KR100965551B1 - Method and apparatus for compression and decompression of lattice arrayed images - Google Patents

Abstract

PURPOSE: A virtual pixel value estimating method and an image compression / restoring method using the same are provided to offer image quality of various quality by applying various image processing methods in improvement of resolution using a restored reference image and an additional image. CONSTITUTION: An image defining unit(701) defines a lattice arrayed image(401) as a reference image and an additional image. An empty pixel prediction unit(702) predicts an empty pixel value in the additional image. An image converter(302) reconstructs the reference image and pixel values of an empty pixel-predicted signal as one image frame. A coding unit(305) performs quantization processing or allocation of a binary code about a converted signal.

Description

Method for estimating virtual value and image compression and decompression using same method and apparatus {Method and apparatus for compression and decompression of lattice arrayed images}

The present invention relates to a method for compressing and reconstructing an image. In particular, the present invention relates to a method of compressing and restoring an image, including an additional image arranged in a grid with the reference image to improve the number of samples of the reference image. A method of estimating an empty pixel value of an image and an image compression and reconstruction method using the same are provided.

The super-resolution technology, when photographing the same subject and securing a plurality of image information having different shooting angles, has an effect of improving the number of samples of the image information, and ultimately through image processing techniques such as matching, noise reduction, and interpolation. Using the principle that high resolution can be overcome to overcome the limitations of the present invention, a technique of implementing super resolution from images having various characteristics such as moving images, stereoscopic images, and satellite images has been introduced.

In particular, in the field of aviation and satellite imagery, the demand for high resolution images continues to increase as commercialization of observational images proceeds.Increasing the size of imaging equipment including lenses, reflectors, etc. Although lowering is one method of improving the resolution of the observed image, in the former case, an increase in the size of the imaging device is inevitable, and a method of lowering the operating altitude also inevitably shortens the satellite life and decreases the observation width.

Therefore, as shown in FIG. 1A, the array of image detection elements such as CCD and CMOS (101) is long (102) in the direction in which the aircraft and the satellite move (106), and T2, T4, and T6. As shown in FIG. 1B, an image device of an aircraft and satellite field is characterized by obtaining a set of (102A, 102B, 103C) of linearly arranged pixels by controlling an acquisition point of an image as shown in FIG. It is common to use a 'grid array linear image detector' (108), which is characterized by arranging the pairs (102) to (104) half of the inter-pixel spacing (108), which uses the lattice linear image detector. By specifying the (102 to 103) pixel output (103-A, 102-A, 103-B, 102-B, 103-C, 102-C) times of the linear image detection device (T1, T2, T3, T4) , T5, T6) two sets of images, i.e., the number of samples that increases the number of samples of the reference image and (107) the reference image. It is possible to acquire an image (105). That is, like a camera mounted on an aircraft and a satellite, a linear image detection device in which 104 half of the interval between pixels is arranged at right angles to the camera's moving direction (106) at the right angle (102, 103) has 103 arrays operated at the time T1. To obtain one linear pixel (103-A), and at time T2, array 102 operates to obtain another linear pixel (102-A), and at time T3, array 103 to operate to obtain another linear pixel. (103-B), at time point T4, 102 arrays operate to obtain another linear pixel (102-B), and at time T5, 103 arrays operate to obtain another one linear pixel (103-C). At the time T6, 102 arrays operate to acquire another set of linear pixels (102-C), and through this series of function iterations, we acquire two sets of images (105, 107) of different locations. As shown in 1c, zeros are obtained using a set of linear image detectors. A set of comparison because it is effective to increase sample size is twice the acquired image for the same area of the subject, by applying the aforementioned image processing technique is ready to improve the resolution of the image. Meanwhile, the reference image and the additional image are relative to each other. For example, when the reference image is defined as the image set 105, the image set 107 may be considered to contribute to an increase in the number of samples of the image set 105. Of course, it can be defined as an additional video.

