JPH027178A - High-speed picture data enlarging/reducing system - Google Patents

Abstract

PURPOSE:To promptly enlarge and reduce picture data by referring to the value of each rotation shift register with successively shifting the register, controlling the input/output of the data and plural concentration calculating parts, and simultaneously executing a concentration operation for the row or the column of the continuously converted picture. CONSTITUTION:A concentration calculating part 116 calculates the value of a converted picture element from the respective latest two bits overflowing an the result of shifting input shift registers 112 and 113, where the original picture data are stored, and the horizontal and vertical components of intra-lattice coordinates, which are stored at the highest-order of rotation shift registers 118 and 119. Simultaneously, a concentration calculating part 117 calculates the value of the converted picture element from the respective latest two bits overflowing as the result of shifting input shift registers 114 and 115, and the horizontal and vertical components of the intra-lattice coordinates, which are stored at the highest-order of the rotation shift register 118 and at the second highest-order of the rotation shift register 119. Thus, by simultaneously processing the continuous rows or the columns, even when the original picture is small, since the plural concentration calculating part can be effectively applied, the processing time can be made shorter.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to an apparatus for enlarging and reducing digital images.

This relates to the scaling method.

[Conventional technology]

Image scaling refers to the process of resampling the original image using a new grid interval.

First, terms and concepts related to scaling will be explained with reference to FIG. The original pixel in the i-th row and j column of the original image is represented by Pij, the converted pixel in the m-th row and n column after scaling is represented by Qmn, and the set (Qmn) is called a converted image. The grid spacing in the horizontal and vertical directions is A and B, respectively, and the scaling factor is α.

When β, α=A/a and β=B/b hold.

In the scaling process, it is necessary to interpolate the density of a converted pixel from the density value of a neighboring original pixel, and this is called density calculation. Various methods have been proposed for concentration calculation.
For example, the SPC method, the logical sum method, the 9-division method, the projection method, etc. are known. Although the density calculation itself differs between the above methods, the input information to the density calculation is common to each method.

This input information will be explained below with reference to FIG.

Qmn is a conversion pixel. Pxmyn is the upper left point of the four neighboring points of Qmn. (Xm, Yn) is defined as the interlattice coordinate of Qmn. Also, 4 points, Pxmyn, Pxm+, yn.

Pxmyn+1. A rectangular area surrounded by Pxm+1yn+.

It is called the lattice area of Q m n, and the coordinates (um, n) of Qmn within this lattice area are defined as intra-lattice coordinates.

The information required to calculate the value of Qmn is um.

vn, 4 neighboring points P X my n , P X m+1
y n HPxmyn+1. Pxm+1yn+.

Xm, Yn, Llm, and vn can be determined by linear equations.

Xm=1 nteger I: (m-1)
Xb/Bl +IYn=integer ((n-1)
Xa/A)+1 m= (m-1) Xbm
odB'l, Ln= (n-1)XamodA also. Q
Lattice area spacing △X between the lattice area of mn and the lattice area of Qm+,n
The lattice area interval ΔYn between the lattice area of m, Qrnn and the lattice area of Qmn◆1 can be determined by a linear equation.

△Xn=Xm+, -Xm ΔYn:=Yn+, -Yn Next, a conventional scaling method will be described below.

A HB g a g If b is limited to integer values, u
m is a periodic sequence with a period of B, and the subsequence U is (inverse □ + +'
jz +...+"a).?/'n is also a periodic sequence with period A, and the subsequence U is (V, 7h,...r v
-A). Furthermore, ΔXm is a synchronous sequence with a period of B, and the partial sequence ΔX is (ΔX1.ΔX 2 + ”' rΔX
B). ΔYn is also a periodic sequence with a period of A, and one partial sequence Δ is now (ΔY 2 ΔY 2 . . . , ΔYA). Focusing on this periodicity, the m-th and m-th + A of the converted image
, m + 2 A, . . . The conventional method is to commonly control the calculations for each pixel in the rows column by column and execute them simultaneously.

That is, it has a plurality (K) of concentration calculation sections, and the first concentration calculation section has a Q factor of 0. Q1□,...+Qi1. I;Qz
□.

Q2□, ... yQ2N; ... are processed one pixel at a time in the order of
.

Q1◆^2+ ”'r Q1÷＾,;Q2+□l l
This is a method in which the processing is performed in the order of Qz+Az+"'Qz"AN:..., and the second density calculation section is processed in the same way in parallel.

