JPS6268366A - Picture element interpolation system - Google Patents

Abstract

PURPOSE:To obtain a reproduced picture with less deterioration in picture quality even for a picture where picture elements change rapidly by selecting a picture element used for interpolation operation while the quantity of correlation by differences is checked and the picture element having the largest correlation is used selectively. CONSTITUTION:A picture element shown in a white round mark is inserted by referring to sub-sample picture elements 201 shown in 4 hatched lines around the said picture element obliquely. Then the picture element shown in the white round mark is interpolated by referring to the sub-sample picture elements 201 and four picture elements around the said element in horizontal and vertical directions comprising picture elements finished for the interpolation shown in round marks on hatched lines. With respect to the picture element not subjected to interpolation at the peripheral part of the picture, the interpolation is applied by using values of adjacent picture elements subjected to sub-sampling or the weight mean value with respect to the ratio of distances from plural adjacent picture elements.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention utilizes the redundancy of image data to estimate the value of a missing pixel from the values of neighboring pixels to reconstruct an image. Regarding the insertion method.

BACKGROUND OF THE INVENTION In recent years, with the development of communication technology, the demand for image transmission and storage has increased rapidly. In contrast, these demands are being realized by the development of memory elements with large-scale storage capacity brought about by advances in computer technology, and by high-efficiency encoding technology using digital signal processing technology.

One of the simplest methods for encoding images, especially gradation images, is to thin out pixels in advance by sampling a band-limited image signal at a frequency that is a fraction of its Nyquist rate. There is a method of reducing the number of pixels (hereinafter referred to as subsampling). Although this method may cause some eriazingi distortion, it is an effective method for reducing the amount of information.

As a method for obtaining a reproduced image close to the original image from this subsampled pixel information, the co-order interpolation method (supervised by Toshifumi Sakata, published by Combita Imaging Doshinsha Kogyo, special edition, 1983) is known.

This co-order interpolation method will be explained below using FIG. 4.

FIG. 4 is a diagram showing the relationship between subsampled pixels (hereinafter referred to as subsampled pixels) and non-subsampled pixels (hereinafter referred to as non-subsampled pixels) in a certain portion of an image.

In FIG. 4, 401 indicates a subsample pixel, and 402 indicates a non-subsample pixel. In the figure, there are 4 pixels in the horizontal direction,
Subsampling is performed every 4 pixels in the vertical direction (normally called supline in the vertical direction)
. Four sub-sample pixels Poo, P2O, PO4, P
The reproduced pixel value "J ('IJ is Po.

The subscript indicating the pixel position in the horizontal and vertical directions with reference to , is given by Equation 11.

qIJ ==(1u)(1v)Poo 1u(1-v)
P2O-4-(1-u)vPo4+ LIVP44-
= (1) (However, Poo, P2O, PO4, P44
are sub-sample pixels Poo, P2O, PO4, respectively.
P44 represents the pixel value, and u and v represent the internal division ratio values in the horizontal and vertical directions for the line segments POOP40 and POOPO4 of Qij, respectively. ) This method calculates the value of the pixel to be interpolated by using the weighted average of each subsample pixel value, with the weight being the ratio of the distance from the four subsample pixels. In this case, the reproducibility is good because there is a high correlation between the pixel to be interpolated and each subsample pixel used for interpolation.

Problems to be Solved by the Invention However, in the above method, there is a problem in that image quality deteriorates significantly when the sub-sample pixels used for interpolation include those having a low correlation with the pixels to be interpolated. A specific example will be shown and explained below.

FIG. 3 is a diagram showing the relationship between sub-sampled pixels and non-sub-sampled pixels when subsampling is performed every two pixels in both the horizontal and vertical directions in a certain part of the image.

In Fig. 3, an extreme case is considered in which the image before subsampling is bordered by the pixel on the dashed line ll, where all pixels above the dashed line 11 are black and all pixels below the dashed line 11 are white. think. At this time, the reproduced pixels Qll, Q
33 has a gray-white color because three of the four surrounding sub-sample pixels used for interpolation are white and one is black. Conversely, three of the reproduced pixels Q31 are black and one is white, resulting in a gray-black color. For this reason, in the reproduced image, white and gray-white appear alternately on the broken line 12, and black and gray-black appear alternately on the broken line 13, resulting in a significant deterioration in image quality at the boundary between white and black.

