JP2809656B2 - Image motion vector detection device and shake correction device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention mainly relates to an apparatus for detecting a motion vector of an entire screen due to a vibration or shaking of a video camera from an image signal, and an unnecessary screen of the entire screen due to vibration or shaking of a video camera. The present invention relates to a device that corrects shaking.

2. Description of the Related Art A conventional image motion vector detecting device is disclosed in, for example, Japanese Patent Application Laid-Open No. 61-269475.

FIG. 13 shows a block diagram of an image fluctuation correcting apparatus to which the conventional motion vector detecting apparatus is applied.
Is an image signal input terminal. Reference numeral 2 denotes a vector detection circuit for each detection area. Reference numeral 3 denotes a vector reliability determination unit, and reference numeral 4 denotes a motion vector determination unit. Reference numeral 5 denotes a shake correction unit.

In the conventional image motion vector detecting device configured as described above, first, an input terminal 1 is supplied with a temporally continuous image signal of at least two fields.
In the vector detection circuit 2 of each detection area, FIG.
As shown in 2, 17, and 18, four vector detection areas are defined in the image, and for each detection area between two fields, a predetermined range of amount (i, j): imin <i <imax, jmin <J
<The difference Σ | ΔL | (i, j) of the signal at the position shifted by jmax is obtained as a correlation value, and the shift (i ′, j ′) that gives the minimum value is detected. Vector. In addition, the minimum value, average value, and maximum value of the correlation value are obtained,
Along with the value of the vector, it is output to the vector reliability determination circuit means 3 and the motion vector determination means 4. The vector reliability determining means 3 and the motion vector determining means 4 determine the reliability of the vector from the minimum value, the average value, and the maximum value of the correlation values of each area as shown in the flowchart of FIG. If the magnitude and directionality of the vector of the area determined to be the same, the average value thereof is determined and output as a motion vector of the entire screen. If they do not match, a zero vector is output. The shake correcting means 5 has an image signal storing means of one or more fields, and by controlling a reading position of a signal therefrom, when an image having a large fluctuation as shown in FIGS. With respect to the input image signals of the (n-1) th field 19 and the nth field 20, based on the motion vector 10 input from the motion vector determination means 4, the screen of the image signal to be read in the direction 21 for correcting the motion is displayed from 29. It moves in parallel to 30 and performs operations such as enlargement of the screen by interpolation or the like, and outputs image signals 27 and 28 of the size of one screen with little shaking.

In the above description, the reliability of the vector is determined from the minimum value, the average value, and the maximum value of the correlation values. In addition, the difference between the value of the minimum point of the correlation value and the surrounding points, that is, the minimum point There are various methods for judging the reliability from the state of the correlation value, such as a method of judging the reliability of the vector from the surrounding inclination and a method of judging only the minimum value.

In some cases, the determination of reliability is represented by a weight, and a vector obtained by weighting and averaging the vectors with the weight is output as a motion vector of the entire screen.

However, in the above configuration, the motion vector detection device outputs a zero vector when a moving object enters the screen, or detects a vector to which the motion of the moving object is added. Often. Therefore, in the shake correction device, the correction is interrupted or the vector to which the movement of the moving object is added is detected, and the screen is corrected in a direction different from the direction intended by the photographer, which often causes a malfunction.

