JPS5819187B2 - Shading correction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a shape distortion correction method in film transport, and more specifically, to a method for correcting shape distortion in film transport.
The present invention relates to a method for repairing shape distortion caused by an FSS tube or a television camera and optical system when image information is obtained from a subject such as a film or an opaque using an apparatus or a television camera.

Various methods for correcting shape distortion have been proposed and attempted in the past.

For example, there is a method of performing approximate correction by transforming the scanning signal using a time function.

Although the correction device is simple and inexpensive, it is not possible to achieve sufficient approximation, and it is not possible to obtain correction results that are suitable for practical use.

Further, as illustrated in FIG. 1, the light beam from the FF 5t'1 is split by a half mirror 2, one of the light beams is made to enter a phototube 5 via an optical system 3 and a subject 4, and its output is sent to a computing unit 6. The other light beam is made to enter directly into another phototube 8 via another optical system 7, and its output is inputted to the arithmetic unit 6, and then divided by the arithmetic unit 6 and converted into image information. There is a method to output it.

This method of splitting the light rays from the light source and comparing the signal that passes through the subject with the signal that does not pass through the subject to remove the shape distortion cannot be completely corrected because the shape distortion caused by the two optical systems 3 and 7 is not the same. No effect can be obtained, and two sets of optical systems are required, making the device complicated and expensive.

Another shape distortion correction method uses memory.

This memory method, as illustrated in Figure 2, is based on the FFS tube 1
The pixel information from the phototube 13 which receives the light beam from 0 through the optical system 11 and the subject 12 is input to the arithmetic unit 15 by the switch 14, and when the subject 12 is removed, it is stored in the memory 16 in advance via the switch 14. The stored comparison pixel information is input to the arithmetic unit 15 and output as image information.

Although this memory method is excellent in principle, as will be explained below, it requires a memory with a large storage capacity and requires division, so the device tends to be large and expensive.

FIG. 3 is a diagram for explaining the principle of a memory-based shading correction method.

If the light source 20 is an FSS tube or a television camera, it is an ideal light source with its illumination and no shading, and L(x
.

Suppose that a beam of energy y) is emitted.

Here, X and y indicate the coordinates of each pixel on the screen.

It is assumed that the next shading distortion source 21 includes all the shading distortions caused by the FSS tube or the television camera, the optical system, etc., and is multiplied by S(x, y) as a result.

The subject 22 is a type of film that uses transmitted light and an opaque or standard screen that uses reflected light (hereinafter referred to as "opaque, etc.").

Collectively called.

) may be used, and is shown as a film for convenience of illustration.

Assume that the loss rate of this object 22 is D (x 7 y ).

The last photoelectric conversion unit 23 such as a phototube or a television camera is
Assuming ideal photoelectric conversion, the output is E(x,
y) -.

Now, if the output of the photoelectric conversion unit 23 when the subject 22 is placed is Es(x, y), it can be expressed as follows.

Es(x,y)=L-8-D(x,y)"-(1)
.

The output Eo (x, y) of the photoelectric conversion unit 23 when the subject 22 is removed (in the case of film, when the film is removed; in the case of opaque, etc., it is replaced with white standard opaque, etc.) is as follows. can be expressed as

Eo (x, y) = L-8 (x, y)...-
(2) Therefore, if we divide equation (1) by equation (2), then Es
(X, y)-L-8-D(X, y) Eo(x, y) L-8, (x, y)=D(x, y
) ・・・・・・・・・・・・(3) ・It becomes.

This is subject information from which the influence of shading has been removed.

In this way, the memory method can theoretically perform complete shading correction.

However, the memory must store comparison pixel information for each unit pixel corresponding to one frame, and a large capacity memory is required.

Furthermore, in order to store the data of Eo (x, y) with high precision, the number of memory bits must be further increased.

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a memory-based shading correction method that minimizes the required memory storage capacity.

Furthermore, it is an object of the present invention to provide a memory-based shading correction method that can obtain the necessary correction accuracy with a minimum number of memory bits.

Furthermore, the present invention avoids the use of complicated and expensive division circuits and allows the use of simple and inexpensive multiplication circuits.

The present invention will be explained below.

FIG. 4 is a basic configuration diagram of an apparatus for carrying out the method of the present invention, and for convenience, it is drawn for monochrome use.

The switch 30 corresponds to the switch 14 in FIG. 2, and when a subject is placed, the pixel information L-8-D (X, y) from the photoelectric exchange unit is input to one side of the multiplication circuit 31. When there is no subject, comparison pixel information L-8 (x, y) from the photoelectric conversion section is sent to the sample and hold circuit 32.

The output of the sample hold circuit 32 is sent to an A/D converter 33.
and converted into digital data N(x,y).

The digital data N(x, y) is sent to the arithmetic path 34, and as described later, the arithmetic mean N(X, Y) within the small interval is taken, and 1/N(X, Y) calculation is performed. The data is input to the memory 35 as m-bit data M(X, y) with only the necessary digits below the decimal point.

The memory 35 stores the input data M(X, Y) as comparison data at an address designated by an address signal from an address control circuit that receives the horizontal synchronization signal HD and the vertical synchronization signal VD.

When the object is placed, the switch 30 is turned upward and the pixel information L-8-D(X,y) is connected to one input of the multiplier.

On the other hand, data M (X, Y) from the memory 35 specified by the address signal from the address control circuit 36 is read out, sent to the D/A converter 37, and subjected to D/A conversion.

The output of the D/A converter 37 is sent to the restoration circuit 38, where a voltage corresponding to a constant is applied to obtain 1/N(x, y
) is restored and sent to the other input of the multiplier circuit 31.

The multiplication circuit performs multiplication shown in the following equation and outputs desired image information D'(x, y).

L-8-D(X, y)×□ゝD'(x, y)...(4
)N(X, Y) As mentioned above, a large capacity memory is required to completely store one frame of data.

Focusing on the fact that shedding changes continuously and gradually, we divided the screen into small sections of appropriate size and assumed that the comparison data was almost uniform within each small section. By digitizing and storing the average value or representative value of the ratio data within the small section, the required storage capacity can be further reduced.

Therefore, as shown in Figure 5, the screen of A horizontally and vertically is divided into p sections horizontally and 9 sections vertically, subdivided into a/pxb/q subdivisions. Comparison data L-8(x
, y) of the average value or representative value of N(X,
Y).

Then, the comparison digital data N(x,=y) at any point (x, y) within the subdivision is N(x, y)2N
(X, Y) ・・・・・・・・・・・・(5) It is considered as shi.

For this purpose, in the example of this configuration diagram, the sample and hold circuit 32 samples and holds pixel information at the right end or left end of each small section for each horizontal scan, outputs it to the A/D converter 33, and outputs it to the A/D converter 33. In the arithmetic unit that receives the output of the D converter, the arithmetic mean N(X, Y) for each subdivision is determined,
And 17N (X.

An approximate calculation of Y) is performed, and m-bit data M(X, Y) with only the necessary digits below the decimal point is sent to the memory and stored.

Then, when the subject is placed, the multiplier 31 is (4
) and outputs desired image information D'(x, y).

L:5-D(X,y)×□D'(x,y)N(X,Y
) As can be seen from the above equation (3), division is necessary to obtain image information without shading, but the division circuit is complicated and expensive, so as mentioned above, in the present invention, The comparison data is digitally converted, the reciprocal of the approximate calculation is stored in memory, and then multiplied by the image information.

Therefore, use a multiplication circuit instead of a division circuit.

Assuming that the shedding normally does not exceed 50 inches and that L-8(x,y)>0.5, the following equation holds true.

■≦□ku2 ・・・－・・・・・・・・・・・・(7
)N(X, Y) Therefore, the number above the decimal point is always 1, so as mentioned above, the memory stores the m-bit digital data portion M( X, Y) only.

M(X・Y)−N(X,Y)′=a・2′+b・2−2+−+m・2”−(
8) When obtaining image information, M(
By adding 1 to X, Y), 1/N(X, Y) can be restored, which is converted into pixel information E(x, y), that is, L・5−D(x',
y), the required shading-corrected pixel information D'(x, y) can be obtained.

As the shedding becomes smaller, M(X, Y) approaches O, and a, b, and c in equation (8) become O in this order.

Therefore, in such a case, the same shading correction accuracy can be obtained even if the number of memory bits is reduced.

For example, if the shedding is less than 25%, M(X,
Y)≧□-1=0.333 0.75 , ”, M(X, Y)>02-1+b ・2 ”+c ・
2 ”・・・・・・m・2-m ・・・・・・・・・
...(9).

Therefore, bit 2-1 is always 0. It is sufficient to store only data of 2-2 or less.

The error rate ε in the above approximate calculation is as follows.

,=D(x,y)-D'(xyy)×L-8(x,y)
D(x, y) -2m ・・・・・・・・・・・・(10) However, D'(x, y): Image information D(x
, y): Image information not based on approximate calculations If we look at the case where the shading coefficient is 40 and m=4, ε=3.7% or less.

This value of ε is satisfactory for the shading correction accuracy, and it can be said that sufficient shading correction can be performed with a 4-bit memory.

When the shedding is small and close to 0, the following simple approximation method can be adopted, and the device can be made inexpensive.

Equation (8) is transformed as follows.

1 M(X・Y)-N(X, Y) ' 1-N(X, Y) ・・・・・・－・・・・・・αη
(N(X, Y) Since the shedding is close to 0, N(X.

Y) has a value close to 1, and therefore can be approximated as follows.

M(X, Y), 1-N(X, Y) --------・
(12) Since the logic circuit for this operation is simpler than the logic circuit for the operation in equation (8), it is advantageous to use this method when the shedding is small.

FIG. 6 shows an embodiment in which the present invention is applied to a continuous running FSS type film transport device, and only the configuration of the shading correction circuit portion is shown.

In this type of film transport device, image signals are obtained by scanning by moving the raster position on the FSS tube in response to the continuously running film, but the shading in conventional FSS for still images is difficult. In this case, it is perceived as flicker, so it is necessary to perform strict shading correction.

Pixel information from each of the three primary color phototubes is sent to correction circuits 40, 41, and 42, respectively.

Each correction circuit has constant addition and multiplication functions.

Comparison pixel information when the subject is excluded is switched by the input switching circuit 43 and sequentially sent to the sample and hold circuit 4.
4, and is digitally converted by an A/D converter 45.

An arithmetic circuit that receives the digitized comparison pixel information calculates the reciprocal of the input data.

Normally, it is not possible to adjust the siedengu to below 25 units.

Since it is easy, the above equation (9) can be used, and 2-
It is sufficient to output four digits of digital data of a reciprocal number of two terms or less to the memory 47.

The memory 47 receives input data according to an address signal from a deflection position detection circuit 48 that receives a horizontal synchronization signal HD and a vertical synchronization signal VD.

The data is stored separately for R, G, and B.

When the object is placed, comparison pixel information is read out from the memory 47 according to the address signal from the deflection position detection circuit 48, and converted into converters 49 and 5 separately for R, G, and B.
Sent to 0.51.

The outputs of the D/A converters 49, 50, 51 are connected to corresponding correction circuits 40, 41, 42.

Then, these correction circuits 40, 41, and 42 add 1 to the output of the D/A converter, and output the sum together with the pixel information from each phototube.

In this example, it is divided into 16 horizontally and 32 vertically, and 5
Comparison pixel information was divided into 12 small sections and stored in memory.

As a result, in this example, a sufficiently practical shading correction effect was obtained.

As described above, according to the present invention, memory-based shading correction can be performed using a device with a small storage capacity and a simple configuration.

Furthermore, the necessary correction accuracy can be obtained with the minimum number of bits.

Therefore, an apparatus without unnecessary redundancy can be provided at low cost in accordance with necessary correction accuracy conditions.

Furthermore, since R, G, and B can be corrected separately, color balance can be easily achieved in the case of color.

Furthermore, since shading correction can be performed regardless of the type of subject, it has a wide range of applications.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram showing one of the conventional shading correction methods, FIG. 2 is a schematic diagram of an apparatus for implementing the memory-based shading correction method, and FIG. 3 is the principle of the memory-based shading correction method. An illustrative diagram, FIG. 4 is a diagram illustrating the principle of the method of the present invention, FIG. 5 is a divided screen diagram, and
FIG. 6 is a block diagram of a shading correction device embodying the present invention.

Claims (1)

[Claims] 1. When the subject is a type that uses transmitted light, the subject is removed; when the subject is a type J that uses reflected light, it is replaced with a standard subject, and a television camera or FSS device is used. The screen obtained by is divided into several subsections, comparison pixel information representative of each section is stored in memory in advance, and the pixel information when the subject is placed is compared with the comparison pixel information. In the method of obtaining image information that has undergone shape correction, when storing comparison pixel information in memory, digitize the comparison pixel information and find its reciprocal,
Of the data below the decimal point of the reciprocal number, only the data portion 2 sufficient to obtain the necessary shape and distortion correction accuracy is stored in memory as comparison pixel information, and when performing shape and distortion correction of the image information of the subject, it is used as the pixel information of the subject. A shape-and-shape correction method characterized by adding a constant to the comparison pixel information stored in a memory and multiplying it by a restored inverse number. =2 When the shape deviation is small, only the data part sufficient to obtain the necessary shape deviation correction accuracy is stored in the memory as comparison pixel information, excluding the upper digit or multiple digits of the data below the decimal point of the reciprocal number. A shape-and-shape correction method according to claim 1J, characterized in that: 3. When the shape deviation is close to O, a value obtained by subtracting this comparison pixel information from 1 is stored in a memory as comparison pixel information below the decimal point of the reciprocal number. Method.