On the other hand, in relation to a method for obtaining a grid array image, as well as the grid array linear image detection device 108, Masaru Osada et al., According to US Patent 6,847,397 B1, characterized in that a plurality of image detection elements arranged in a grid array The method of using the planar image detection device with this is also well known in the art. Fig. 2 shows a conventional 'plane image detection device' and 200-A, which are arranged in a row and column direction with N and M image detection elements linearly on a two-dimensional plane, 201 A, and additionally in a lattice form. By explaining the (200-B) structure of the 'lattice array planar image detection device' characterized by arranging the image detection elements 202 together, and the concept of implementing super resolution from the grid array image obtained therefrom, Comparing the distance between the image detection elements included in the array-type image detection device (200-B), when the image detection elements are arranged in a lattice shape, the interval between the pixels in the diagonal direction is increased due to the increase in the number of elements per unit area. A to 205-B) shows that there is a feature that becomes smaller compared to the spacing between the pixels in the row 203 and the column direction 204, which can be understood as the effect of increasing the number of images per unit area, The lattice-array image detection device 108 to lattice-array image detection device 200- than the non-lattice-type image obtained by using the conventional non-lattice linear 102 to the conventional non-lattice planar image detection device 200-A. B) This is also one reason that the correlation between pixels is improved in a grid array image obtained by using such a method. That is, as shown in (c) of FIG. 1, when the acquisition positions of pixels included in two different image sets (105 to 107) are found and aligned in one image frame, a well-known image processing method such as interpolation and noise filtration is applied. Due to the increasing effect of the image sample, it is possible to estimate the (206) pixel value of the 'empty pixel' region without the pixel value more accurately in the case of non-lattice general image, which is a conventional non-lattice image set (105, 107). To 200-A) is a method of overcoming the limitation of resolution, and is well known as an example of a super resolution implementation technique. However, the grid array image is inevitable to increase the amount of image to be stored and transmitted than the non-lattice type normal image, it is necessary to compress the video signal according to the storage and transmission capacity.

3 is an example of a concept of image compression, which is widely known in the art, and calculates a (301) pixel value of an input image signal and expresses the information of the image signal as a signal different from the conventional pixel value. (301) compression of the video signal through step (302) and (304) an encoding step (305) of assigning a specific code to a result of calculating the entropy of the pixel value or the 303 quantized image converted signal. It is characterized in that it consists of procedures for generating a code string (306), and the image conversion uses (302) a well-known feature of an image signal having a high correlation between adjacent horizontal and vertical pixel values, and is used in JPEG to MPEG and the like. Like the discrete cosine transform (DCT) method and the wavelet image conversion method, a 'transition region' exists in a region range of 'horizontal 8 pixels x vertical 8 pixels' or a specific two-dimensional color space pixel region. Is a method for calculating the horizontal and the vertical correlation between pixels, represented in the frequency domain to the pixel domain naejineun the pixel value of the zone frequencies. The encoding (305) includes a procedure of allocating binary codes having different lengths according to the probability of occurrence of each signal by calculating the entropy of the signal, as in Run-length, Huffman, Lempel-ziv, etc. In this case, the entropy of the pixel value is calculated directly and the specific binary code is assigned without performing the image conversion.

In the case of a lattice array image, as in the example of FIG. 2, in addition to the correlation between the pixel values in the vertical direction 204 and the horizontal direction 203 in the image conversion step, correlations between different diagonal directions (205-A to 205-B) are calculated. Since this operation is difficult to perform the image conversion, it is not effective to apply the conventional compression method to compress. Therefore, the grid array image is divided into a base image 107 which is a non-grid type image as illustrated in FIG. 1, and an additional image 105 arranged in a grid and having an increase in the number of samples of the image. Although a method of independently compressing the reference video and the additional video using the conventional compression method as illustrated in FIG. 3 may be considered, in this case, the size of the compression device is increased due to the need to implement a function of amplifying separately. There is a disadvantage.

Thus, Quincunx discrete cosine (DCT) conversion method has been introduced, Quincunx discrete cosine conversion method, as shown in Figure 4, the grid to match each pixel (105 to 107) of the two sets of images, corresponding to the respective image acquisition position In the image conversion step, the pixel region to be discrete cosine transformed is set in a rhombus shape as shown in FIG. 4 to improve the correlation between the pixels to be converted (402). In the case of some pixels in which pixels are located at the beginning or end of an image column or row and are difficult to be included in a rhombus-shaped 'conversion area', the image conversion becomes difficult (403). Meanwhile, according to US Patent No. 6,847,397 B1 of Masaru Osada et al., (Binary pixel) calculated by applying a grid array image obtained through a grid array planar image detection device (200-B) and an interpolation technique, etc. 206) After adding the values to improve the resolution of the image, the compressed image is compressed.In this case, the high-speed corresponding to the increase in the amount of information due to the addition of the empty pixel operation procedure and the increase in the amount of information of the image to be compressed is achieved. There is a disadvantage that a compression device must be provided.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and the method of estimating the virtual pixel value of the reference image using consecutive adjacent pixel value information of the additional image and the method of compressing and restoring the image using the same and at a simple and low cost. It is an object to provide a compression method and apparatus.

