KR970011541B1 - Auto-focusing control method for video-cameras - Google Patents

Abstract

A method for automatically controlling a focus of a video camera is disclosed. First, differences between pixels are calculated to provide a set of gradient vectors. And then, each of the gradient vectors is changed to a double gradient vector. By accumulating the double gradient vector, the motion of the object is determined to provide a similarity. The similarity is compared to a reference value to determine whether the motion exits or not.

Description

Auto focusing method of video camera

FIG. 1 is a view for explaining a process of making a gradient vector into a double angle vector and summing in the present invention.

The present invention relates to a method for automatically adjusting the focus of a video camera, and more particularly to a passive type of auto focus control method capable of precisely automatic focus control without a common focus control function even for a moving object. .

In general, two types of automatic focusing methods for video cameras have been put to practical use.

One of them is called an active type, and installs an infrared or ultrasonic transmitting and receiving device on one side of the video camera, analyzes the infrared or ultrasonic wave transmitted and received, and measures the distance, and uses the measured distance to focus lens. A method of adjusting the focus by operating the driving means, and the other method is a passive method, the method of adjusting the focus by moving the focus lens to the position where the contrast is the largest using the contrast signal among the image signal from the image pickup device to be.

This passive method does not need to transmit infrared or ultrasonic waves of sufficient intensity as in the active method, so it consumes less power and analyzes the video output signal, so there is no parallax, when the distance of the subject is far or inclined, and when the video camera It is convenient because the normal focusing function is made even when there is glass in front, and there are many advantages such as miniaturization because there is no need to install separate parts such as infrared or ultrasonic transceiver. In recent years, many of these passive methods have been adopted.

This passive technique is described in Korean Patent Application Nos. 91-2807 and 91-6109 of the present invention, which were invented by the inventor of the present invention, and US Patent No. 5,200,828 (issued April 6, 1993) corresponding thereto. Well explained. In addition, related technologies are described in the applicant's prior patent applications No. 1992-22762, 1994 patent application 11069 and 1994 patent application 12938.

In patent application No. 92-22762, it is determined that the camera focuses on the front or back of the subject to be photographed, so that it is not in the confocal state, and the moving direction of the focus lens is determined as the confocal direction. This section describes how to enable focusing while following automatically when the instrument moves. This method is a video camera automatic focus control method that automatically adjusts the focus by using the video signal extracted from the video camera image. (1) The screen is divided into a plurality of blocks and the video signal is processed to reduce the contrast in each block. Obtaining the indicated focus value, (2) comparing and determining the change in the focus value to determine whether it is in a blur state, and if it is a blur, executes a blur adjustment step, and if it is not a blur, executes a general confocal adjustment step. Including, and the refocusing step; (3) determining whether the change in the focus value is smaller than the predetermined value or becomes "0" to move the lens to the position of the lens where the focus value is maximum, and (4) the lens motor driving direction and the driving motor speed. Driving the focus control drive motor; The blur adjustment step includes (5) calculating deviations of the block focus values obtained from a plurality of blocks; and (6) if the deviation is less than a predetermined threshold, the focus control drive motor is set so that the lens focuses farther than the present. Determining the driving direction of the focus control driving motor so that the focus of the lens is closer to the current focus point if the deviation is greater than the predetermined threshold, and (7) setting the driving control motor in the predetermined direction. It is a method consisting of driving steps. In the first step, the deviation between the block focus value of each block and the average focus value of the entire screen is obtained, and the portion smaller than the threshold value is the first zone, and the portion larger than the threshold value is the second zone. Divide the screen and adjust the focus value of the first zone to the maximum.

This application includes an A / D converter for sampling a video signal of a video camera at predetermined intervals to create a digital parallel signal, a clock generator for dividing the screen into predetermined regions using horizontal and vertical synchronization signals, and generating a sampling clock; A focus value generation circuit that receives a digital parallel signal of the A / D converter and a sampling clock of the clock generator and generates a focus value, and an automatic control unit that controls the focus of the camera by driving the lens motor of the video camera using the focus value. An automatic focus control apparatus for a video camera comprising: a difference signal generation circuit for generating a difference signal representing a contrast by obtaining a difference between the digital parallel signals and receiving a clock of the clock generator; A cumulative circuit which accumulates the difference signals by a predetermined section and makes a focus value. It is configured. The difference signal generation circuit includes a first latch 72 that receives a digital parallel signal of the A / D conversion unit, a second latch 74 connected to the first latch, an output of the first latch, and an output of the second latch. Or a difference signal generation circuit for receiving the digital parallel signal of the A / D converter; n, a plurality of latches 82 connected in series and an output of the first latch 82-1 from the plurality of latches on one input, and the second, third,... Comprising a ROM (86) for receiving the output of the n latches at the other pressure through the first MUX (84) to generate a plurality of period difference signal, furthermore, the difference signal generation circuit receives the digital parallel signal of the A / D conversion section A ROM for receiving a first latch 92, a horizontal scan line memory 94 connected to the first latch, an output of the first latch, and an output of the horizontal scan line memory, and generating a first scan bow difference signal. What is necessary is just to comprise (96). Another difference signal generation circuit includes a first latch 102 for receiving a digital parallel signal of an A / D converter, a plurality of horizontal scan line memories 104 connected in series with the first latch, and an output of the first latch. Is received from one input, and the outputs of the plurality of horizontal scan line memories are received from the other input through the second MUX 108, and consists of a ROM 106 for generating a multiple scan bow difference signal. The accumulator circuit is an adder 120 for receiving and adding difference signals of the difference signal generating circuit, and a third MUX for connecting the output of the adder and the accumulated value of the data bus to the output by the accumulator / * exchange signal 112. (121), the latch 122 which latches the output of the third MUX and connects to the input of the adder, and the output of the adder and the input of the buffer 124 are connected to the output by the cumulative / * exchange signal 112. A fourth MUX 125 for latching, a latch 2 123 for latching the output of the fourth MUX and connecting it to the input of the buffer, and a buffer 124 for connecting the output of the latch 2 to the data bus by an output transmission signal. It can be made, including, allowing the accumulation of a plurality of sets of separated time intervals independently of each other.

Patent Application No. 94-11069 describes a method for automatically adjusting the focus of a video camera, in particular, a passive type of auto focus that enables automatic focus control even for a high-luminance subject such as a light source. It is described in the adjustment method. This method uses the auto focus control device described in patent application No. 92-22762, 1) divides the screen into a plurality of blocks, accumulates and obtains luminance signals from the brightness of each block and the brightness of the entire screen, and Identify the positions of the bright blocks that are brighter than the average brightness of the screen, 3) analyze the connections of the bright blocks, and 4) if the bright blocks are spread across the screen, the total number of bright blocks is greater than the total number of blocks. If the ratio is less than a certain ratio, instead of calculating the focus value of the bright block, only the focus value of the remaining blocks is calculated by the conventional method, and the total focus value for automatic adjustment is set. Otherwise, the inverse of the sum of the brightness of the bright block The focus value of the dark block other than the light block is calculated by the conventional method, and then the total focus is obtained by adding the light block focus value and the dark block focus value together. To the value. In the fourth step, the method of adding the light block focus value and the dark block focus value is performed by multiplying the light block focus value by the area ratio of the light block to the entire screen, and the dark block focus value by the area ratio of the dark block. In the first step, it is recommended to divide the entire screen into 64 blocks, and in the third step, the connection state of the bright blocks is: (a) counting the number of blocks in which the bright blocks are continuous, and (b) Calculate the size of the largest block of bright blocks, and (c) use the count of the number of blocks and the size of the largest block of bright blocks to see if the bright blocks are spread or clustered on the screen.

In addition, a method of obtaining a focus value for video camera automatic focus control that automatically adjusts focus by using an image signal extracted from an image of a video camera is described.Average luminance signal values corresponding to two or more adjacent pixels are averaged to at least one pixel interval. Focusing method of generating a focus value by subtracting each other and accumulating each other as a focus value, and focusing on a video camera that averages the luminance signal values of two adjacent pixels and subtracts the average value by two or more pixel intervals and accumulates the accumulated value as a focus value. The method of generating values is also described.

Patent Application No. 94-12938 describes a method for automatically adjusting the focus of a video camera, especially for a subject that emits high brightness, such as a light source, as well as when the light source is in a blinking state. A passive auto focusing method is described which enables the adjustment. In this method, 1) the screen is divided into a number of blocks, each block is divided, the brightness of each block and the brightness of the entire screen are accumulated and obtained by the luminance signal, and 2) a bright block that is brighter than the average brightness of the entire screen. 3) Analyze the connected state of the bright blocks, and 4)-(a) If the bright blocks are spread on the screen and the total number of the bright blocks is less than a certain ratio compared to the total number of blocks, 4)-(b) Other than the above (a): (b-1) Brightness of the bright block is calculated by calculating only the focus value of only the remaining blocks without calculating the focus value. The inverse of the sum is set as the focus value Vb of the bright block, (b-2) the variation period of the focus value Vb of the bright block is calculated while storing the focus value Vb of the bright block of each field, and (b-3) the variation Largest bright block focus during cycle Calculate Vb ', (b-4) calculate the focal value Vd of the dark block other than the light block, and then (b-5) add the bright block focal value Vb' and the dark block focal value Vd to make the total focal value Vf. The method of adding the bright block focus value and the dark block focus value in the above steps 4)-(b-5) is multiplied by the area ratio of the bright block to the entire screen, and the focus value of the dark block is multiplied. Multiply the area ratios of the dark blocks to sum them to obtain the total focus value, and divide the entire screen into 64 blocks in the first step. The method of claim 1, wherein in the third step, the connection state of the bright blocks includes (a) counting the number of blocks in which the bright blocks are continuous, (b) calculating the size of the largest chunk of the bright blocks, and The number of blocks and the size of the largest chunk of bright blocks can be used to determine whether the bright blocks are spread or clustered on the screen. In step 4)-(b), the period of changing the focus value of the bright block is longer than the blinking period of the normal Christmas bulb or neon sign, and finds the smallest bright block focus value within the period, and 4)-( In step b-1), the sum of the brightnesses of the bright blocks is subtracted from the constant number as the focal value Vb of the bright blocks.

Further, by further improving the auto focus control device, the screen is divided into a plurality of blocks, the brightness signal generating means for accumulating the brightness signals of each block to obtain the brightness of each block and the brightness of the entire screen, and from the brightness signal generating means. Bright block positioning means for checking the position of the bright block that is brighter than the average brightness of the obtained full screen, and analyzing the connected state of the bright block, and if the bright block is spread on the screen, the total number of bright blocks Comparative analysis means for judging whether or not it is less than a certain ratio compared to the number, a first focus value generating means for calculating the inverse of the sum of the brightness of the bright blocks, and a focus value for the dark blocks other than the bright blocks. The value Vd is the sum of the second focal value generating means for calculating, the focal value Vb of the light block and the focal value Vd of the dark block, and the total focal value. And focusing means for focusing on the basis of the total focusing value, thereby enabling automatic focusing even when the light source is included in the subject. The second focal value generating means averages the luminance signal values corresponding to two or more adjacent pixels, subtracts and accumulates each other at one or more pixel intervals, and sets the focal value, while maintaining the bright value Vb of the bright block of each field. A minimum calculation means for calculating the fluctuation period of the focus value Vb of the block and obtaining the smallest bright block focal value Vb 'during this fluctuation period, the focus control means further comprising the focal value Vb' and the darkest value of the smallest bright block. The focus value Vd of the block is added to make the total focus value, and the focus control is performed based on the total focus value to enable automatic focus adjustment even when the flashing light source is included in the subject.

The invention of the present application is a further development and use of the techniques of the above-mentioned prior applications, which is hereby incorporated by reference in its entirety. If all of these specifications are re-written in this specification, the amount of this specification will be excessive, so that the contents of this specification are omitted, but the contents are cited and used as necessary.

Even in the above-described inventions of the present application, the blur phenomenon caused by the blur and the sudden movement caused by the motion blur and the focus were not solved well.

There are two types of blur (blur) that degrades an image, which are basically caused by movement and blur due to out of focus. Both blurring results in attenuation of high frequency components in the spatial frequency domain, but the human visual impact is quite different. The motion blur is due to the finite shutter speed of the camera and the sensitivity of the image pickup device.

