JPH07123364A - Image shake correcting device - Google Patents

Abstract

PURPOSE:To obtain an excellent pictures at all times by preventing an error from being occurred in the operation of correcting the shake of the images by the presence of superimposed characters in the pictures. CONSTITUTION:A superimposed character detection circuit 6 for detecting the signal components of characters superimposed on reproducing video signals s1 is provided, the detection signals s6 of the characters outputted from the superimposed character detection circuit 6 are supplied to a microcomputer 4, a motion vector in a detection area where the character components are detected in the superimposed character detection circuit 6 is excluded from the motion vectors s2 in the respective detection areas detected by a motion vector detection circuit 3 by the control of the microcomputer 4 and the shake is corrected. Thus, influence by the motion of the characters to be the cause of generating the error at the time of calculating the shake amount of the pictures is not received, only the shake amount of the pictures at the time of actual photographing is accurately calculated and the shake is corrected without any malfunction at all times.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image shake correction device, and more particularly, to a device suitable for use in a video player or an image receiver having a function of correcting image shake caused by camera shake during shooting. Is.

[0002]

2. Description of the Related Art Japanese Patent Laid-Open No. 63-16637 discloses an example of a video player having a function (anti-vibration function) for correcting image shake caused by camera shake during shooting.
There is one described in Japanese Patent No. 0. An image blur correction device to which this video player is applied is configured as shown in FIG.

In FIG. 9, 1 is a video player, 2 is an image memory, 3 is a motion vector detecting circuit, 5 is an image enlarging circuit, and 30 is a microcomputer.
Further, s1 is a video signal reproduced by the video reproducing device 1, s2 is a motion vector detected by the motion vector detecting circuit 3, s30 is an address signal representing the read position of the video signal stored in the image memory 2, and s4 Is digital data indicating an image enlargement ratio, and s5 is a video signal output from the image shake correction apparatus.

In the image blur correction device thus constructed, the video signal s1 reproduced by the video reproducer 1 is reproduced.
Are temporarily stored in the image memory 2. On the other hand, the video signal s1 reproduced by the video player 1 is also input to the motion vector detection circuit 3, and the motion vector detection circuit 3 detects the motion vector s2, which is the information of the motion speed of the image, from the video signal s1. To be done.

The motion vector s thus detected
2 is given to the microcomputer 30. Then, the microcomputer 30 determines the address signal s3 indicating the read position of the video signal stored in the image memory 2 based on the motion vector s2.

The video signal stored in the image memory 2 is read according to the address signal s3. As a result, the video signal is read from the image memory 2 so that the image is vertically and horizontally shifted so that the image shake caused by the camera shake during shooting is eliminated.

Next, the video signal thus read out is
It is input to the image enlarging circuit 5. In the image enlarging circuit 5, the image read from the image memory 2 is corrected according to the image enlargement ratio data s4 so that the image has a normal screen size. Then, the video signal s5 whose screen size has been corrected in this way is output to the signal processing circuit provided in the next stage.

As described above, since the video signal s1 reproduced by the video reproducing device 1 is subjected to a series of processes, even if the reproduced video signal s1 has an image shake, there is no shake. A stable video signal s5 is output from this image blur correction device.

By the way, if the subject on the screen is deformed,
In order to remove a subject that cannot be processed, it is generally performed to divide a screen into a plurality of calculation areas (motion vector detection areas) and detect a motion vector for each of the plurality of calculation areas.

In this case, the motion vector detection circuit 3 detects a motion vector for each calculation area. Then, the microcomputer 30 obtains the average value of the motion vector amounts thus detected, or combines each motion vector by a method such as median processing to calculate the shake amount of the entire screen. The shake amount of the entire screen represents a difference value between frames or fields at the image position.

Next, the shake amount of the entire screen is subjected to integration processing or low-pass processing by a low-pass filter to obtain the final shake correction amount of the image, that is, the image stored in the image memory 2. The address signal s30 representing the signal read position is determined. Note that a fixed value is usually used as the image enlargement ratio data s4.

[0012]

However, in the conventional image shake correction apparatus as described above, when correcting the shake of an actually broadcast television image or an image recorded by a home video camera, There was a problem as described below.

That is, in a television image that is actually broadcast, character information such as a broadcast time, a broadcast title or an annotation, or a Japanese translation may be superimposed. In addition, an image recorded by a home video camera may have superimposed date and time at the time of shooting, or text information indicating shooting content.

When the video signal s1 containing the character information thus superimposed is input to the motion vector detecting circuit 3 and used as it is for detecting the motion vector, the superimposing is performed. There was a case that the movement of the written character was detected as it was.

For example, when there is a character whose display position does not change on the screen such as date and time, a motion vector detection area around the display position,
In the motion vector detection area including the display position, the image shake may be detected as being smaller than the actual shake.

Further, in the case of a title or telop indicating a shooting condition, etc., if there is a character that moves in the screen in a flowing manner, a motion vector detection area around the display position and a motion including the display position are displayed. In the vector detection area, such a moving speed of characters may be detected.

Therefore, if the image shake amount of the image is calculated by using the conventional image shake correction apparatus as described above, and the image shake correction is performed based on the calculated shake amount of the image, it occurs at the time of actual photographing. In addition, there is a problem in that a process unrelated to the process for correcting the shake of the image is performed, resulting in an unsightly image.

The present invention has been made to solve such a problem, and prevents the occurrence of an error in the image shake correction operation due to the presence of superimposed characters in an image. However, the objective is to always obtain a good image.

[0019]

An image blur correction apparatus of the present invention detects a motion vector from a reproduced video signal in a plurality of detection areas, and based on the plurality of motion vectors detected in each of the detection areas. In an image blur correction device for correcting image blur by means of a character detection means for detecting a signal component of a character superimposed on the video signal, and a result of detection of the character component by the character detection means, A control means for changing the shake correction processing for each of the detection areas is provided.

Another feature of the image blur correction apparatus according to the present invention is that the image blur correction apparatus detects a motion vector from a reproduced video signal and corrects the image blur based on the motion vector. , A character detecting means for detecting a signal component of a character superimposed on the video signal, and a process for correcting image shake according to a result of detection of the character component by the character detecting means is stopped. A control means for controlling the above is provided.

In the image shake correcting apparatus having the above-mentioned configuration, the character detecting means includes first detecting means for detecting the edge strength of the image based on the reproduced video signal, and the reproduced video signal. Of the image signal detected by the first detecting means, and the luminance of the video signal detected by the second detecting means. It may be configured by an encoding means for encoding the video signal.

Further, the character detecting means includes a first detecting means for detecting a motion amount of an image based on a reproduced video signal, and a second detecting means for detecting a luminance of the reproduced video signal. Encoding means for encoding the reproduced video signal based on the amount of motion of the image detected by the first detection means and the brightness of the video signal detected by the second detection means. It may be configured by.

Further, the character detecting means includes a first detecting means for detecting a motion amount of an image based on the reproduced video signal, and a first detecting means for detecting an edge strength of the image based on the reproduced video signal. 2 detecting means, the amount of motion of the image detected by the first detecting means, and the edge strength of the image detected by the second detecting means, and the reproduced video signal is encoded. It may be configured by an encoding means for performing.

[0024]

Since the present invention comprises the above technical means, the motion vector in the area in which the signal component of the character superimposed on the video signal is detected is excluded from the target when calculating the shake amount of the image, This eliminates the influence of the character movement that causes an error when calculating the image shake amount, and makes it possible to accurately calculate only the image shake amount at the time of actual shooting.

According to another feature of the present invention, the movement of the character causes an error in calculating the shake amount of the image based on the detection result of the character superimposed on the video signal. When it is determined that the image blurring is likely to occur, the image blurring correction operation is stopped, so that it is possible to prevent the inconvenience that an unsightly image is obtained by performing the image blurring correction.

[0026]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the first embodiment of the image shake correction apparatus of the present invention. Note that, in FIG. 1, components denoted by the same reference symbols as those shown in FIG. 9 have the same functions, and thus detailed description thereof will be omitted.

The image blur correction apparatus of this embodiment is obtained by adding an impose character detection circuit 6 to the configuration of the conventional image blur correction apparatus shown in FIG. Then, the microcomputer 4 determines the address signal s3 indicating the read position of the image based on the detection signal s6 of the imposing character output from the imposing character detecting circuit 6.

Now, the principle of the imposing character detection circuit 6 will be described. In general, an image captured by a video camera is distributed in a frequency region considerably lower than the frequency band of a video signal. On the other hand, the above-mentioned image of the imposing character is generally a white image with sharp edges and is concentrated in a narrow range to some extent on the screen, and therefore has an outstanding frequency spectrum. There is.