[Problem to be solved by the invention]

In the above conventional technology, when the original image is small, only the first density calculation section is used even if there are multiple density calculation sections.
Processing takes time because other concentration calculation units are not used effectively. In addition, even though it is possible to perform density calculations using a common original pixel for the m-th row of the converted image and multiple consecutive rows during scaling, the conversion process for each row requires reading the original image separately. As a result, the same original image is read many times, resulting in a large number of memory reads and time-consuming processing. There was a problem.

An object of the present invention is to eliminate the above-mentioned drawbacks of the prior art and to perform scaling of image data at high speed.

[Means to solve the problem]

The above purpose is to limit A, B, a, and b to integer values,
We focused on the fact that um, 1)-n, ΔXm, and ΔYn each have periodicity, and that by sequentially adding the values of ΔXm or ΔYn, it is possible to obtain the row or column address necessary for concentration calculation. and each subsequence 'nr 9r
Δx, A'? This is achieved by providing a rotating shift register for storing the data, a plurality of concentration calculation sections, and a control section for controlling data input/output and the plurality of concentration calculation sections.

[Operation] The control section sequentially shifts each rotating shift register to refer to the values, controls data input/output and a plurality of density calculation sections, and performs density calculations for consecutive rows or columns of converted images. By executing these simultaneously, image data can be enlarged or reduced at high speed.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG.

FIG. 1 shows the overall configuration of a circuit that implements this scaling method. This circuit is an example in which two concentration calculation sections are provided,
Density calculations for two consecutive rows of converted images are performed simultaneously and in parallel column by column.

101 is an original image memory for storing IjK images; 104;
105 is a counter that sequentially adds overflow values as a result of shifting the rotating shift register 104. 102 is a rotating shift register 104. Then, referring to the value of the counter 105, input the original image data necessary for converting the m rows and m+1 rows of the converted image from the original image memory 101, and input the original image data to the input buffers 107 and 1081.
09. Input control units 107 and 108 that transfer to 110 input the original image data Xm rows and X
Input cover sofas each storing m+0 rows;
09, 110 are input buffers that store original image data necessary for converting m+1 rows of the converted image; 112, 113, 1;
14,115 is.

Input shift registers 118 and 119 input and shift parallel data from input buffers 7107, 108, 109, and 110, respectively, and rotation shift registers 118 and 119 store horizontal component subsequences and vertical component subsequences of grid coordinates, respectively. , 116 and 117 are the original pixel data input from the input shift registers 112, 113, 114, and 115, respectively, and the rotating shift register 11.
Density calculation units 120 and 121 calculate the value of the converted pixel based on the intra-lattice coordinates input from 8°119.

Output shift registers 122 and 123 store the converted pixel values calculated by the density calculation units 116 and 117, respectively, and convert them into parallel data; output buffers 122 and 123 store the converted image data of the output shift registers 120 and 121, respectively; 125 is.

A converted image memory 124 for storing converted images is an output control unit that sequentially stores the converted image data in the output buffers 122 and 123 in the converted image memory 125.

106 is a rotating shift register that stores a subsequence ΔY of the horizontal component of the grid interval; 111 is an output of the rotating shift register 106;

A timing control section 103 controls the shift timing of each shift register and the timing of concentration calculation, and a system control section 103 controls the entire operation of one circuit.

Next, the operation of this embodiment will be explained.

First, the system control unit 103 inputs the partial sequence Δ to the rotation shift registers 104, 106, 118, and 119 according to the scaling factors α and β.

ΔY,? A and L are calculated and loaded, the counter 105 is set to 1, and the input control unit 102 is activated.

The input control unit 102 inputs the first and second lines of the converted image data.
In order to calculate the row number, the value of the counter 105 and the most significant bit of the rotating shift register 104 are referred to. The value of the counter 105 indicates the line number of the original image data to be input to the input buffer 107, and the line next to the line of the original image data to be input to the input buffer 107 is input to the input buffer 108. Further, the value obtained by adding the value of the most significant bit of the rotate shift register 104 to the value of the counter 105 indicates the row number of the original image data to be input to the input buffer 109. The line next to the line of original image data is input. Now, the value of the counter 105 is 1, so if the most significant bit of the rotate shift register 104 is O, the first row of the original image data is transferred to the input buffers 107 and 10.
9, W bits of the second line are input to input buffers 108 and 110, respectively. Also, if it is 1, JXliXl-
The first line of the data is input to the input buffer 107, the second line is input to the cover sofas 108 and 109, and the third line is input to the input buffer 1.
If it is 2, the first line of the original image data is input to the input buffer 107, the second line is input to the input buffer 108, the third line is input to the input buffer 109, and the fourth line is input to the input buffer 109. W bits are respectively input to the input buffer 110.