The present invention is to make it possible to perform the operation of automatically focusing to distinguish from the blur caused by the motion blur when there is such a motion blur.

This method automatically adjusts the focus of the video camera.

1) A gradient vector is obtained by obtaining a difference signal of neighboring pixelizations from an arbitrary pixel in a video image signal,

2) Convert this gradient vector into a double angle vector,

3) Accumulate these double angle vectors to find the equivalence to judge the movement of the subject.

4) Compare the size of the same value with the standard value to determine whether the subject is moving and to perform the focus adjustment operation.

Here, in step 3), the display apparatus may be divided into a plurality of zones of the screen, and the equal value may be obtained for each divided zone, and the presence or absence of a moving subject may be determined for each zone in comparison with the reference value in step 4).

Also, in the second step, the method of converting the gradient vector into a double angle vector is assuming that the values of the x and y axes of the gradient vector are x and y, respectively.

To save and

We find and calculate the x and y values of the double angle vector.

In a simpler way

1) From the video image signal, obtain a difference signal with a pixel neighboring a pixel in a vertical direction, and obtain a vertical focus value therefrom, and obtain a difference signal from a pixel neighboring with a pixel in a horizontal direction. Find the horizontal focus value,

2) The vertical focus value and the horizontal focus value in all fields of the video signal are compared with the vertical focus value and the horizontal focus value of the current field, and the variation thereof is obtained.

3) If the focus value in the same direction as the movement direction is decreased but the focus value in the direction perpendicular to it is not affected, it is determined that the movement is in the vertical direction when the focus value suddenly decreases for all the fields. It is also possible to determine the movement of the focusing operation.

In more detail with respect to the present invention, the three-dimensional object is projected on the two-dimensional imaging surface inside the imager. At this time, each point of the three-dimensional object surface corresponds to each point of the two-dimensional imaging surface. The brightness of the point corresponding to each point on the image pickup surface is accumulated for the opening time of the shutter and output as an electric signal for each point, that is, the pixel. If the subject moves while the shutter is open, each point on the subject surface does not correspond to a point on the imaging surface, and the corresponding point forms a trajectory. Therefore, the image seems to flow. To prevent motion blur, the shutter open time is infinitely small, but this is not practical. When the direction of motion is called the x-axis, the motion blur has an effect of attenuating high frequency components in the x-axis direction (having a flat characteristic in the vertical direction of motion).

Considering this filter effect as one dimension along the x-axis and comparing the characteristics of the spatial domain with those of the spatial frequency domain, it is directional that the motion blur can be distinguished from the blur when it is not in focus. have.

If the focus is blurred due to misalignment, the high frequency component of the image decreases as the blur is diffused because the dot on the subject surface does not correspond to a point on the imaging surface as described above. However, in the case of motion blur, only the image of the part with a moving subject is blurred and the image of the other part is in the same state as it is in focus.

In case of blur due to poor focus, image quality is reduced, but in case of motion blur, it is advantageous for humans to see moving images. Because if any point on the image plane receives flux of light, which is a one-dimensional signal with respect to time, if the shutter speed is infinitely fast and the reduction of the image pickup device is infinite, then the output of the image pickup device is the result of sampling (sampling) the input. In this case, aliasing occurs over time. However, if the shutter speed is finite, this results in sampling the signal averaged over a certain interval. In this case, since the average acts as a low pass filter, it has an antialiasing effect, which is more advantageous for a human being to see a moving image.

This results in the same reduction in focus, but in the case of motion blur, there is no need to determine that the focus is blurred. Therefore, it is necessary to distinguish these two cases and make an appropriate response. It is possible to use the method of investigating the correlation between image blocks in different fields to detect motion. It is almost impossible to apply to. Therefore, the present invention is invented the following methods.

The simplest way is to use a variation of the vertical and horizontal focus values. This takes advantage of the fact that, as described above, the focus value in the same direction as the direction of motion decreases but the focus value in the direction perpendicular to it is not affected. For example, when the horizontal focus value suddenly decreases but the vertical focus value is maintained at a similar level, it can be seen that the horizontal motion occurs. On the contrary, when only the vertical direction suddenly decreases the focus value for the previous fields, it may be determined that the vertical motion. This method can be implemented with very simple hardware, and considering that the general direction of movement of the general subject is mostly horizontal, it is judged that it is worth applying to the experimental results without any big problems in practical application.