This will be described in detail with reference to the drawings. First, FIG. 7 shows a state of a correlation value from the screen and each area when a moving object having a size equal to or larger than the detection area passes through the screen. The first time a moving object comes in.
When both the background and the moving object 13 are present in a certain detection area 6 as shown in FIG. 7A, the state of the correlation value 7 in that area (such as an increase in the minimum value) as shown in FIG. , The area is determined to be unreliable, but when the moving object 13 occupies most or all of the detection area 6 as shown in FIG. 7B, the detection area 6 A motion vector 15 obtained by adding the motion of the video camera and the motion of the moving object is detected, and the state of the correlation value 7 is determined to be reliable. At this time, if the motion vector 8 of the video camera is detected from the other area 12, since these two vectors do not match, 0 vector is output as the motion vector of the entire screen. (Areas 17 and 18 are areas 6 and 12, respectively.
Since the state is the same as that described above, the description is omitted. Then, as the moving object moves, the situation in which both the background and the moving object 13 are present again in the area 6 {not shown in FIG. Fig. 7 (d),
The situation is as shown in FIG. Therefore, the motion vector detected at the time points (a) to (e) in FIG.
As shown in FIG. In FIG. 15, reference numeral 8 denotes a motion vector due to camera shake, which vibrates up and down around the 0 vector. Numeral 15 is a vector obtained by adding the motion of the moving object, and a substantially constant value is added to the vector. 9, 16 are areas 6,
In the vectors detected from 12, the motion vector 8 due to the camera shake, the vector 15 to which the motion of the moving object is added, or other erroneous vectors are detected depending on how the moving object is applied to the area. If the shake correction is performed based on the output motion vector 10, the shake correction is interrupted in the state of FIG. Such interruption of vector detection can be improved to some extent by increasing the number of detection areas in the screen, but it causes an increase in the circuit scale and complexity of the device. If the size of each detection area is reduced in order to increase the number, the feature amount of an image necessary for detecting a motion vector for each area decreases, and the probability of detecting an erroneous vector in each area increases.

Also, as shown in FIG. 9, when the moving object 13 occupying most or all of the entire screen passes through the screen, the detection area 6 first becomes as shown in FIG. When both the background and the moving object exist in the area, the area 6 is determined to be unreliable according to the state of the correlation value 7, but as shown in FIG. 9 (b). When the detection area 6 occupies most or all of the detection area 6, a motion vector obtained by adding the motion of the video camera and the motion of the moving object is detected from the detection area, and the state of the correlation value 7 is determined to be reliable. You. When the moving object occupies most or all of the screen as shown in FIGS. 9C and 9D, the motion vector obtained by adding the motion of the video camera and the motion of the moving object is used as the motion vector of the entire screen. Is output. Therefore, the motion vectors detected at the time points (a) to (e) in FIG. 9 are as shown in FIG. 16, and when the shake correction is performed based on this vector, the motion vector is corrected in the state of (b). Is interrupted, and the screen is corrected in a direction different from the direction intended by the photographer in the states (c) and (d), resulting in a malfunction.

Means for Solving the Problems In order to solve the above problems, the present invention provides a plurality of detection areas in a screen, and for each of the detection areas, means for calculating a correlation value between a current field and a previous field at a predetermined shift. And means for calculating a motion vector for each detection area from the correlation value, and determining the reliability of each detection area, wherein at least the reliability determination result in each frame is recorded as a history. From the state of the correlation value in the frame, the determination result is changed from reliable to unreliable, and from the state of the correlation value in each frame and the degree of coincidence of the detected motion vector with the motion vector of the entire screen. Means for judging the motion of each detection area, which is characterized in that the judgment result is changed from unreliable to reliable, based on the judgment of the reliability. A means for determining a motion vector of the entire screen using a motion vector of each area is provided. The invention according to claim 2 is characterized in that a plurality of detection areas are arranged in a horizontal direction so as not to be arranged vertically in the screen. Means for determining a correlation value at a predetermined shift in the detection area for each of the detection areas, and detecting each of the detection areas based on the correlation value. Means for obtaining a motion vector for each area, means for determining the reliability of the motion vector of each detection area at this time, and motion of the entire screen using the motion vector of each area based on the reliability determination there. Preferably, a means for determining a vector is provided. In the invention according to claim 1, preferably, the detection areas are provided so as to be shifted in the horizontal direction so as not to be aligned in the vertical direction, or each of the detection areas is horizontally shifted. Providing shifted in the longitudinal direction, characterized in so that it does not line up.

Then, based on the motion vector detecting device, the image motion vector detecting device, at least one field of image signal storing means, and the motion vector detected from the image motion vector detecting device, the image signal storing is performed. It is characterized by an image shake correction device comprising means for controlling the reading position of the means.

