JPS6314552B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a television image conversion device useful for generating special effects such as image rotation in, for example, television broadcasting stations. When producing programs for television broadcasting, various special effects such as moving, reversing, and rotating images are used. For example, when the image pattern 1 is rotated about a point 2 as shown in FIGS. 1a and 1b, this is generally done by controlling the rotation of the optical system of a television camera. However, when applying this type of effect to a television image recorded on a VTR or a television image sent from another station, the television image must first be displayed on a monitor, etc. , it was necessary to image this again with a television camera. However, as mentioned above, once the television signal is converted into an optical system, the image quality deteriorates significantly. Further, there are disadvantages in that the device configuration required for the conversion process is large-scale, and the operation thereof is also very complicated. Because of this, there was a limit to the ability to use special effects to produce a wide range of programs. The present invention was made in consideration of the above circumstances, and
The purpose is to provide a television image conversion device that can easily perform image conversion such as movement, inversion, and rotation of a television image without causing deterioration in image quality. At the same time, the present invention provides a television image that allows special effects (movement, inversion, rotation) to be obtained by processing television signals in real time using simple hardware, without requiring signal processing using a large-scale computer. The purpose is to realize a conversion device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the apparatus will be described along with the principle (function) of television image conversion according to the present invention with reference to the drawings. The image indicated by the coordinates (x, y) is expressed by the coordinates (0,
0) as the center of rotation by an angle α, the coordinates (x', y') of the above image are given as follows. On the other hand, suppose that a television image is sampled and shown as shown in FIG. This sampling unit is M samples performed at a pixel interval h in the horizontal scanning direction H, and N samples performed at a pixel interval v in the vertical scanning direction V. Therefore, each pixel coordinate (x, y) of the input television image is denoted as x=mh m=0, 1, 2, . . . My=mv n=0, 1, 2, . Therefore, the coordinates (x', y') of the above image mapped by the angle α mentioned above are x' _(n,o) = mh・cosα+nv・sinα y′ _(n,o) = −mh・sinα+nv・cosα shown. Since the coordinate transformation calculation process shown in this formula can be performed only by addition and subtraction,
Generally, if sampling is performed at a sampling frequency _s that is three to four times the color subcarrier frequency _sc of the television signal, each pixel coordinate can be converted at each sampling interval with sufficient time. By the way, when _s = 3 _sc , the sampling interval is about 93 nsec, and when _s = 4 _sc , the sampling interval is about 93 nsec.
This is 70 nsec, and it is easy to perform addition and subtraction processing within this period. Therefore, if the rotational transformation coordinates (x' _(n,o) , y' _(n,o) ) of the input coordinate position (m, n) are known, the rotational transformation coordinate (x' _(n+1,o) ,
y′ _(n+1,o) ) is x′ _(n+1,o) =x′ _(n,o) +h cosα y′ _(n+1,o) =y′ _(n,o) −h sinα is derived by simple addition and subtraction processing. Rotation transformed coordinates (x′ _(n,o+1) ,
y′ _(n,o+1) and (x′ _(n+1,o+1) , y′ _(n+1,o+1) ) are derived, respectively. Now, assuming that the sampling coordinate position at which the rotationally transformed image is displayed takes the same coordinate axis as the sampling coordinate of the input image, the rotational coordinate position described above does not necessarily match the display coordinates (X, Y). Not exclusively. Rather, it generally causes deviations. When the coordinates (x', y') are normalized by the sampling intervals h and v, respectively, the above-mentioned transformed coordinates are shown as follows. x′ _(n,o) /h=X _n,o +ΔX _n,o y′ _(n,o) /v=Y _n,o +ΔY _n,oNote that X _n,o and Y _n,o are integers. The display coordinates shown, ΔX _n,o and ΔY _n,o, indicate the deviation in decimals. Therefore, the display coordinates (X _(n,o) , Y _(n,o) )