The present invention is intended to solve the above-mentioned problems, and aims to obtain a good image by interpolation even for an image in which pixel values change rapidly.

Means for Solving the Problems In the present invention, after subsampling the sampled image so that the pixel spacing is even in both the horizontal and vertical directions, For those that constitute the vertices of a square or a quadrilateral close to a square, and there is a pixel to be interpolated at the center, calculate the value of the pixel to be interpolated by calculating the difference between the pixels at the vertices for each diagonal. select a set of pixels whose difference is smaller when
Next, for the pixels that have already been interpolated and the subsampling, find out whether the four neighboring pixels constitute the vertices of a square or a quadrilateral close to a square, and there is a pixel to be interpolated at the center of the square. The value of the pixel to be interpolated is determined in the same manner as described above, and the interpolation is performed sequentially using the pixels that have already been interpolated.

Effect of the Invention The present invention uses the method described above to select pixels to be used in interpolation calculations by examining the magnitude of correlation by difference and selecting pixels in the direction with the strongest correlation, which is suitable for images with rapid changes in pixel values. This makes it possible to obtain reproduced images with less deterioration in image quality.

Embodiment FIG. 1 is a block diagram of a device for realizing a pixel interpolation method in an embodiment of the present invention.

In FIG. 1, 101 is an address controller that instructs addresses at which sample pixels should be stored and read out in an image memory (described later); 102 is an image memory that reads and writes sample pixels based on instructions from the address controller 101; 103 is a subtracter that subtracts every two pixels from the pixel value sent from the image memory 102; 104 is a comparator that compares the values sent from the subtracter 103; and 105 is a pixel value sent from the image memory 102. An adder adds the pixel value every two pixels; 106 is a divider that divides the value sent from the adder 105 by 2; 107 temporarily holds the result of the divider 106; A register that selectively outputs the value held in accordance with the comparison result; 108 is an interpolation controller that controls the address generation of the address controller 101 and the operation timing of the subtracter 103, comparator 104, and divider 106; It is.

The operation of the above configuration will be explained below.

The address controller 101 selects four sample pixels PI, P2, P3, P4 to be used for interpolation based on the initial setting data sent from the interpolation controller 108.
and the address of the pixel Q to be interpolated, and calculate the address of the sample pixel PI, P2, P3,
For P4, along with the read permission signal, P4,
The addresses are stored in image memory 1 in the order of Pi, P3, and P2.
Send to 02. Furthermore, regarding the pixel Q to be interpolated, the address of the pixel Q is sent to the memory 102 along with the write permission signal. The image memory 102 receives the read permission signal and address from the address control unit 101, and reads P4. PI, P3
, P2 pixel P4. PI, P3. P2 is sent to subtracter 103 and adder 105 in this order. Also,
In response to the write permission signal from the address controller 101 and the address of the pixel Q, the value q of the interpolated pixel sent from the register 107 is written to the address of the pixel Q. Subtractor 1
03, the pixel values sent from the image memory 102 are subtracted every two pixels, the absolute value is taken, and the value is sent to IF4-P.
ll, IF5-P21, and the sequential comparator 10
Send to 4. The comparator 104 compares the magnitude of the correlation in the diagonal direction by comparing the magnitudes of the two values sent from the subtracter 103, and uses the result as a selection signal according to the timing control signal from the interpolation controller 108. Send to register 107. On the other hand, the adder 105 adds the pixel values sent from the image memo IJ 102 every two pixels, and sequentially sends the results P4+P1°P3+P2 to the divider 106. The divider 106 divides the value sent from the adder 105 by 2 and calculates the average value P4 +Pt P3
Send IF2 to each Regis month 07.