Effect of the Invention With the configuration described above, the present invention determines not only the state of the correlation value and the value of the vector in each detection area at that time when determining the reliability of the vector in each detection area, but also the vector before the detection area. Is also determined using the result of the determination of the reliability and the value of the vector. This is when the above problem occurs
When a moving object first enters the detection area, the area is determined to be unreliable, and then when the moving object occupies most of the detection area, the motion of the video camera and the movement of the moving object are detected from the detection area. Is added and the motion vector is detected, and it is determined that there is reliability from the state of the correlation value. Therefore, attention is paid to the fact that the above problem occurs, and the first determination result is determined in the subsequent fields. By using it sometimes, the above problem can be solved. In other words, not only the state of the correlation value and the value of the vector at a certain time but also the reliability using the state of the correlation value and the value of the vector or the determination result of the reliability of the vector before that are used to determine the reliability. Solvable. Therefore, even when a moving object having a size equal to or larger than the detection area enters the screen, if a motion vector of the video camera is detected from other areas, only the vector can be detected. Even when a moving object that occupies most or the whole of the entire screen passes through the screen, the shake correction is interrupted for that time, and no malfunction occurs such that the screen is corrected in a direction different from the direction intended by the photographer. .

Also, when a plurality of detection areas are provided on the screen, the detection areas are shifted in the horizontal direction so as not to be arranged in a uniform manner in the vertical direction. Simultaneous detection of an erroneous vector from the arranged area can be prevented. Therefore, interruption of detection of a motion vector of a screen and detection of an erroneous vector can be suppressed.

Embodiment 2 FIG. 2 is a block diagram of a motion vector detecting device according to a first embodiment of the present invention, wherein 1 is an image signal input terminal. Reference numeral 2 denotes a vector detection circuit for each detection area. Reference numeral 3 denotes a vector reliability determination unit, and reference numeral 4 denotes a motion vector determination unit.

In the image motion vector detecting device of the first embodiment configured as described above, first, an input terminal 1 is supplied with a temporally continuous image signal of at least three fields or more. In the vector detection circuit 2 for each detection area, four vector detection areas are previously defined in the screen as shown in 6, 12, 17, and 18 in FIG. (I, j): imin <i <imax, j
min <j <jmax Difference of signal at shifted position Σ | ΔL | (i, j)
Are obtained as correlation values, a shift (i ′, j ′) that gives the minimum value is detected, and this is set as a vector of each region. In addition, the minimum value, average value, and maximum value of the correlation value are obtained, and the reliability value of the vector is determined as the vector reliability determination circuit means 3 together with the vector value. The vector reliability determination circuit means 3 determines the reliability of each area from each value input from the vector detection circuit 2 of each detection area and the value of the motion vector fed back from the motion vector determination means 4, The result of the determination and the motion vector of each area are output to the motion vector determining means 4. The motion vector determining means 4 determines a motion vector of the entire screen from the determination result of the reliability of each area and the motion vector of each area.

FIG. 1 shows the operation of the vector reliability determining means 3 and the motion vector determining means 4 of the image motion vector detecting device.
In other words, for an image signal input continuously in time,
5 is a flowchart for determining a motion vector of a screen at that time from a state of a vector value and a correlation value up to that time obtained from a vector detection circuit 2 of each detection area. First
These operations will be described with reference to the drawings. First, the vector reliability determination means 3 of each area first determines the first reliability based on the state of the correlation value (minimum value, average value, maximum value). It is. Next, in the second determination, the logical product of the first determination result and the third determination result of the previous field (the third determination will be described later) is obtained (the two determination results are both reliable). Only those that are reliable are determined to be reliable in the second determination.) However, the second determination is the same as the first determination since the third determination result of the previous field is not obtained in the first field. The motion vector determining means 4
The intermediate value of the vector of each area determined to be reliable in the second determination (a value located in the middle when arranged in descending order,
However, in the case of an even number, the average value of the two values near the middle) is determined and output as the motion vector of the entire screen. Further, the motion vector determination means 4 feeds back the motion vector of the entire screen to the third determination of the reliability determination means 3 of the vector of each area. In the third determination, a region in which the vector fed back and the vector of the region match within a predetermined range and the first determination result is reliable is determined to be reliable, and the result is determined in the next field. Used for the second determination.