In order to achieve the above object, the present invention includes an additional image arranged in a grid with a reference image to obtain an effect of improving the number of samples of the reference image. A method of estimating a pixel value, wherein when there is one adjacent pixel in one column or row of one pixel row of the additional image and the column or row is adjacent, the adjacent pixel value is converted into a virtual value. And a virtual value estimation method for calculating the average value of the adjacent pixel values as a virtual value when two adjacent pixels are used.

In addition, an image defining step of reading a pixel value of the grid array image signal and defining it as an additional image contributing to the improvement of the number of image samples of the reference image, the virtual value estimation step described above, and converting each pixel value into color space data On the other hand, there is another feature of an image compression method using a virtual value estimation method including an image conversion step of transforming an entire image into a frequency space and a quantization step of integerizing each component of the frequency domain.

In addition, a method of reconstructing the original pixel value of the reference image by the empty pixel estimation method using the pixel value and the empty pixel value of the additional image generated by inverse image conversion of the grid array compressed video signal, wherein one pixel column of the additional image If there is only one adjacent pixel used for estimating the virtual pixel value corresponding to the first or last empty pixel of one to one pixel row, the corresponding pixel value is calculated as the virtual pixel value, and two consecutive bins are used. If the original pixel is located between the pixels, the corresponding pixel value is reconstructed by using the virtual pixel value estimation method calculated by Equation 16 using two successive empty pixel values (a, b). There is another feature.

According to the present invention, by compressing a grid array image by using virtual image values and a reference image computed from an additional image, each conventional image set is independently compressed, or a Quincunx discrete cosine method or a 'blank pixel' estimation procedure is performed. Overcoming shortcomings such as the increase in the size of the compression device, the incompleteness of the conversion area, and the increase of the calculation amount by applying the compression method of Masaru Osada, etc., as well as the well-known conventional compression and compression restoration methods such as JPEG and MPEG This is possible. In addition, the restoration of the compressed and compressed image according to the present invention is that, as described in the virtual pixel and the original pixel estimation procedure, the original value of the pixel at a specific position is the estimated value of the virtual pixel or the original pixel as the first to the last pixel value of the additional image. It is not only a means for effectively compressing a grid array image, but also restores the original grid array image from the grid array compressed image by simple calculation. This has a possible advantage. In particular, Masaru Osada et al. (US Pat. No. 6,847,397 B1) compresses the resolution of a grid array image in advance to improve the resolution and the amount of computation during compression. In the case of the grid array image restored by the present invention, the reference image and the additional image can be faithfully restored in the compression restoring step. It is possible to apply a variety of image processing techniques in improving the resolution, thereby providing a variety of image quality.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings and examples.

 The present invention relates to a virtual value estimation method, an image compression and restoration method and apparatus using the same as described above. First, the virtual value estimation method will be described with reference to the first embodiment.

Example 1

The virtual value estimating method of the present invention includes an additional image arranged in a grid with a reference image to improve the number of samples of the reference image, and the bin of the reference image using consecutive adjacent pixel value information of the additional image. As a method of estimating a pixel value, when the pixel corresponds to the first or last pixel of one pixel column to one pixel row of an additional image and there is only one adjacent pixel in the column or row, the adjacent pixel value is converted into a virtual value. In the case of two adjacent pixels, the average value of the adjacent pixel values is calculated as a virtual pixel value.

On the other hand, with reference to Figures 1 and 5 the embodiment of the virtualization estimation method of the present invention in detail,

A first step of setting a reference image (107) and an additional image (105) causing an effect of improving the number of samples of the reference image from the output image of the grid array image to the grid array image obtaining apparatus;

In the additional image of the first step, when the acquisition position of the pixels of the reference image and the additional image is found and aligned to one image frame (Fig. 1 (c)), a blank pixel having no actual pixel value between pixel values is obtained. (206) is present, and the virtual pixels included in one of the pixel rows or the pixel columns in the reference image are defined as 'virtual pixels' of the present invention, and (501, 502, 503) ) If there is only one pixel of the adjacent additional video (105-C), the second step of taking the pixel value of the (501 to 502) neighboring original additional video (501) as the value of the (105-B) virtual,

In case that there are two pixels of the virtual source to estimate the value of (503) and the adjacent additional video (105-A, 105-B), the virtual value estimated in the previous step (501) and the virtual step in the previous step (501) A third step of estimating the average of the pixel values of the unused additional video (105-A) and calculating the average thereof as the value of the virtual server 503,

And estimating values of a virtual virtual machine, including a fourth step of repeatedly performing the above steps with respect to the additional video.