Also, even if there is a bright subject at the position where the imposing character is displayed, the brightness of the imposing character is automatically increased so that the imposing character can be easily read, A superimpose circuit has been put into practical use, which has a function to add a black edge or blur to the background around the characters.

Therefore, when the video signal taken by the video camera is processed by such a superimposing circuit, the image taken by the video camera and the image of the imposed character are displayed. Then, the difference in the frequency components becomes more remarkable.

The imposing character detection circuit 6 pays attention to the difference in the frequency components of both images and extracts only the imposing character component from the video signal s1 reproduced by the video reproducing device 1. It was done.

Next, FIG. 2 shows an example of the configuration of the imposing character detecting circuit 6. In FIG. 2, 7 is a differentiating circuit, 8 is a low-pass filter, 9 is a binarizing circuit, and 10 is an isolated point removing circuit. Further, s1 is a video signal reproduced by the above-described video reproducing device 1, s7 is a video signal output from the differentiating circuit 7 with edges emphasized, s8 is a luminance information signal output from the low-pass filter 8, and s9 is 2 A binary video signal output from the binarization circuit 9, s6 is the above-mentioned detection signal of the imposed character.

In the imposing character detecting circuit 6 having such a configuration, the input video signal s1 is given to the differentiating circuit 7 and the low pass filter 8. Then, based on the video signal s1, the differentiating circuit 7 generates a video signal s7 with an edge emphasized. Further, the low-pass filter 8 generates a brightness information signal s8 having average brightness information around the scan pixels.

These video signal s7 and luminance information signal s8
And are input to the binarization circuit 9, respectively. In the binarization circuit 9, the threshold level is appropriately changed according to the level of the luminance information signal s8 given from the low pass filter 8. Then, the edge-enhanced video signal s7 given from the differentiating circuit 7 is binarized by this threshold level.

The video signal s9 binarized by the binarizing circuit 9 is input to the isolated point removing circuit 10,
The noise that occurs suddenly is removed. Then, a portion where the video signal s9 is concentrated in a certain narrow range on the screen is detected, and the signal thus detected is
The detection signal s6 of the imposed character is output to the microcomputer 4 provided in the next stage.

As described above, according to the imposing character detection circuit 6 of the present embodiment, the position of the imposing character multiplexed on the video signal s1 reproduced by the video reproducing device 1 on the screen can be almost accurately determined. Can be detected.

Next, the configuration of another embodiment of the imposing character detection circuit 6 is shown in FIG. In FIG. 3, 11 is an image memory, 12 is a comparison circuit, 13 is a low-pass filter, 1
Reference numeral 4 is a threshold circuit, and 15 is an AND circuit. Also, s
Reference numeral 10 is a comparison signal output from the comparison circuit 12, and s11 is a binary signal output from the threshold circuit 14.

In the imposing character detection circuit 6 having such a configuration, the input video signal s1 is input to the image memory 1
1 and one input terminal of the comparison circuit 12. The image memory 11 has an F capacity having a storage capacity for one frame.
This is an IFO (First In First Out) memory, and the image signal of one frame before is always output from the image memory 11.

The video signal thus output from the image memory 11 is applied to the other input terminal of the comparison circuit 12. As a result, the comparison circuit 12 receives the two video signals of the current frame video signal and the previous frame video signal.

The comparison circuit 12 compares, for each pixel, the two video signals thus input, that is, the video signal of the current frame and the video signal of the previous frame. When the two pixel signals to be compared are equal, “1” is given to one input terminal of the AND circuit 15 as the comparison signal s10. On the other hand, when the two pixel signals are not equal, “0” is given to one input terminal of the AND circuit 15 as the comparison signal s10.

That is, the comparison circuit 12 calculates the inter-frame difference of the video signal between the current frame and the previous frame. Then, "1" is output as the comparison signal s10 in the pixel in the portion where there is no difference between frames, that is, in the portion where there is no image movement. On the other hand, "0" is output as the comparison signal s10 in the pixels of the part where the image moves.

On the other hand, the video signal s1 reproduced by the video reproducer 1 is processed by the low-pass filter 13 and the threshold circuit 1.
4 is also input to one of the input terminals. The low-pass filter 13 and the threshold circuit 14 constitute a so-called adaptive binarization circuit.

