the histogram after the histogram stretching process is as shown in FIG. 8, and when compared with FIG. 7C, the graph is stretched to the left as a whole to confirm that the luminance value is entirely extended in the 0 to 255 section.

Here, the histogram stretching is based on Equation 9.

(식 9)

(Eq. 9)

On the other hand, as a result of the histogram stretching (step R-5), when a pixel having a low gradation value is moved to the left on the histogram, the visibility of a region having a low gradation value (section P in FIG. 9) may be degraded. In the present invention, a gamma conversion (step R-6) for adjusting gamma is performed as a method for widening the dynamic range of this portion.

9 is a histogram after stretching the histogram of the present invention, and FIG. 10 is a graph of the gamma signal conversion form when the general gamma conversion basic formula is applied, and FIG. 11 is a graph of the gamma signal conversion form when the gamma conversion formula according to the present invention is applied. 12 is a photograph of a histogram stretching image (a), a global gamma transform image (b), and a partial gamma transform image (c) according to the present invention.

In general, the basic formula for gamma conversion is to obtain an output signal by using gamma as the square root of the input signal (x) as shown in Equation 10 (where 0≤x≤1 and 0≤y≤1).

(식 10)

(Eq. 10)

However, in the case of a video signal having a value from 0 to 255, the gamma conversion formula is modified and used as in Equation 11.

(식 11)

(Eq. 11)

In FIG. 9, the gamma value may be lowered to less than 1 to increase the signal of the P section. At that time, the gamma signal shape is the same as the green solid line of FIG. 10, so that the signal of the P section can be increased, but the globally increased signal value is obtained for the remaining sections other than the P section due to global processing for 0 to 255. This is an area that can be judged as properly combined in the luminance-combined image is also signal processing, which is a side effect of reducing the contrast when viewed throughout the image.

Therefore, in the present invention, gamma conversion is performed only on a desired P section using Equation 12 to increase visibility of a region having a low signal value.

(식 12)

(Eq. 12)

By setting the Gamma Max Point (GMP) in Equation 12 and performing gamma conversion only for signals below it, gamma conversion is performed only for the P section with low gradation values, and the input signal is applied for the remaining sections with high gradation values. As shown in FIG. 11, as shown in FIG. 11, the visibility of the region having a low signal value is increased and an appropriate contrast effect can be expected for the entire image.

Comparing the histogram stretching result image (a) with the image (b) after the global gamma conversion processing with reference to FIG. 12, it can be seen that the contrast of the image is rather deteriorated after the global gamma conversion processing. It can be seen that the contrast between the low visibility area and other areas is stable and superior to the above two images.

On the other hand, since image combining for dynamic range expansion is an operation of compressing a wide range of gray scale data into a narrow region, it is inevitably accompanied by side effects of making a large gray scale change into a small gray scale change. This is indicated by edge weakening. Therefore, after the gradation compression, the side effect can be canceled out through the edge emphasis.

Here, edge enhancement compensates for the edge attenuation effect through a mask operation such as equations (13) and (14).

Y '(i, j) = (9 * Y _gamma (i, j) -Y _gamma (i-1, j) -Y _gamma (i + 1, j) -Y _gamma (i, j-1)-

Y _gamma (i, j + 1)} / 5 (Equation 13)

Y '(i, j) = 5 * Y _gamma (i, j) -Y _gamma (i-1, j) -Y _gamma (i + 1, j) -Y _gamma (i, j-1)-

Y _gamma (i, j + 1) (Equation 14)

Here, i and j are coordinate data of the x-axis and y-axis, respectively, and Y _gamma and Y 'mean gamma-adjusted image and output image, respectively. Equation 14 has a stronger edge than Equation 13, and can be selected and used for a stronger sharpening effect.

Finally, after all processing for the chrominance and luminance signal is finished, the RY _F , GY _F , and BY _F obtained in Equation 4 are restored to RGB. The RGB restoration is based on Equation 15 in which Y 'processed up to the edge emphasis is added to the color difference signals RY _F , GY _F and BY _F.

R _HDR = RY _F + Y '(Equation 15)

G _HDR = GY _F + Y '

B _HDR = BY _F + Y '

1 is a block diagram of a conventional camera video signal processing apparatus ①.