When the input is completed, the system control unit 103 transfers the original image data in the input buffers 107, 108, 109, and 110 to the input shift registers 112, 113, 114, and 115.
Transfer each. When the transfer is completed, the system control unit 103 activates the input control unit 102 in preparation for the first transfer, and inputs the next W-bit original image data in the same row to the input buffers 107, 108, 109, and 110. .Also,
When the transfer is completed, the system control unit 103 activates the timing control unit 111.

The timing control section 111 has an input shift register 112.
, 113, 114, and 115 to the left by 2 bits, and the density calculation units 116 and 117 are activated.

The density calculation unit 116 uses the latest 2 bits that are overflowed as a result of shifting the input shift registers 112 and 113 in which the original image data is stored, and the lattice data stored at the top of the rotate shift registers 118 and 119, respectively. The value of the converted pixel is calculated from the horizontal and vertical components of the target coordinates. At the same time, the concentration calculation unit 117 overflowed as a result of shifting the input shift registers 114 and 115. The latest two bits, the most significant bit of the rotating shift register 118, and the second bit from the top of the rotating shift register 119 are respectively stored. The value of the converted pixel is calculated from the horizontal and vertical components of the grid coordinates.

Then, the timing control section 111 shifts each of the output shift registers 120 and 121 to the left by 1 bit, and inputs the value of the converted pixel calculated by the density calculation sections 116 and 117 to the least significant bit, respectively.

The timing control unit 111 sequentially shifts the rotation shift register 106 to the left by 1 bit, and transfers only the overflowing most significant data value to the input shift registers 112 and 113.
, 114, 115 to the left]. Furthermore, the rotation shift register 118 is shifted one bit to the left, and the density calculation units 106 and 107 are activated. concentration calculation section 106,
Similarly, 107 sequentially calculates the values of the converted pixels. However, the system control unit 103 controls the input shift register 112
．． Every time data 113, 114, 115 disappears,
Original image data is input from input buffers 107, 108, 109 and 110, respectively.

Furthermore, the timing control section 111 sequentially shifts the output shift registers 120 and 121 to the left by 1 bit, and inputs the converted pixel value calculated by the density calculation sections 116 and 117 to the least significant bit. Then, the system control unit 103 controls the output shift registers 122 and 1 every time the converted pixels of the output shift registers 120 and 121 accumulate W bits.
23 respectively.

When the transfer is completed, the system control unit 103 outputs the output control unit 1.
24.

The output control unit 124 transfers the converted image data of the output sofas 122 and 123 to the converted image memory 1 every time it is activated.
Write W bits each to the first and second lines of 25.

When the conversion process for the first and second lines of the converted image data is completed, the rotation shift register 104 is shifted 2 bits to the left in order to calculate the third and fourth lines of the converted image data. The obtained 2-bit value is added to the value of the counter 105, and the rotate shift register 119 is shifted 2 bits to the left. Similarly, the counter 105
and the most significant bit of the rotate shift register 104, the input buffers 107, 108, 109 .
The original image data is input to 110 to perform density calculation, and the converted image data of the output sofas 122 and 123 are respectively stored in the third row of the converted image memory 125.

Write on the 4th line.

Thereafter, by repeating the above process in the same way, a converted image can be obtained at high speed.

〔Effect of the invention〕

According to the present invention, by processing consecutive rows or columns simultaneously, even if the original image is small, it is possible to effectively utilize a plurality of density calculation units, thereby reducing the processing time. Furthermore, since the original image in the same row or column can be used in density calculation processing for converted images in a plurality of rows or columns, the number of times the original image is read from the memory can be reduced, and the processing time can be shortened.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of an embodiment of the present invention, and FIGS. 2 and 3 are explanatory diagrams of image enlargement/reduction. 101...Original image memory, 102...Input control unit. 103... System control unit, 104, 106, 118
．． 119...Rotate shift register, 1o5...
- Counter, 107, 108, 109, 110... Input buffer, 112, 113, 114, 115... Input shift register, 111... Timing control unit, 1
16,117...concentration calculation section, 120,121...
Output shift register, 122, 123... Output sofa, 124... Output control section, 125... Conversion image memory. lσj λtepewo 1 tsubo mark/2φ output tide l# part

Claims (1)

[Claims] 1. A buffer memory for buffering multiple lines of image data, a control unit that controls input and output of image data, multiple density calculation units that perform density calculations, buffer memory shift timing, and density calculation. By providing a control unit that controls the activation timing of the unit, and a rotating shift register that stores intra-grid coordinates and grid area spacing, concentration calculations for rows or columns of consecutive exchanged images can be performed simultaneously in parallel. A high-speed image data scaling method featuring: