JP3084866B2 - Lens distortion correction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a lens distortion correction method for correcting a geometric distortion of an input image caused by a lens system. It is suitable for use as a TV intercom that can be taken with a TV camera and checked on a monitor in a house.

[0002]

2. Description of the Related Art The image pickup range of a conventional television intercom is about 60 degrees in a horizontal angle and about 45 degrees in a vertical angle. There was a drawback that I could not see it. In view of this, it is conceivable to widen the image pickup range by widening the lens of the television camera, but in this case, there is a drawback in that the barrel is distorted by the wide-angle lens and the screen becomes difficult to see. In general, the geometric distortion of the input image caused by the lens system is not so large as to be noticeable to the human eye except for the wide-angle lens, and the lens distortion is a problem mainly in the case of an ultra-wide-angle lens or a fisheye lens. Was. As a means for correcting this kind of lens distortion, it is conceivable to prepare an address correction table in which addresses before and after correction are associated in advance, and correct the distortion by converting the address of the pixel according to this table. However, the address conversion table generally needs to have distortion correction values for all points on the screen. For example, if the screen size is 512 × 512 pixels, an 18-bit address is required, and the memory for storing the address correction table is large. As a result, the cost increases.

[0003]

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and an object of the present invention is to address a barrel-shaped lens distortion generated in an image picked up by a wide-angle lens by using an address correction table. Another object of the present invention is to reduce the capacity of an address correction table in a lens distortion correction method in which correction is performed by using an address correction table.

[0004]

FIG. 1 shows a basic configuration of the present invention. The input image signal is input to the synchronizing signal separating means 1 and the A / D converting means 2. The synchronization signal separating means 1 extracts both horizontal and vertical synchronization signals from the image signal. Other circuits operate in synchronization with this synchronization signal.
Further, the synchronizing signal separating means 1 generates a pixel sampling clock synchronized with the synchronizing signal. This clock is
The A / D converter 2 is used to sample pixels. The address generation means 3 counts pixel sampling clocks and counts addresses (X
Address), and a vertical address (Y address) is generated by counting the horizontal synchronization signal. The X and Y addresses are input to the address conversion means 4, and the converted addresses x and Y corresponding to the addresses are input.
Output y. The converted addresses x and y are input to the address correction table 5 next. The address correction table 5 outputs address correction values ΔX and ΔY indicating how many pixels are to be moved in the X and Y directions from the pixel in accordance with each pixel of the converted address x and y. The X and Y addresses from the address generation means 3 and the address correction values ΔX and ΔY from the address correction table 5 are added in the address correction means 6 to obtain correction addresses X + ΔX, Y +
ΔY is input as an address of the frame memory 7. Note that the address correction table 5 is created in advance for each lens. The address correction table associates the address before correction with the address after correction so that the coordinate system of the image before correction including the distortion becomes the coordinate system of the image without distortion. When the distortion changes smoothly, the correspondence between the addresses before and after the correction can be approximated by a general polynomial. Therefore, an appropriate polynomial is obtained and calculated.

On the other hand, the image signal is converted into digital image data by an A / D converter 2 in accordance with a pixel sampling clock. The digital image data is input to the dual-port frame memory 7 and stored in the location specified by the above-described correction addresses X + ΔX, Y + ΔY. The frame memory means generally called a dual port has two asynchronous ports, one of which is a port that can be accessed randomly, and the other port that performs serial access in units of rows. The above-described writing using the correction addresses X + ΔX and Y + ΔY uses a random access port. At the same time, the image is read from the asynchronous serial port in accordance with the X and Y addresses generated by the address generation means 3 while performing the writing while correcting. The read image data is converted into an analog image signal by the D / A conversion means 8, mixed with a synchronization signal by the synchronization signal mixing means 9, and displayed on a monitor television.

In the above description, the address correction processing is performed at the time of writing to the frame memory 7. However, the writing may be performed serially, and the reading may be performed while correcting the address from a random port. By using such a dual-port frame memory including a serial port and a random port, lens distortion can be corrected at a video rate.

[0007]

The operation of the present invention will be described below with reference to FIG. As shown in FIG. 3, the barrel-shaped distortion due to the wide-angle lens has a property that the distortion increases as the distance from the center of the screen increases, and the barrel is displaced toward the center of the screen from a correct position without distortion. To correct this, on the contrary, if the pixels are moved in the direction of driving out of the screen, the distortion is equivalently corrected. Of course, forcibly correcting an image that is extremely distorted like a fisheye lens will degrade the image quality of the corrected part, but if the distortion is not so large, it may be necessary to restore the curved straight line to the original It is possible. In this case, it is naturally possible to perform more appropriate correction by correcting in the horizontal direction (X direction) and the vertical direction (Y direction). Specifically, address correction data is written in advance in the address correction table according to the distortion characteristics of the lens to be used. That is, a certain point (X, Y)
If the coordinates after distortion correction are (X ′, Y ′), the output address when the point (X, Y) is input to this address correction table is (X ′, Y ′).

In order to have an address correction table for one entire screen, as described above, a very large capacity memory is required and the cost is high. Memory capacity is reduced. In other words, since the distortion characteristic of the lens generally has a point-symmetric characteristic with the origin at the center of the screen, the address correction processing may be performed point-symmetrically. Therefore, even if the address correction table for the entire screen is not provided, the address correction for the entire screen can be performed by repeatedly using the address correction table in a half or そ の of the area. However, for this purpose, it is necessary to change the direction in which the address correction table is accessed.

Now, consider the case where an address correction table having a capacity of 1/4 of one entire screen is used. FIG. 4 shows an example of an address correction table having a 1/4 capacity obtained by dividing the screen into upper, lower, left, and right, respectively. In this example, as shown in the figure, at the upper left of the screen, horizontal access is performed from left to right as indicated by an arrow, and vertical access is performed from top to bottom. In the upper right part of the screen, horizontal access is performed from right to left as indicated by arrows, and vertical access is performed from top to bottom. Next, in the lower left portion of the screen, horizontal access is performed from left to right as indicated by arrows, and vertical access is performed from bottom to top. Furthermore, in the lower right part of the screen, horizontal access is performed from right to left as indicated by arrows, and vertical access is performed from bottom to top. The address correction table contains a relative displacement Δ from the input addresses X and Y.
By storing X and ΔY, the capacity of the table can be reduced to ４ as compared with the case where the address correction table of the entire screen is stored. Also, the capacity of the address correction table can be reduced by storing the output addresses X + ΔX, Y + ΔY as relative values instead of storing them as absolute values.

More specifically, X,
The address conversion means at the next stage performs the above-described address conversion on the Y address, and generates x and y addresses. In the case where the screen size is 512 × 512 pixels, for example, considering the first horizontal line, if the address generation unit sequentially outputs the X and Y addresses from (1, 1) to (512, 1), the address conversion unit Received this (1,
1) to (256, 1) are output as they are, and (25)
From 7, 1), the x address is decremented conversely,
The address of (1,1) is output from (256,1).
That is, when 1 ≦ X ≦ 256, the converted horizontal address x is x = X, and when 257 ≦ X ≦ 512, the converted horizontal address x is x = 513-X. Next, regarding the conversion of the Y address by the address conversion means, 1 ≦ Y
When ≦ 256, the converted vertical address y is set to y = Y, and when 257 ≦ Y ≦ 512, the converted vertical address y is set to y = 513-Y. The address correction table is accessed using the converted addresses x and y. In the address correction table, displacement data Δ indicating how many pixels each of X and Y are to be moved relative to the input address.
X and .DELTA.Y are output, and the address correcting means adds the original addresses X and Y and the displacement data .DELTA.X and .DELTA.Y to obtain corrected addresses X + .DELTA.X and Y + .DELTA.Y, and writes the image data to the addresses in the frame memory. is there.

However, when adding the displacement data .DELTA.X and .DELTA.Y to the original addresses X and Y in the address correcting means, the displacement data .DELTA.X and .DELTA.Y are each given a minus sign at the upper left of the screen, that is, the subtraction is performed. In the lower right part of the screen, it is necessary to add the displacement data ΔX and ΔY with a plus sign. Similarly, the displacement data Δ
It is necessary to add a sign of + to X and a sign of − to displacement data ΔY. At the lower left of the screen, a sign of − is given to displacement data ΔX, and a sign of + is given to displacement data ΔY. Must be added. The reason why the signs of the displacement data ΔX and ΔY need to be switched in this way is that the distortion characteristic of the lens is always displaced toward the center of the screen from the correct position, and the distortion is corrected in the opposite direction, that is, This is because it is necessary to displace the screen toward the outside of the screen. Therefore, the address correction means has a flag indicating which of the four divided areas the current address is in, and accordingly, each of the displacement data ΔX and ΔY to be added to the original address X and Y. It is necessary to switch the sign.