That is, the low-pass filter 13 calculates the average luminance information around the scanning pixels from the input video signal s1. The brightness information calculated by the low-pass filter 13 is given to the other input terminal of the threshold circuit 14, and the threshold value in the threshold circuit 14 is appropriately changed according to the brightness information. The video signal s1 input to one input terminal of the threshold circuit 14 is binarized by this threshold.

As a result, in the pixel of the high brightness portion in the image, the binary signal s11 is applied to the other input terminal of the AND circuit 15.
"1" is given as, and "0" is given to the other pixels. Then, the logical product circuit 15 performs a logical product of the comparison signal s10 and the binary signal s11 (AN
D) is calculated, and the calculation result is output as the detection signal s6 to the microcomputer 4 provided in the next stage.

The detection signal s6 represented by "1" represents the position of a pixel having no image motion and having high brightness. Therefore, according to the imposing character detection circuit 6 of the present embodiment, it is possible to almost accurately detect an imposing character whose display position does not change within the screen, such as date and time.

The amount of motion of the image is detected by first obtaining the inter-frame difference of the video signal and dividing the value by a parameter indicating the complexity of the image (for example, image gradient). Is common. However, in the case of the imposing character detection circuit 6 of this embodiment, it is not particularly important to detect the exact movement of the image.

Therefore, in this embodiment, only the above-mentioned interframe difference is calculated, and the image gradient calculation and the division by the image gradient are not performed. Thereby, the gradient calculation circuit and the division circuit can be omitted, and the circuit scale can be reduced. In the case of this embodiment,
It goes without saying that the amount of motion of the image may be detected by the general method described above.

Further, for example, when the reproduced video signal s1 is 8-bit digital data, the comparison circuit 12
The comparison operation by does not necessarily have to be performed using all of the 8-bit data. For example, it is often sufficient to perform the comparison operation using only the upper 3 bits or 4 bits of data.

Further, in the above description, the case where all the movements of the image are detected in the frame unit has been described as an example, but the movement may be detected in the field unit. In this case, the storage capacity of the image memory 11 is half that in the case where the movement of the image is detected in frame units.

Next, FIG. 4 shows the configuration of still another embodiment of the imposing character detection circuit 6. In FIG. 4, 16 is an addition circuit, 17 is a multiplication circuit, 18 is an image memory, 19 is a differentiation circuit, and 20 is a binarization circuit. Further, s12 is a video signal obtained by multiplying the video signal of the previous frame by a predetermined coefficient, and s13 is a video signal output from the adding circuit 16.

Here, the adder circuit 16 and the multiplier circuit 1 described above are used.
7 and the image memory 18,
This is a circuit portion that detects the amount of movement of an image. Further, the circuit portion configured by the differentiating circuit 19 and the binarizing circuit 20 described above is a circuit portion that detects edge information of an image.

In the imposing character detecting circuit 6 having such a configuration, the input video signal s1 is added by the adding circuit 16
Is applied to one of the input terminals. Also, this adder circuit 1
The video signal s13 output from 6 is subjected to predetermined processing by the multiplication circuit 17 and the image memory 18, and then applied to the other input terminal of the addition circuit 16.

The above-described multiplication circuit 17 has a coefficient of 0 or more and 1 or less. Then, the multiplication circuit 17 multiplies the video signal s13 output from the addition circuit 16 by the coefficient. Then, the video signal multiplied by the coefficient by the multiplication circuit 17 is given to the image memory 18 and temporarily stored therein. The image memory 18 is a FIFO memory having a storage capacity of one frame, like the image memory 11 described above.
The video signal before the frame is always output.

Therefore, in the adder circuit 16, the reproduced video signal s1 of the current frame applied to one input terminal,
The video signal s12 obtained by multiplying the video signal of the previous frame applied to the other input terminal by the coefficient is added. Then, the video signal s13 obtained as a result of this addition is given to the differentiating circuit 19.

As described above, the adder circuit 16, the multiplier circuit 17, and the image memory 18 form a temporal low-pass filter for the video signal. Then, such a temporal low-pass filter has such a characteristic that a pixel signal in a portion of the input video signal s1 having a high luminance and a small image motion outputs a larger value. Further, it has a characteristic that the output level of an image having a slight movement is lowered and the edge of the image is extremely dull.