2' The register 107 temporarily stores the two average values sent from the divider 106, and transfers the contents of the register designated by the selection signal from the comparator 104 to the image memory 102 as the interpolation pixel value q. send. The pixel value q at this time is the (2nd
) can be expressed as the following equation.

The interpolation controller 108 sends initial setting data for address calculation to the address controller 101, and also sends the address controller 101, the subtracter 103, the comparator 104, the adder 105,
A timing control signal is sent to adjust the operation timing of the divider 106 and register 107.

As a result of the above, when the interpolation process for a certain interpolation pixel is completed, the interpolation controller 108 sends the next initial setting data to the address controller 101, and the process for the next interpolation pixel is executed.

Further, when the -way interpolation is completed, the interpolation control unit +08 changes the initial setting data so as to select four pixels including the pixel that has already been interpolated. In this way, the values of pixels to be interpolated are sequentially determined.

FIGS. 2A and 2B are conceptual diagrams schematically showing how interpolation is performed in an embodiment of the present invention.

In FIG. 2, 201 represents a subsample pixel, 202 a non-subsample pixel, 203 a pixel to be interpolated, and 204 a pixel for which interpolation has been completed.

First, in the first step, the pixel indicated by the white circle in FIG. In FIG. 2(b), the pixel indicated by a white circle is interpolated with reference to the four pixels around it in the horizontal and vertical directions, consisting of the sub-sample pixel and the pixel for which interpolation has been completed, indicated by the diagonally shaded circle.

For pixels for which interpolation is not performed in the peripheral area of the image, interpolation is performed using the values of adjacent pixels subjected to subsampling or a weighted average value based on the ratio of distances from a plurality of adjacent pixels.

According to this embodiment, as described above in the conventional example in FIG. black,
Even for an image with sudden gradation changes where all pixels below the broken line 11 are white, the reproduced pixels Qll and Q33 become white because they are interpolated from pixels with high correlation in the diagonal direction, and Q31 Since the image becomes black in the same way, the quality of the reproduced image is greatly improved compared to the conventional method. Furthermore, the image quality can be improved compared to conventional methods even for images where pixel values change gradually, and the effect is significant.

Effects of the Invention As described above, the present invention can determine the value of a pixel by examining the magnitude of the correlation using the difference and selecting the pixel in the direction with the strongest correlation. Even for images with rapid changes, it is possible to obtain reproduced images with less deterioration in image quality, which is highly effective.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a device that implements a pixel interpolation method in an embodiment of the present invention, FIG. 2 is a conceptual diagram schematically showing how interpolation is performed in an embodiment of the present invention, and FIG. 3 and 4 are conceptual diagrams showing the conventional interpolation method. 101... Address control unit, 102... Image memory, 103 Subtractor, 104... Comparator, 105...
Adder, 106...Divider, 107...Register, 1
08...Interpolation control unit, 201, 301, 401...
Sub-sample pixels, 202 , 302 . 102... Non-subsample pixel, 203... Pixel to be interpolated, 204... Pixel for which interpolation has been completed. Name of agent: Patent attorney Toshio Nakao (1st person)
Figure 2 (α〕/2C”) Figure 3゛・7゜

Claims (1)

[Claims] After subsampling the pixels obtained by sampling the gradation image to reduce the number of pixels, the pixels that were not subsampled are estimated from the subsampled pixels and interpolated to reproduce the original image. are performed so that the pixel spacing is an even number in both the horizontal and vertical directions, and the four adjacent pixels constitute the vertices of a square or a quadrilateral close to a square, and whose center is the pixel to be interpolated. The value of the pixel to be interpolated for those that match the position of is the pixel of the set of pixels for which the difference is smaller when the difference between the pixels at the vertices is calculated for each diagonal line using those four pixels as reference pixels. Perform sequential interpolation by giving the average value of the values of
Next, regarding interpolated pixels and subsampled pixels, for those where four adjacent pixels constitute the vertices of a square or a quadrilateral close to a square, and whose center coincides with the position of the pixel to be interpolated, A pixel interpolation method characterized in that the value of the pixel to be interpolated is determined, and interpolation is performed sequentially based on four surrounding reference pixels.