As described above, the correction of the entire screen can be performed by using the address correction table having a capacity of 1/4 of the entire screen four times, thereby reducing the capacity of the address correction table. It can be reduced to 1/4.

It is apparent that the lens distortion correction method of the present invention can obtain the same effect even when using an address correction table having a half capacity obtained by dividing the screen into half vertically and horizontally. It is.

[0014]

FIG. 2 shows an embodiment of the present invention. In this embodiment, for example, in a general house or an apartment house, a lens distortion having the configuration shown in FIG. The correction means is used. The TV intercom includes a TV camera 10 equipped with a wide-angle lens and a handset 12 having a microphone 11 for collecting the voice of a visitor, a signal transmission system 13 for transmitting an image signal and a sound signal from the handset 12, and a handset. An image signal transmitted from the slave unit 12 by using a wide-angle lens, comprising a monitor television 14 for displaying an image from the slave unit 12 and a speaker 15 for outputting sound from the slave unit 12. 3 includes a barrel-shaped lens distortion as shown in FIG. 3, which is corrected using the lens distortion correcting means 17 having the configuration shown in FIG. 1 to monitor a high-quality image with little distortion. be able to. For example, the shape of an object that is about 50 cm away from the television camera, such as a visitor's face, can be corrected to make it easier to see.

[0015]

According to the present invention, there is an effect that lens distortion included in an image obtained by a lens system can be corrected, and the visibility of the image can be improved. Reduce the capacity of the address correction table by 1 /
This can be reduced to 2 or 1/4, and the cost can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of the present invention.

FIG. 2 is a block diagram showing a configuration of one embodiment of the present invention.

FIG. 3 is an explanatory diagram illustrating a state of lens distortion caused by a wide-angle lens.

FIG. 4 is an explanatory diagram showing an access direction of an address correction table used in the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Synchronization signal separation means 2 A / D conversion means 3 Address generation means 4 Address conversion means 5 Address correction table 6 Address correction means 7 Frame memory 8 D / A conversion means 9 Synchronization signal mixing means

Continuation of the front page (56) References JP-A-5-30512 (JP, A) JP-A-4-61570 (JP, A) JP-A-62-81180 (JP, A) (58) Fields studied (Int .Cl. ⁷ , DB name) H04N 5/232 H04N 7/18

Claims (3)

(57) [Claims] 1. A synchronizing signal separating unit for separating a synchronizing signal from an image signal of a television camera using a lens system having a point-symmetric distortion about an optical axis, and an A / D converting unit for digitizing the image signal Address generation means for generating a screen address from a synchronization signal; and a screen address before distortion correction and a screen address after distortion correction only in an area divided in at least one of a horizontal direction and a vertical direction passing through the center of the screen. An address correction table for associating the screen address generated by the address generation means with an address in the area of the address correction table, and an image at the screen address before distortion correction based on the output of the address correction table. A frame memory in which the digital value of the signal is written to the screen address after distortion correction, A lens comprising D / A conversion means for D / A conversion of a digital value read from a frame memory, and synchronization signal mixing means for mixing a synchronization signal with an output of the D / A conversion means. Distortion correction method. 2. The address correction table stores a difference between a screen address before distortion correction and a screen address after distortion correction as a relative displacement, and stores the screen address before distortion correction in a plurality of divided areas. 2. The lens distortion correction method according to claim 1, further comprising address correction means for adding a relative displacement having a different sign to a screen address before distortion correction in accordance with which area the pixel belongs to. 3. A child device having a television camera equipped with a wide-angle lens and a microphone for collecting voice of a visitor,
Lens distortion correction in a signal transmission system for transmitting an image signal and a sound signal from a child device, and a television interphone including a monitor television for displaying an image from the child device and a parent device having a speaker for outputting audio from the child device. 3. The lens distortion correction method according to claim 1, wherein the means is provided in the parent device.