Therefore, if there is a portion having a steep edge characteristic in the video signal s13 outputted from this temporal low-pass filter, that portion originally has a sharp edge, there is no movement of the image, and it is high. It can be inferred that the pixel has brightness. Therefore, the differentiation circuit 1 of the next stage
It is possible to detect a signal component having the above-mentioned steep edge characteristic in the circuit portion constituted by 9 and the binarization circuit 20, and encode the video signal s13 according to the result of the detection. It has such a structure.

That is, the video signal s13 output from the adder circuit 16 is given to the differentiating circuit 19, and the differentiating circuit 19 extracts the edge gradient of the video signal s13. Next, the binarization circuit 20 outputs the video signal s13 to 2 according to the signal strength level of the edge gradient.
Valued. Then, the binary signal thus obtained is output to the microcomputer 4 provided in the next stage as the detection signal s6 of the imposed character.

As described above, according to the imposing character detection circuit 6 of this embodiment, from the input video signal s1,
It is possible to extract a component having no image movement, high brightness, and a sharp edge. Accordingly, it is possible to detect an impose character whose display position does not change within the screen, such as date and time, almost accurately.

In this embodiment also, the video signal s1
It is not always necessary to detect the motion of the image using the full bit width of. Further, it is not always necessary to detect the movement of the image in frame units, and it may be detected in field units. By doing so, the circuit scale can be significantly reduced.

As described above, according to the imposing character detecting circuit 6 shown in FIG. 2, FIG. 3 or FIG. 4, only the signal component of the imposing character is almost accurately extracted from the video signal with a simple circuit configuration. can do.

Next, the operation contents of the entire image shake correcting apparatus according to the first embodiment shown in FIG. 1 will be described with reference to the flowchart of FIG. Note that, here, the image enlarging circuit 5
It is assumed that the data s4 of the image enlargement ratio given to is constant.

First, in step P1 of FIG. 5, the motion vector detection circuit 3 specifies from which motion vector region in the image the motion vector is to be detected. Then, in step P2, the motion vector s2 of the area thus designated is detected.

Further, in step P3, it is detected whether or not an imposing character is multiplexed in the video signal s1 reproduced by the video reproducing device 1 within the area designated in step P1. The detection of this impose character is
As described above, it is performed by the imposing character detection circuit 6 as shown in FIG. 2, FIG. 3 or FIG.

Next, in step P4, the detection signal s6 of the imposing character detected by the imposing character detecting circuit 6 is changed to the motion vector detecting circuit 3 in step P2.
It is taken into the microcomputer 4 together with the motion vector s2 detected by.

Further, when there are a plurality of motion vector detection areas in the image, the process returns from the branch of step P5 to step P1 and the processes of steps P1 to P5 are repeated. As a result, the motion vector s2 and the detection signal s6 of the impose character are fetched into the microcomputer 4 for the required area.

Then, when the motion vector s2 and the detection signal s6 of the imposed character are fetched into the microcomputer 4 by the required area, the microcomputer 4 causes
In the following steps P6 to P10, image shake correction processing is performed.

That is, in step P6, the motion vector of the area in which the imposing character is detected is determined in step P7.
It is instructed not to be used in the subsequent processing. This instruction is performed, for example, by setting an invalid flag for the motion vector of each area.

Next, in step P7, the area to be subjected to the image shake correction is determined based on the variation value of the motion vector fetched by the microcomputer 4 and the degree of alignment in the image. In this determination, only the motion vector of the area where the invalid flag is not set, that is, the area where the imposing character is not detected is used.

Next, at step P8, the average value of the motion vectors in the area where the invalid flag is not set, that is, the shake amount of the image on the entire screen is obtained. In the next step P9, an integral operation is performed on the average value of the motion vectors obtained in step P8 to calculate the final shake correction amount of the image.

Further, in step P10, step P9
The address signal s3 representing the read position of the video signal stored in the image memory 2 is determined based on the image shake correction amount calculated in step S3, and the address signal s3 is output to the image memory 2. The video signal at the position designated by the address signal s3 is output from the image memory 2 to the image enlarging circuit 5 in the next stage. Then, the processing of steps P1 to P10 as described above is repeated for each field.

Here, the process in step P6 is equivalent to the process of removing the motion vector of the area including the imposed character from the motion vectors detected by the motion vector detection circuit 3. Therefore, in the process after step P7, only the motion vector other than the removed motion vector is used, and the address signal s3 determined in step P10 is influenced by the motion amount of the character such as date and time. The value is not received.