As an example, a virtual image exists between a pixel column (or row) of an additional image having A, B, C, D, and E values and arranged in the order of A, B, C, D, and E, and the reference image 'pixel row'. Referring to the procedure of estimating the pixel, the virtual pixels a, b, c, d, and e estimated from the pixel values of the additional video are a, A, b, B, c, C, d together with the pixels of the additional video. According to the method for estimating a virtual pixel value according to the present invention, the value of 'a' is the first virtual pixel having only one pixel of an additional image adjacent to the virtual pixel. Is calculated by Equation 1,

[Equation 1]

a = A

It can be estimated by the relation of, and the next virtual 'b' can be calculated by Equation 2.

[Equation 2]

b = (a + B) ÷ 2

Likewise, it is estimated that the average value of the virtual value 'a' estimated in the previous step and the pixel value 'B' of the next additional image of the virtual image 'b' to be estimated are calculated. It can be calculated by equation 3,

&Quot; (3) "

c = (b + C) ÷ 2

Likewise, it is assumed that the average value of the virtual value 'b' estimated in the previous step and the pixel value 'C' of the next additional image of the virtual image 'c' to be estimated are calculated. 4 can be calculated in the same way as the previous method,

&Quot; (4) "

d = (c + D) ÷ 2

Likewise, it is estimated that the average value of the virtual value 'c' estimated in the previous step and the pixel value 'D' of the next additional image of the virtual image 'd' to be estimated are calculated. By 5,

[Equation 5]

e = (d + E) ÷ 2

Likewise, it is estimated that the average value of the virtual value 'd' estimated in the previous step and the pixel value 'E' of the next additional image of the virtual image 'e' to be estimated are calculated. It is an example of an effective calculation method when the pixels (A, B, C, D, E) need to be processed sequentially, and it is advantageous to simultaneously read and process the above-described pixels as in the planar image detection device shown in FIG. In this case, except for the virtual 'a', which should be calculated by Equation 1, the remaining virtuals can be estimated to calculate a simple average of pixel values of two additional images adjacent to the virtual, and the virtual 'b' For example, Equation 2 for estimating 'can be calculated as' b = (A + B) ÷ 2' in this case. The virtual sources 'c' and 'd' to 'e' estimated by Equations 3 to 5 may also be estimated by calculating an average of pixel values of two additional images adjacent to the virtual source.

Example 2

The present invention to be described in this embodiment is an image compression method (S100) using the above-described virtual value estimation method.

That is, the compression method (S100) includes an image definition step (S110) of defining an additional image contributing to the improvement of the number of sample images of the reference image by reading pixel values of the grid array image signal, and one pixel column to one pixel row of the additional image. If there is one adjacent pixel in the column or row corresponding to the first or last pixel of the pixel, the adjacent pixel value is calculated as a virtual pixel value, and when there are two adjacent pixels, the average value of the adjacent pixel value is calculated as the virtual pixel value. A virtual value estimating step (S120) for calculating a value, an image conversion step (S130) for converting each pixel value into color space data, and converting an entire image into a frequency space, and integerizing each component of the frequency domain. A quantization step S140 is included.

The image defining step S110 is a step of defining a reference image and an additional image contributing to the improvement of the number of image samples of the reference image by reading pixel values of the grid array image signal as described above.

The virtual value estimating step (S120) corresponds to the first or last pixel of one pixel column to one pixel row of an additional image, and when there is only one adjacent pixel in the column or row, the adjacent pixel value is virtualized. It calculates a value, and when there are two adjacent pixels, the average value of the adjacent pixel values is calculated as a virtual value, which has been described in Embodiment 1, and thus detailed description thereof will be omitted.

The image converting step (S130) is a step of converting each pixel value into color space data and converting an entire image into a frequency space.