The disadvantage of the above method is when the motion between the horizontal and vertical directions occurs. In this case, since the focus values in both directions will decrease, there is a high possibility that the subject will be wrongly determined to have changed. In addition, the accuracy of the judgment is not practically large, but it is inferior in accuracy. Therefore, the following method is suggested.

This method also estimates motion using the fact that motion blur only blurs in the direction of motion.

If the image has contrast in only one direction, the directions of the gradient vectors (the result obtained by subtracting adjacent pixel values) of the image all point to the contrasted direction. Depending on the degree of directionality in the contrast, the gradient vectors may be aligned or distributed. If you create a certain amount of offset between different directions when accumulating with contrast direction information, you can use it as a criterion (value) to determine if the contrast is directional in one area.

Among the findings used to determine the moving value, the most useful value is the double angle vector that doubles the angle while maintaining the size of the gradient vector. FIG. 1 shows a process in which gradient vectors in opposite directions are easily accumulated in the same direction by using a single angle vector. In other words, when the vectors Ⅳ1Ⅰ∠α having an angle of α with respect to the horizontal axis and the vectors Ⅳ2Ⅰ∠180 + α having an angle of α + 180 degrees with respect to the horizontal direction are combined, respectively, when the double angle vector is made, the V1 vector becomes IV1Ⅰ∠2α The V2 vector becomes IV2I∠2α and is in the same direction. Thus, when the single-angle vectors are combined, it becomes the V3 vector of IV1 + IV2I∠2α. Thus, when V1 and V2 are added as double angle vectors, the result is a V3 vector, which is illustrated in FIG. If we can allow the magnitude of the gradient vector to be squared, then an operation that doubles the angle is possible with three multiplications. The mathematical solution is as follows.

《How to double the angle of Vector》

Some vector (x, y) can be expressed as follows using angle θ between size r and the positive x-axis. If the vector doubling the angle θ of this vector is (x ', y'), x 'and y' are represented as follows.

So if we take a vector (x, y) twice that angle

Therefore, if the (x, y) vector is calculated by (x ', y') and summed, the sum of double angle vectors obtained by doubling the angle of the (x, y) vector is summed.

As described above, by using the method of the present invention, it is possible to appropriately perform the adjustment operation without causing confusion of the automatic focusing operation when photographing a moving object.

Claims (4)

A method of automatically adjusting the focus of a video camera, comprising: 1) obtaining a gradient vector by obtaining a difference signal from an arbitrary pixel to a neighboring pixel in a video image signal, 2) converting the gradient vector into a double angle vector, 3) accumulate the double-angle vector to determine the equivalence of judging whether the subject is moving, and 4) compare the size of the equivalence with the reference value to determine whether the subject is moving and to perform the focus adjustment operation. Focus auto adjustment method. The method of claim 1, wherein the screen is divided into a plurality of zones in step 3), and the equality is obtained for each divided zone, and the presence or absence of a moving subject is present in each zone compared to the reference value in step 4). Automatic focusing method of the video camera characterized by judging. The method of claim 1, wherein the method for converting the gradient vector into a single-angle vector in the second step is assuming that the values for the x and y axes of the gradient vector are x and y, respectively. To save and The automatic focusing method of a video camera, characterized in that to obtain the x value and the y value of the double angle vector. A method of automatically adjusting the focus of a video camera, the method comprising: 1) obtaining a difference signal from an arbitrary pixel to a neighboring pixel in a vertical direction in a video image signal, and obtaining a vertical focus value from the pixel; Obtain the difference signal with neighboring pixels in the direction, and obtain the horizontal focus value from this, and 2) the vertical focus value and the horizontal focus value in all fields of the video signal and the vertical focus value and the horizontal focus value in the current field. 3) When the focus value in the same direction as the movement direction decreases but the focus value in the direction perpendicular to the direction is not received, the focus value suddenly decreases for all the fields. It is determined that the movement in the vertical direction, to determine whether the movement of the subject to focus the video How to automatically adjust the focus of the camera.