As described above, the image blur correction apparatus according to the first embodiment is provided with the imposing character detecting circuit 6 for detecting the position of the imposing character on the screen, and the imposing character detecting circuit 6 detects the position. The feature is that the motion amount of the image is calculated by excluding the motion vector of the area including the generated imposing character.

By configuring the image shake correction apparatus in this way, even if characters such as date and time are multiplexed in the video signal, only the shake amount of the image at the time of shooting can be correctly detected. It is possible to perform image blur correction without malfunction.

Next, a second embodiment of the image blur correction device according to the present invention will be described. The image blur correction device according to the second embodiment is made in consideration of the following circumstances.

That is, among the imposing characters that are most commonly used in the television image actually broadcasted and the image taken by the home video camera, the most common and frequently used one is the date or It is a time character. Then, these characters are usually multiplexed mainly in the peripheral portion of the screen, such as in the lower right or lower left of the screen.

On the other hand, the imposing character detection circuit 6 as described in the above-mentioned first embodiment always includes some detection error. Therefore, the image blur correction operation becomes unstable depending on the type of captured image. Therefore, in the second embodiment, the above-described inconvenience is prevented by performing the following calculation in the process of step P8 in the flowchart shown in FIG.

FIG. 6 shows an example of a television screen for explaining the principle of this embodiment. The television screen of FIG. 6 shows a total of 60 motion vector detection areas of 6 rows by 10 columns. Hereinafter, the positions of these motion vector detection areas will be represented by (i, j), where the row number is i (i = 1 to 6) and the column number is j (j = 1 to 10).

Further, in FIG. 6, 21 is an effective pixel area of the television image, 22 is the first outermost detection area out of 60 detection areas in the screen, and 23 is the first detection area 22. The second detection area is the innermost detection area, and 24 is the innermost third detection area. Next, 25 is a date character that is multiplexed in the image, 26 is a time character, and 2 is a time character.
Although the area 7 is not originally the area of the impose character, it is assumed that the area 7 is an erroneous detection area erroneously detected as the area of the impose character.

By the way, in the above-mentioned first detection area 22, second detection area 23, and third detection area 24, for example, according to the probability that the imposing characters are multiplexed in each area, for example, We set weighting factors such as 1.0, 0.9, and 0.1 that increase toward the outside of the screen.

Therefore, in the present embodiment, in step P8 of the flow chart shown in FIG. 5, the weighting coefficient set in this way is used to calculate in each detection area according to the following equation (1). The motion vector averaging calculation is performed.

[0081]

[Equation 1]

Here, MVa is the calculation result of averaging, MV
Is a motion vector value of each detection area, T is an invalid flag (0 or 1) indicating whether or not an imposing character is detected in each detection area, and W is the first detection area 22 to the third detection area 24. The weighting factors set in the table are shown respectively.

According to the averaging operation shown in (Equation 1), the invalid flag T (= 1) has a weight of 1.0 in the detection area of (5, 10) in which the date character 25 is multiplexed. It is multiplied by the factor W so that the motion vector in this detection area is not used in the averaging operation. Similarly, for the detection area of (6, 10) in which the time character 26 is multiplexed, the invalid flag T (= 1) is multiplied by the weighting coefficient W of 1.0, and the motion vector of this detection area is subjected to the averaging calculation. Will not be used.

On the other hand, in the detection area of (4, 5) including the erroneous detection area 27, the invalid flag T (= 1) is multiplied by the weighting coefficient W of 0.1, and as a result, the motion vector is 0.9.
Will be evaluated with the weight of. In other words, this (4
In the detection area of 5), although the imposing character is erroneously detected, its motion vector is used for the averaging operation almost normally.

As described above, according to the second embodiment, it is determined whether or not the motion vector detected by the imposing character detection circuit 6 is an erroneous detection based on the weighting coefficient set in each detection area. be able to. Therefore, even if the imposing character detection circuit 6 has a certain detection error, it is possible to prevent the image shake correction operation from becoming unstable.

By applying the method of the second embodiment, the threshold level of the binarization circuit 9 shown in FIG. 2 is lowered in the outermost first detection area 22 in the screen. To set. Alternatively, in the first detection area 22, the isolated point removing circuit 10 is set to have a low ability to remove isolated points. By doing so, it is also useful to make it easy for the imposing character detection circuit 6 to detect a small imposing character or an imposing character having a low brightness level.