In this case, the conversion step (S130) is a first conversion step (S131) for converting each pixel value into color space data, divided the entire image into sections having a size of 8x8 pixels, and then the data of each 8x8 section is two-dimensional A second transform step S132 of converting into a frequency space using a discrete cosine transform may be included.

Referring to the quantization step S140 of integerizing each component of the frequency domain, the human eye may distinguish small differences in a relatively wide region with respect to brightness. However, the change in the brightness of the high frequency is less distinguishable.

Based on this fact, much of the information of the high frequency component can be discarded. This is done by dividing by a certain constant and taking only integer quotients for each component in the frequency domain. This is the most costly part of the overall compression process. In this way, the high frequency components are almost zero, zero or near positive or negative.

According to the present invention (S100) as described above, the image can be compressed simply and easily.

Example 3

The present invention relates to an apparatus implemented by an electric machine and a computer program for converting an image including the empty pixel estimation method. The apparatus for compressing a grid array image, the pixel value of the grid array image signal being read, An image defining unit defined as an additional image contributing to the improvement of the sample number of the reference image, and one adjacent pixel in the column or row corresponding to the first or last pixel of one pixel column to one pixel row of the additional image And a virtual prediction unit for estimating the adjacent pixel value as a virtual pixel value and estimating an average value of the adjacent pixel values as a virtual pixel value when there are two adjacent pixels, a pixel value of the reference image and a value of the virtual pixel. Image frames in the color space according to the acquisition position, Includes the image conversion unit for converting the area to be represented in the mixed region, it said image conversion expression region encoding unit to quantize the transformed output image signal processing portion.

As shown in FIG. 7, the apparatus for compressing a grid array image includes an image defining unit 701 which defines an input grid array image 401 as a reference image 105 and an additional image 106, and the additional image. A bin pixel prediction unit 702 for predicting a bin pixel value by a bin pixel prediction method, and the reference image 105 reconstruct the pixel values of the bin pixel predicted signal into one image frame by estimating respective acquisition positions. After the image 703, the image conversion unit for calculating the correlation using the same (302)

Finally, in order to generate the compressed code sequence 704 of the grid array image, the transformed signal may be configured by an encoder 305 for assigning a quantized process or a binary code.

Example 4

 In the present embodiment, a method (S200) of reconstructing a compressed image using the above-described virtual value estimation method will be described.

That is, the method (S200) is a method of reconstructing the original pixel value of the reference image by the virtual pixel value estimation method using the pixel value and the empty pixel value of the additional video generated by inverse image transforming the grid array compressed video signal. And when there is only one adjacent pixel used for estimating the virtual pixel value corresponding to the first or last empty pixel of one pixel column to one pixel row of the additional image, the original pixel value is calculated as the virtual pixel value. and,

When the original pixel is located between two consecutive empty pixels, the corresponding original pixel value is calculated by Equation 16 using two consecutive empty pixel values (a, b).

[Equation 16]

Original pixel value = (2 × b)-a

As shown in FIG. 8, the method for estimating the original pixel according to the present invention will be described in detail. The method for setting the reference image 105 and the empty pixel image set estimated by the empty pixel estimation method of the present invention from the grid array compressed image Perform step 1 (S210).

Subsequently, in one pixel column to one pixel row of the empty pixel image set in the first step S210, as the first empty pixel 501 value estimated at the time of compression, the first original pixel to estimate the value is 104. -B) take a second step (S220),

If there are two or more empty pixel values estimated at the time of compression before the original pixel 104-A to be estimated (502 and 503), a third pixel for calculating the original pixel value using the two previous empty pixel values is calculated. Perform step S230,

It is characterized by consisting of a series of steps of estimating original pixel values by repeatedly performing the above steps on the image set estimated by the empty pixel estimation method of the present invention.

More specifically, the pixel values in one pixel column (or row) obtained by reverse image conversion of the compression code sequence of the grid array image in the second step S220 and the third step S230 of the method of reconstructing the grid array compressed image. Assuming that the values of a, b, c, d, and e are arranged in the same order and the estimated original pixel values are A, B, C, D, and E, respectively (504), According to the method of estimating the original pixel value, the value 'A' of the first original pixel having only one original pixel (A) of the adjacent pixel and (a) the neighboring subpixel can be calculated by Equation 6.

&Quot; (6) "

A = a

It can be estimated by the relation of, and the next pixel 'B' is expressed by Equation 7,

[Equation 7]

B = (2 × b)-a

For example, the two previous empty pixel values 'a' and 'b' are calculated, and the next original pixel 'C' is represented by Equation 8 in the same way as in Equation 7,

[Equation 8]

C = (2 × c)-b

For example, the two previous empty pixel values 'b' and 'c' are calculated, and the next original pixel 'D' is expressed by Equation 9,

[Equation 9]

D = (2 × d)-c

For example, the two previous empty pixel values 'c' and 'd' are calculated, and the next original pixel 'E' is also expressed by Equation 10,

[Equation 10]

E = (2 × e)-d

As shown above, the two empty pixel values 'd' and 'e' can be easily estimated. As illustrated above, the pixels are read sequentially (A, B, C, D, E). It is an example of an effective method for estimating an original pixel when it is calculated by Equations 1 to 5, and when the above-described pixels are read and processed at the same time, except for the original pixel 'A', which is unavoidable by Equation 6 For example, the remaining original pixels can be easily estimated by doubling the empty pixel value immediately before the original pixel to be estimated by subtracting an already known original pixel value. In this case, it can be calculated using the empty pixel 'b' and the original pixel value A previously known and estimated as 'B = (2 × b)-A'. The original pixels of Equations 8 to 10 may also be estimated by subtracting a known original pixel value by doubling the empty pixel value immediately before the corresponding pixel.

The variables disclosed in Equations 1 to 6, that is, A, B, C, D, and E, are preferably maintained to have a value corresponding to the original original pixel value. Of course, it can be changed to have a value added or multiplied.

Example 5

On the other hand, it is also possible to configure a method and apparatus for reconstructing the grid array compressed image to which the empty pixel estimation method is applied.

That is, in an apparatus for reconstructing the additional image and the reference image from a lattice-arranged compressed image signal composed of specific binary codes, the binary image is dequantized to decompress the compressed image signal into a frequency domain or a frequency domain and a spatial domain. In the mixed region or an inverse encoding unit for restoring the image frame by analyzing the binary codes, and the output signal of the inverse encoding unit is a video signal expressed in a frequency domain or a region in which the frequency domain and the spatial domain are mixed. An inverse image converter configured to calculate pixel values of one image frame including pixel values and virtual pixel values of the reference image from the decoded signal, and virtualization in which the pixel values of the image frame are estimated from the additional image. In the case of a small value, the pixel estimating unit which estimates the original pixel value by the above-described virtual pixel value estimation method. A lattice arrangement compressed image reconstruction device is also included.

In the present embodiment, as shown in FIG. 8, the compressed code sequence input unit 704 of the lattice array image, the decoded coder 801 which analyzes or dequantizes the quantized or binary processed compressed code sequence 704, An inverse image conversion unit 802 for calculating a pixel value from the inversely encoded signal and the inverse image transformed signal to which the empty pixel estimation method of the present invention is applied to determine whether an additional image requires an original pixel (807). In the case of the additional video, the processing unit 805 estimates the original pixel value by using the original pixel estimation method of the present invention (805). Electric machines and devices that are configured are also possible.

On the other hand, a method and apparatus for restoring a grid array compressed image to which the original pixel estimation method is applied may also be configured. As shown in FIG. 8, the compressed code sequence input unit 704 of the grid array image and the compressed code sequence as described above are used. An inverse encoding unit 801, characterized in that it directly interprets or inversely quantizes the quantized or binary processed compressed code string 704, and an inverse image converter that calculates pixel values in a color space from the inverse quantized signal. 802 and whether the inverse image transformed signal is the result of the additional video requiring the original pixel estimation as the virtual pixel estimation method of the present invention is applied (807) or the result of the reference video that is not (806). In the case of the result of the additional video including the judging procedure (803), in the case of the result of the original pixel estimation method of the present invention, the original pixel estimator 805 estimates the original pixel value (804). System and apparatus are possible.

1A to 1C are conceptual views of acquiring a super-resolution image set using a linear image detection device arranged in a grid in the related art.

2 is a conceptual diagram of implementing a super resolution using a planar image detection device arranged in a grid conventionally;

3 is an explanatory diagram of an image compression method including a conventional image conversion step;

4 is an explanatory diagram of a compression method of a conventional grid array image having an improved image conversion area (Quincunx method).

