JPS62139485A - Subsample interpolation system - Google Patents

Abstract

PURPOSE:To secure the continuity of image data by using a value obtained by multiplying each picture element by a function for using plural picture elements around the picture element concerned as a signal value after interpolating at every block on the receiving side. CONSTITUTION:A signal thinned based on an ordinary method and sent from the transmission side is inputted to a decoding circuit 10 and interpolated similarly to the ordinary method. The output of the circuit 10 is applied to an interpolating circuit 20 to execute interpolating operation for removing discontinuity generated among respective bloks formed at the time of thinning. When the value of the remarked picture element is Z0, a value just before the Z0 is Z<-1>, a value separated by 2 before the Z0 is Z<-2>, a value just after the Z0 is Z<1>, and a value separated by 2 after the Z0 is Z<2>, a function f(z) expressed by the shown equation is formed and the value of the function f(z) is set up to a new signal value for the remarked picture element Z0. Consequently, the continuity of the image data can be secured.

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] A discontinuous image data string is converted into a continuous image data string by performing simple interpolation processing.

[Industrial application field]

The present invention relates to a band compression method for image No. 13, and in particular to a sub-sampling method for reducing the amount of information generated by thinning encoded image data, and relates to an interpolation method as a method for reproducing thinned image data. be.

[Conventional technology]

It is well known that when transmitting an image signal, the transmission time or the transmission band is reduced by compressing the band of the image signal.

Generally, when transmitting an image signal after band compression, the image signal is PCM encoded into a digital signal in units of 8 bits, and a DPCM encoding format is adopted in which the difference from the previous signal is calculated. The amount of information transmitted is further reduced by sub-sampling (thinning out) without transmitting the information during the period.

According to the conventional technology, when performing subsampling to reduce the amount of information transmitted in an image like this (A), due to hardware constraints, the entire screen is divided into small screen units (block units), and this block Normally, the positions of pixels to be thinned out are determined according to a standard within the range.

FIG. 3 is a diagram illustrating an example of conventional subsampling.

As is well known, each row obtained by scanning the entire screen F (original image) with a plurality of scanning lines is composed of a plurality of pixels E1 to En. Note that this one row of pixel signal columns is continuous data.

To explain a simple example of subsampling in 4HL, a plurality of pixels E and -E in one row are divided into several blocks. For example, 5 consecutive pixels are set as 1 block and 1 block,
The pixels E, ~E, are the first block, and the pixels E6~E. Let be the second block...

After dividing into individual blocks in this manner, the second and fourth pixels of each block are thinned out.

Therefore, in the first block, pixels E2 and E4 are thinned out, and pixels E, E3, and ES are sent out, and in the second block, pixels E7 and E are thinned out, and pixels E are sent out.

, 211, El. is sent.

The decoding circuit on the receiving side needs to receive the thinned out image signal and restore (interpolate) it.

In order to restore the thinned out pixel E2, for example, the data of the pixel E1 is used at the position of the pixel E2, or the average of the data of the pixel E1 and the pixel E is used. Similarly, to restore pixel E4, the data of pixel E3 is used at the position of pixel E4, or the average of the data of pixel E3 and pixel E is used.

Generally, interpolation of the image data of the portion thinned out by subsampling is obtained by multiplying the previous and subsequent image data by an arbitrary function.

[Problem that the invention seeks to solve]

However, due to the sampling frequency, the number of pixels included in one block is, for example, 5.7 due to the multiple of the entire screen.
．． It is often set to an odd number such as 13. Therefore, when the subsampling is performed, the boundaries of each block become discontinuous data, so interpolation multiplied by an arbitrary function cannot be directly applied. This has the disadvantage that stripes appear on the boundaries of blocks.

[Means for solving problems]

The problem mentioned above is that in a band compression method that divides continuous image data into multiple blocks and performs prescribed thinning for each block, the receiving side performs conventional interpolation for each block, and then This can be solved by setting the signal value of the pixel to be a value multiplied by a function that uses a plurality of pixels centered on the pixel.