Next, a third embodiment of the image blur correction device according to the present invention will be described. As described above, the imposing character detection circuit 6 erroneously detects even an ordinary subject as an imposing character and rarely generates the imposing character detection signal s6. If the detection signal s6 erroneously detected in this way is used as it is for the image blur correction process, the image blur correction operation (anti-vibration operation) becomes unstable.

Therefore, the image blur correction device of this embodiment is
It is configured to prevent such inconvenience. Specifically, the image blur correction device of this embodiment is configured as shown in FIG.

In FIG. 7, the microcomputer 28 determines whether or not the detection signal s6 of the imposing character detected by the imposing character detecting circuit 6 is correct, and based on the result of this judgment, the vibration prevention is performed. Decide whether to continue the work.

The operation contents of the image shake correcting apparatus according to the present embodiment will be described below with reference to the flowchart of FIG. In FIG. 8, the processes of steps P1 to P5 are as follows.
This is the same as the processing of steps P1 to P5 of the flowchart shown in FIG. Then, when the processes of steps P1 to P5 are completed, the microcomputer 28 performs the processes of steps P6 to P10 as described below.

That is, the microcomputer 28 first determines in step P6 whether or not the motion vector detecting circuit 3 has generated an imposing character detection signal s6, that is, the imposing character detecting circuit 6 detects an imposing character. Determine whether or not. At this time, if it is determined that the imposing character is not detected at all, the process proceeds to step P9 to continue the image stabilizing operation.

On the other hand, if it is determined that the imposing character is detected, the process proceeds to step P7, and it is determined whether or not there are n or more areas where the imposing character is detected. In addition,
For example, when there are 60 motion vector detection areas in the image, n is about 10.

If it is determined in step P7 that there are n or more areas in which the imposing character is detected, it is considered that the imposing character is the title character displayed in large size on the screen. , Step P10
Go to and stop the anti-vibration operation. On the other hand, if there are less than n areas in which the imposing characters have been detected, step P8
Proceed to.

In step P8, the motion vector value MVt of the area in which the imposing character is detected is compared with the motion vector value MVn of the other area. Here, when MVt≈MVn (Equation 2) holds, it means that “the subject in the background and the imposing character are moving at the same speed”.

That is, in this case, it can be inferred that the imposing character detection circuit 6 has erroneously detected an ordinary subject as an imposing character. Further, even if the imposing character is correctly detected by the imposing character detecting circuit 6, as described above,
It can be considered that the imposing character and the normal subject are moving in the screen at the same speed.

Therefore, in such a case, even if the image stabilization operation is continued as it is, an unsightly image in which only the imposed characters move around the screen does not appear. Therefore, when the above formula (2) is satisfied, the process proceeds to step P9 to continue the image stabilization operation.

On the other hand, when the above equation (2) does not hold, the following two situations can be considered. That is, when MVt << MVn ... (Equation 3) holds, it is considered that “the image has a shake and imposing characters are multiplexed at a fixed position”. If the image stabilization operation is continued in such a case, there arises a problem that image shake correction cannot be accurately performed.

On the contrary, when MVt >> MVn ... (Equation 4) is satisfied, "impose characters displayed while moving in a flowing manner are multiplexed on an image with little shake. Therefore, it is impossible to perform the correct shake correction in the same manner as described above.

Therefore, in these cases, that is, when the above equation (2) does not hold, step P
Proceed to 10 to stop the image stabilization operation. Then, the processing of steps P1 to P10 as described above is repeated for each field.

The image shake correcting apparatus according to the third embodiment described above is made on the assumption that the imposing character detecting circuit 6 has some detection error. However, if the detection accuracy of the imposing character in the imposing character detecting circuit 6 is completely reliable, the image stabilizing operation may be stopped as soon as the imposing character is detected.

In this case, in step P6 of FIG.
If it is determined that the imposing character has been detected, the processing of steps P7 and P8 is not performed, and immediately step P10 is performed.
Go to and stop the image stabilization operation.

As described above, according to the third embodiment, for example, even if a superimposed character is superimposed on the video signal during recording, only the portion of the superimposed character moves around the screen. It is possible to prevent an uncomfortable image.