2 is a block diagram of a conventional television camera and television signal processing method ②.

Figure 3 is a block diagram of a color image processing apparatus according to the present invention

4 is an image processing flowchart of a color image processing apparatus according to the present invention.

5 is a general HSI coordinate system

FIG. 6 is a photograph of a low light image (a) and a high light image (b) and an image of average image combining two images in a conventional brightness combining step.

FIG. 7 is a histogram (a) of the low light image of FIG. 6, a histogram (b) of a high light image, and a histogram (c) of an average combined image of two images.

8 is a histogram after the histogram stretching process according to the present invention

9 is a histogram after histogram stretching of the present invention,

10 is a graph of gamma signal conversion when the general gamma conversion basic formula is applied.

11 is a gamma signal conversion graph when the gamma conversion formula according to the present invention is applied.

12 is a photograph of a histogram stretching image (a), a global gamma transform image (b), and a partial gamma transform image (c) according to the present invention.

Claims (5)

delete A long exposure / short exposure image acquisition circuit for acquiring long and short exposure image signals from an image pickup device; A luminance / color difference signal conversion circuit for separating the output image signal of the long exposure / short exposure image acquisition circuit into a luminance and a color difference signal; A luminance signal combining circuit for receiving long and short exposure luminance signals from the luminance / color difference signal conversion circuit and compressing the luminance by a weighted average technique; A luminance signal post-processing circuit for correcting a signal attenuated in the luminance compression process; A color difference signal combining circuit which receives color difference signals of long exposure and short exposure from the luminance / color difference signal conversion circuit and combines the color difference signals by a weighted average method according to the intensity of saturation; And an RGB inverse conversion circuit for synthesizing the output luminance signal of the luminance signal post-processing circuit and the color difference signal of the color difference signal combining circuit. The luminance signal post-processing circuit includes: histogram stretching means for stretching the combined luminance signal of the two images; Gamma adjusting means for performing gamma conversion only on a region having a low gray level as a result of the histogram stretching; And edge enhancement processing means for enhancing the weakened edge of the gamma-adjusted luminance signal. In a method for realizing a wide dynamic range, Long-exposure image I _LE photographed with long exposure time using image pickup devices CCD and CIS is combined with short-exposure image I _SE photographed with short exposure time to obtain an image with extended dynamic range. R-1 step; An R-2 step of converting the acquired two images into a luminance Y signal and a color difference signal (RY, GY, BY); R-3 step of combining the color difference signals by weighted averaging the two images for each color difference signal (RY, GY, BY) pixel by pixel (Pixel by Pixel); An R-4 step of combining and compressing the luminance of the two images through a weighted averaging operation on the luminance (Y) signal; R-5 stretching the histogram to correct the luminance attenuated during the luminance compression process; R-6 performing gamma conversion to compensate for insufficient luminance compression due to the weighted average combination and to increase visibility of dark areas; An edge enhancement process R-7 for compensating for the sharpness of the image which is blurred due to the luminance process; And an R-8 step of synthesizing the chrominance combined image and the edge-weighted luminance signal to provide an RGB image having an extended dynamic range. The method of claim 3, The R-3 step of combining the color difference signal is obtained by weighted averaging for the color difference by pixel by pixel, and the weight α is expressed by Equations 5 and 6-1 below. α = S ₁ / (S ₁ + S ₂ ), if (S ₁ + S ₂ ) = 0 then S ₁ = S ₂ = 0.5 (Equation 5) S ₁ = │R ₁ -Y | + | G ₁ -Y | + | B ₁ -Y | (Equation 6-1) S ₂ = R ₂ -Y | + | G ₂ -Y | + | B ₂ -Y | Acquisition by, wherein S ₁ is the color intensity of the long exposure pixel, S ₂ is the color intensity of the short exposure pixel of the same position. The method of claim 3, In the R-6 step of gamma-converting the histogram stretching result image, Gamma Max Point (GMP) is set.

(Eq. 12) A gamma conversion is performed only on a desired P section (when a pixel having a low gradation value is shifted to the left on a histogram) using Equation 12.

Images