5 is an embodiment of a method for estimating a blank pixel of the present invention for estimating a blank pixel from an additional image of a grid array image;

FIG. 6 illustrates an embodiment of the present invention in which a conversion region is set from the bin pixel values estimated by the bin pixel estimation method of the present invention and the pixel values of the reference image estimated from each of the acquisition positions and displayed in one image frame. Yes

7 is an embodiment of the present invention for the compression method of the grid array image

8 illustrates an embodiment of a method for reconstructing a super-resolution lattice array compressed image.

<Explanation of symbols for the main parts of the drawings>

102 (103): A linear image detector and a simplified symbol of its concept for generating a meaningful sequence of pixel values at a particular point in time

206: A lattice-array image composed of pixels having meaningful values such as pixels 201 to 202, which defines an area that is regularly arranged with the pixels 201 to 202 without actual pixel values.

501 (or 502): As a part of the 'empty pixel', the 'virtual pixel' calculated by the virtual pixel value estimation method of the present invention, and an example of being arranged in a horizontal direction with the pixels of the base image.

703: A function performing unit for arranging virtual pixel values according to the present invention to one image frame in addition to the original desired position between the pixel rows or the pixel columns of the reference image.

Claims (6)

A method for estimating empty pixel values of a reference image using consecutive adjacent pixel value information of the additional image, including an additional image arranged in a grid with a reference image to improve a sample number of the reference image, If the pixel corresponding to the first or last pixel of one pixel column to one pixel row of the additional image is one adjacent pixel in the column or row, the adjacent pixel value is calculated as a virtual pixel value, And calculating the average value of the adjacent pixel values as virtual value when there are two adjacent pixels. An image definition step of defining the additional image contributing to the improvement of the number of image samples of the reference image by reading pixel values of the grid array image signal; When one pixel column corresponds to the first or last pixel of one pixel row of an additional image and there is one adjacent pixel in the column or row, the adjacent pixel value is calculated as a virtual pixel value, and when there are two adjacent pixels. A virtual value estimation step of calculating an average value of the adjacent pixel values as a virtual value; An image conversion step of converting each pixel value into color space data and converting an entire image into frequency space; And a quantization step of integerizing each component of the frequency domain. The method of claim 2, The converting step may include a first converting step of converting each pixel value into color space data; Dividing the entire image into sections having a size of 8x8 pixels, and then converting the data of each 8x8 section into frequency space using a two-dimensional discrete cosine transform. Grid array video compression method. In the apparatus for compressing a grid array image, An image definition unit for defining a reference image and an additional image contributing to improving the number of samples of the reference image by reading pixel values of the grid array image signal; In the case where there is one adjacent pixel in the column or row corresponding to the first or last pixel of one pixel column to one pixel row of the additional image, the adjacent pixel value is estimated as a virtual pixel value, and there are two adjacent pixels. A virtualization prediction unit estimating an average value of the adjacent pixel values as a virtualization value; An image converter for converting the pixel value of the reference image and the values of the virtual device into one image frame in a color space in accordance with an acquisition position so as to be expressed in a frequency domain or a region where a frequency domain and a spatial domain are mixed; Grid array image compression device including an encoder for quantizing the output image signal converted from the representation region of the image conversion unit A method for reconstructing an original pixel value of a reference image by a virtual pixel value estimation method using pixel values and empty pixel values of an additional image generated by inverse image transforming a grid array compressed video signal, If there is only one adjacent pixel used for estimating the virtual pixel value corresponding to the first or last empty pixel of one pixel row to one pixel row of the additional image, the corresponding pixel value is calculated as the virtual pixel value. , When the original pixel is located between two consecutive empty pixels, the corresponding pixel value is calculated by Equation 7 using two consecutive empty pixel values (a, b). Reconstruction method of grid array image using the method. [Equation 7] Original pixel value = (2 × b)-a An apparatus for reconstructing an additional video and a reference video from a grid array compressed video signal composed of specific binary codes, An inverse encoding unit for expressing a compressed video signal by inverse quantization of the binary codes in a frequency domain or a region in which a frequency domain and a spatial domain are mixed or reconstructing an image frame by interpreting the binary codes; When the output signal of the inverse encoder is a video signal expressed in a frequency domain or a region in which a frequency domain and a spatial domain are mixed, one image frame including pixel values and virtual pixel values of the reference image from the decoded signal. An inverse image converting unit calculating pixel values of If the pixel value of the image frame is the value of the virtual pixel estimated from the additional video, the grid array compressed image restoring apparatus comprising a pixel estimator for estimating the original pixel value by the virtual pixel value estimation method of claim 5 .

Images