[0103]

As described above, the present invention is provided with the character detecting means for detecting the signal component of the character superimposed on the video signal, and the character detecting means detects the character component other than the detection area where the character component is detected. Since the shake of the image is corrected by using the motion vector in the detection area, for example, even if characters such as date and time are multiplexed in the video signal, the cause of the error in calculating the shake amount of the image It is possible not to be influenced by the movement of the character that becomes, and it is possible to accurately calculate only the shake amount of the image at the time of actual photographing.

According to another feature of the present invention, character detection means for detecting the signal component of the character superimposed on the video signal is provided, and the character detection result by this character detection means is provided. Therefore, the image stabilization operation is controlled, so when there is a risk that the movement of characters will cause an error when calculating the shake amount of the image, the image stabilization operation can be stopped, and only the character part is displayed on the screen. It is possible to prevent an uncomfortable image such as moving around.

Therefore, according to the image blur correction apparatus of the present invention, it is possible to always perform the image blur correction without malfunction, and furthermore, the image blur correction causes an unsightly image. It can be eliminated, so that a good image can always be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of an image blur correction device of the present invention.

FIG. 2 is a block diagram showing an example of a configuration of an impose character detection circuit in the image blur correction device shown in FIG.

FIG. 3 is a block diagram showing another example of a configuration of an impose character detection circuit in the image blur correction device shown in FIG.

FIG. 4 is a block diagram showing still another example of the configuration of the imposing character detection circuit in the image blur correction device shown in FIG. 1.

FIG. 5 is a flowchart showing the operation of the image shake correction apparatus according to the first embodiment.

FIG. 6 is a diagram showing an example of a television screen for explaining the principle of the second embodiment of the image shake correction apparatus according to the present invention.

FIG. 7 is a block diagram showing the configuration of a third embodiment of the image shake correction apparatus of the present invention.

FIG. 8 is a flowchart showing the operation of the image shake correction apparatus according to the third embodiment.

FIG. 9 is a block diagram showing a configuration of a conventional image blur correction device.

[Explanation of symbols]

 3 motion vector detection circuit 4 microcomputer 6 imposing character detection circuit 7 differentiation circuit 8 low-pass filter 9 binarization circuit 10 isolated point removal circuit 11 image memory 12 comparison circuit 13 low-pass filter 14 threshold circuit 15 AND circuit 16 adder circuit 17 Multiplying circuit 18 Image memory 19 Differentiating circuit 20 Binarizing circuit 28 Microcomputer s1 Reproduced video signal s2 Motion vector s3 Read address signal s6 Impose character detection signal

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H04N 5/91

Claims (5)

[Claims] 1. An image shake correction apparatus for detecting a motion vector from a reproduced video signal in a plurality of detection areas and correcting the shake of an image based on the plurality of motion vectors detected in each of the detection areas, Character detection means for detecting a signal component of a character superimposed on the video signal, and the shake correction processing for each of the detection areas is changed according to the result of detection of the character component by the character detection means. An image blur correction apparatus comprising: a control unit. 2. An image blur correction device for detecting a motion vector from a reproduced video signal and correcting the blur of an image based on the motion vector, wherein a signal component of a character superimposed on the video signal is detected. An image characterized by being provided with a character detecting means for detecting and a control means for controlling so as to stop the processing for correcting the shake of the image in accordance with the result of the detection of the character component by the character detecting means. Shake correction device. 3. A first detecting means for detecting an edge strength of an image based on a reproduced video signal, a second detecting means for detecting a brightness of the reproduced video signal, and the first detecting means. The character detecting means by the encoding means for encoding the reproduced video signal based on the edge strength of the image detected by the means and the brightness of the video signal detected by the second detecting means. The image blur correction device according to claim 1, wherein the image blur correction device is configured. 4. A first detecting means for detecting a motion amount of an image based on a reproduced video signal, a second detecting means for detecting a luminance of the reproduced video signal, and the first detection. The amount of movement of the image detected by the means,
The character detecting means is constituted by an encoding means for encoding the reproduced video signal based on the brightness of the video signal detected by the second detecting means. 2. The image blur correction device described in 2. 5. A first detecting means for detecting an amount of motion of an image based on a reproduced video signal, and a second detecting means for detecting an edge strength of the image based on the reproduced video signal. The amount of motion of the image detected by the first detection means,
2. The character detecting means is configured by an encoding means for encoding the reproduced video signal based on the edge strength of the image detected by the second detecting means. 2. The image blur correction device described in 2.