If this coordinate is indicated, some kind of interpolation operation of the rotated coordinate and its pixel information is required for this coordinate. FIGS. 3 and 4 are pixel layout diagrams for explaining an example of an interpolation operation applied to this device. The coordinates (x', y') connected by solid lines in Figure 3 are:
The rotational coordinates of the angle α are shown, and the coordinates (X, Y) connected by the broken line are the display coordinates. But now,
As shown in the enlarged view of Figure 4, when the display coordinates (X, Y) are specified, the rotational coordinates (x' _(n,o) , y' _(n,o) ) ( x′ _(n,o-1) , y′ _(n,
o-1) , ),
(x′ _(n-1,o) , y′ _(n-1,o) ), (x′ _(n-1,o-1) , y′ _{(n-
1,o-1) )} are calculated respectively. In the following description, the rotational coordinates will be abbreviated as (m, n). Method of pixel interpolation from each pixel information of these coordinates P _n,o , P _n,o-1 , P _n-1,o , P _n-1,o-1 to the display coordinates (X, Y) Although there are various methods such as simple averaging and weighted averaging, for example, the pixel information of the rotated coordinate where the distance between images is the minimum may be approximately replaced. This method is called a permutation approximation method, and will be briefly explained below. As shown in Figure 4, the display coordinates (X, Y) and the four rotating coordinates (m, n), (m, n-
1), (m-1, n), and (m-1, n-1) by square calculation, for example, d ² _n,o , d ² _n,o-1 ,
It is determined as d ² _n-1,o and d ² _n-1,o-1 . These are obtained by trigonometry operations from the decimal parts ΔX _n,o and ΔY _n,o described above. This can be achieved by comparing these distances with each other to find the minimum distance, and interpolating the information on the rotational coordinates to the display coordinates (X, Y) as an approximate value. FIG. 5 is a schematic diagram of the present apparatus which performs interpolation calculations using the permutation approximation described above. The television image signal P is sampled as shown in FIG. 2 and is sequentially input to the switch circuit 11 in time series. At this time, the image signal P is
1 directly, and also via a 1 sample delay circuit 12 and 1 horizontal scan delay circuit 1.
3 and another one-sample delay circuit 14 to the switch circuit 11. Therefore, at the input timing of the pixel signal P _n,o , the switch circuit 11 receives not only the above signal P _n,o but also the image signal P _n-1,o from one sample before, and the image signal P from one horizontal scan before. _n,o-1 , and further, image signals P _n-1 and P o _{-1 one sample before this P n} ,o _-1 are simultaneously obtained. One of these image signals is selectively extracted based on an output signal from an address selection circuit 15, which will be described later. On the other hand, signals of coordinates corresponding to each of the image signals, that is, input coordinate (x, y) information are sent to respective coordinate conversion circuits 16a, 16b, 16c, respectively.
16d. These coordinate conversion circuits 16
a, 16b, 16c, 16d are, for example, adders/subtractors,
It is composed of registers, etc., and performs the above-mentioned arithmetic processing according to preset rotation angle information α to obtain rotation coordinates (m, n). Note that these arithmetic processing methods and coordinate conversion circuits 16a, 16b,
As for 16c and 16d, various methods that have been introduced in the past may be used as appropriate depending on the specifications. Therefore, these coordinate conversion circuits 16a, 16b, 1
The rotational coordinates (m, n) obtained by 6c and 16d, respectively, are supplied to the address selection circuit 15, and are also supplied to the coordinate conversion circuits 16a, 16b, 16.