[Effect]

According to the present invention, an image signal is divided into a plurality of blocks, thinned out within each block, and interpolation is performed within each block, so that discontinuity in the reproduced screen that inevitably occurs between blocks is removed. By performing a process of converting discontinuous data obtained by interpolating within a block into continuous data, there is an advantage that stripes between blocks are eliminated from the reproduced image.

〔Example〕

FIG. 1 is a diagram showing one embodiment of a subsample interpolation method according to the present invention.

FIG. 2 is an explanatory diagram of the subsample interpolation method according to the present invention.

In the figure, 10 is a decoding circuit, and 2o is an interpolation circuit according to the present invention.

In FIG. 1, a decoding circuit 1o is a conventional decoding circuit, and a signal sent from the transmitting side and thinned out according to the conventional method enters the decoding circuit 1o. The decoding circuit 10 performs the same interpolation as in the conventional method and outputs the same.

When an image is later reproduced using the output of this decoding circuit 10, as mentioned above, stripes occur on the boundary lines of the blocks, so the output is input to the interpolation circuit 2o according to the present invention to perform second interpolation. conduct.

The signal sequence shown in Figure 2 is the 13th sequence in which the signals that have been thinned out for each block and then interpolated for each block are continuously created again. It is.

Paying attention to any negative pixel in the above signal string, perform the following arithmetic operation. In this explanation, calculation operations are performed using up to two pixels before and after the pixel of interest, but it is not necessary to stick to this, and up to three pixels before and after, or up to four pixels before and after the pixel of interest may be taken up. The more the scope is expanded, the more continuity is ensured.

The value obtained by multiplying the value of the pixel of interest by the function f (z) is set as the value of the new pixel.

f(z)=a,z−”+azz−’+a,z.

+34 z″+as zl′ where a, ~a, is a weighting coefficient, zo is the pixel of interest, z −1 is the pixel immediately before pixel z0, z−2 is one of the pixels z −1 The previous pixel, z+l, represents the pixel after pixel z0, and Z12 represents the pixel after pixel z1.

Note that a, +a2+a3+a4+a5=l.

For example, let us now focus on pixel E in FIG.

The value of pixel E5 is multiplied by weighting factor a, the value of pixel E4 is multiplied by weighting factor a2, the value of pixel E3 is multiplied by weighting factor a1, the value of pixel E6 is multiplied by weighting factor a4, and the value of pixel E6 is multiplied by weighting factor a4. The value of E is multiplied by the weighting coefficient a5 to find the sum, and this sum is set as the new value of the pixel E5.

Next, in the case of pixel E6, the value of pixel E6 is multiplied by weighting coefficient a3, the value of pixel E5 is multiplied by weighting coefficient a2, the value of pixel E4 is multiplied by weighting coefficient a1, and pixel E, The value of the pixel E is multiplied by the weighting coefficient a4, the value of the pixel E is multiplied by the weighting coefficient a, the sum is determined, and this sum is set as the new value of the pixel E6.

By performing the above calculation on each pixel output from the conventional decoding circuit in this manner, it is possible to remove discontinuities that occur between blocks provided during thinning.

〔Effect of the invention〕

As described above in detail, the present invention has the great effect of ensuring the continuity of image data by performing a simple supplementary operation to the conventional simple interpolation method.

[Brief explanation of drawings]

FIG. 1 is a diagram showing one embodiment of a subsample interpolation method according to the present invention. FIG. 2 is an explanatory diagram of the subsample interpolation method according to the present invention. FIG. 3 is a diagram illustrating an example of conventional subsampling. In the figure, F is the entire screen, E and -E are pixels, 10 is a decoding circuit, and 20 is an interpolation circuit according to the present invention. / # - Tomeiyo 5 Fubufunpuruo car direction IJ, Sea practice, 1st 12th Mere desert aft, Yo 3 Sobus sample Nekojima direction θ theory 8H
Figure 2

Claims (1)

[Claims] In a band compression method that divides continuous image data into a plurality of blocks and performs prescribed thinning for each block, after interpolation is performed for each block on the receiving side, each pixel is A sub-sample interpolation method characterized in that the signal value of the pixel is set to a value obtained by multiplying by a function using a plurality of pixels centered on the pixel.