The signals are supplied to distance calculation circuits 17a, 17b, 17c, and 17d provided corresponding to the signals c and 16d, respectively. Distance calculation circuits 17a, 17b, 17c, 1
7d mainly consists of a squaring circuit and an addition circuit, and as mentioned above, the distance from the display coordinates (X, Y)
d ² _n,o are calculated respectively. Minimum distance judgment circuit 18
is the distance from the display coordinates (X, Y) calculated above
Input d ² _n,o , d ² _n-1,o , d ² _n,o-1 , d ² _n-1,o-1 and compare them with each other to find the minimum distance. There is. According to the minimum distance information determined by the determination circuit 18, the address selection circuit 15 determines a rotational coordinate that provides the minimum distance between the display coordinates (X, Y), and uses this as the selection control signal of the image signal. It is output as . Therefore, the switch circuit 11
If it is determined that the rotation coordinate (m, n) is at the minimum distance from the display coordinate (X, Y), it selects and outputs the image signal P _n,o . If the rotational coordinates (m-1, n) are determined in the same way, the image signal
P _n-1,o is the image signal P _n,o-1 when the rotational coordinate (m, n-1) is determined, and the rotational coordinate (m
-1, n-1), the image signal
P _{n-1 and o-1} are respectively selectively output. Thus, interpolation processing of the image signal of the rotation coordinate with respect to the display coordinate (X, Y) is performed here. By the way, when a television image is rotated by an angle α, it is expected that some areas will protrude from the display screen range. The image conversion process in this protruding area has no meaning. Therefore, in this apparatus, the start address calculation circuit 21 is activated in accordance with the conversion process start signal ST to calculate the display coordinate scanning start position for the rotation angle α. This calculation circuit 2
1 specifies the display coordinates (X=0, Y=0) as shown in FIG. 6, for example. Assuming that the display image consists of a group of sampled pixels represented by (k+1) pixels in the horizontal direction and (l+1) pixels in the vertical direction, the display address range calculation circuit 22
sets a display address range of 0Xk 0Yl according to the scanning start position coordinates. The coincidence detection gate circuit 23 determines whether or not the converted coordinates outputted from the address selection circuit 18 are within the display address range, and selects the display coordinates (X,
Y) is output as the address designation value of the high-speed RAM 24. Therefore, the high-speed RAM 24 writes the interpolated image signal P _n,o only to the display coordinates (X, Y) determined by the coincidence detection gate circuit 23 to be within the display address range. Become. As shown in FIG. 6, the image signal P _n,o is written into the high-speed RAM 24 one pixel at a time, starting from the one whose display coordinates (X, Y) are determined. The image signals written in the high-speed RAM 24 are sequentially read out, for example, two pixels at a time, under the control of the readout control circuit 25, and are sequentially stored in a large-capacity memory having a storage capacity of, for example, one field, that is, the main memory 26. transferred and written. A rotationally converted television image is formed by the group of image signals P _n,o addressed and written into the main memory 26 . The high-speed RAM 24 is configured independently for each horizontal scanning direction or vertical scanning direction of a display image, for example. Here, an example will be explained in which an independent configuration is provided for each horizontal scanning direction, but an independent configuration for each vertical scanning direction will function in exactly the same manner and will not cause any inconvenience. As shown in Fig. 6, if we perform interpolation calculation processing on the television image that has undergone rotational coordinate transformation and sequentially interpolate the display coordinates (X, Y), the display will depend on the scanning direction of the television image. The interpolation order of coordinates (X, Y) is
The display scan is different. In other words, in the example shown in Figure 6, after being interpolated to the display coordinates (0, 0),
(1, 0), (0, 1), (1, 1), (2, 0), (
0,
2) Interpolation is performed in the order of (1, 2) ~, (k, 0) ~. Therefore, as shown in the following table, the above-mentioned image signals and coordinate information are stored at each address of the high-speed RAM 24, which is independently configured in the horizontal scanning direction, as shown in the following table, for example.

【table】

Claims (1)

[Claims] 1. Each pixel coordinate of a sampled television image signal is coordinate-transformed according to a predetermined rule, and each pixel information at this transformed coordinate position is interpolated to display coordinates to create a television image. In the television image conversion device that performs the conversion, there is a main memory having an address corresponding to the display coordinates, a high-speed buffer memory configured independently corresponding to the row or column of the display coordinates, and a pixel at the coordinate position whose coordinates have been converted. means for obtaining pixel information to be interpolated to display coordinates near the converted coordinate position from the information; and means for transmitting the interpolated pixel information to the high-speed buffer memory corresponding to the row or column containing the display pixel near the converted coordinate position. means for writing along with pixel position information in said row or column;
When a predetermined number of interpolated pixel information is written in the high-speed buffer memory, the pixel information written in the high-speed buffer memory is written to the high-speed buffer memory according to the corresponding row or column and the pixel position information. A television image conversion device comprising means for transferring to a specified address of the main memory. 2. The television image according to claim 1, wherein the transfer of pixel information from the high-speed buffer memory is performed each time an amount of pixel information determined by the access time of the main memory is written to the high-speed buffer memory. conversion device. 3. The scope of the claim is that the interpolation of pixel information and the transfer of the interpolated pixel information to the high-speed buffer memory are performed only for those whose converted coordinates fall within a predetermined display coordinate range. 2. The television image conversion device according to claim 1.