Here, the difference from the conventional motion vector detecting device is that the vector obtained as a result of the motion vector determining means 4 is sent to the vector reliability determining means 3 of each area, and the reliability of the vector is determined again. The result is that the result is stored for one field, and is used for determining the reliability of the vector of each area in the next field.

As described above, according to the present embodiment, when determining the reliability of the vector of each detection area, not only the state of the correlation value and the value of the vector in each detection area at that time, but also the past of the detection area. Judgment is also made using the vector reliability judgment result and the vector value, so even if a moving object of the same size or larger than the detection area enters the screen, the video camera shakes from other areas. Is detected, it is possible to detect only the motion vector due to the shake of the video camera.

7 and 8 are diagrams for explaining this. FIG. 7 shows the correlation value between the screen and each area when the moving object 13 having the same size as the detection area enters the screen, and FIG. 8 shows the motion vector 10 detected at that time. It is shown. In FIG. 8, reference numeral 8 denotes a motion vector due to camera shake, which fluctuates up and down around the 0 vector. Fifteen
Is a vector to which the motion of the moving object is added, and a substantially constant value is added. Reference numerals 9 and 16 denote vectors detected from the areas 6 and 12, respectively. Depending on how the moving object occupies the area, the motion vector 8 due to the camera shake or the vector 15 to which the motion of the moving object is added or the other error A vector has been detected. Reference numeral 10 denotes a vector output from the motion vector determining means 4. First, when both the background and the moving object 13 are present in a certain detection area 6 as shown in FIG. 7A, the state of the correlation value 7 is changed as shown in FIG. 7A. It is determined that the area is unreliable (the operations of the areas 17 and 18 are the same as those of the areas 6 and 12 at this time, so the description is omitted.) Next, as shown in FIG. When most or all of the area 6 is occupied, a motion vector 15 that is the sum of the motion of the video camera and the motion of the moving object 13 is detected as the vector 9 of the detection area,
And it is determined that there is reliability from the state of the correlation value 7 (first determination). However, here, since the previously detected motion vector 10, that is, the motion vector 8 due to the camera shake and the vector 9 of the area 6 do not match, the third determination is unreliable, and therefore the second determination is not reliable. Only the area 12 (and 18) that does not contain a moving object is determined. Since the motion vector 8 due to the shaking of the video camera is detected from this area, the motion vector determining means 4 keeps detecting the motion vector 9 due to the shaking of the video camera. Further, when the moving object moves as time elapses, the situation where both the background and the moving object 13 are present again in the area 6 {not shown in FIG. Fig. 7 (d),
The situation is as shown in FIG. In (c) to (e), there is no situation where a moving object occupies most of a certain area.
Since the area where the moving object and the background are present is determined to be unreliable by the correlation value (first determination), only the motion vector 8 due to the camera shake from the area determined to be reliable in the second determination Is detected. Therefore, FIG.
The motion vector 10 detected at the time (e) is as shown in FIG. Here, as the output vector 10, only the motion vector 8 due to the shaking of the video camera can be detected from the vectors 9 and 16 of the areas 6 and 12, respectively.

Next, FIG. 3 is a block diagram of a shake correcting apparatus according to a second embodiment of the present invention, and 1 is an image signal input terminal. Reference numeral 2 denotes a vector detection circuit for each detection area. Reference numeral 3 denotes a vector reliability determination unit, and reference numeral 4 denotes a motion vector determination unit. Reference numeral 5 denotes a shake correction unit.

The operation of the image motion vector detecting device of the second embodiment configured as described above is almost the same as that of the first embodiment. However, if there is no area determined to be reliable in the second determination described in the first embodiment in the vector reliability determination unit 3 and the motion vector determination unit 4, the previous field and the previous field are used. Motion vectors M (n-1), M (n-
From 2), the motion vector M (n) of the current field is obtained as in the following equation.

When n> 2, M (n) = a × M (n−1) + b (M (n−1) −M
(N-2)) where a, b: 0 <a ≦ 1, 0 <b ≦ 1 When n = 2, M (n) = a × M (n−1) When n = 1, M (n) = 0 When the image having large fluctuation is input as shown in FIGS. 19 and 20 in FIG.
For the input image signal of the n-th field 20, based on the motion vector 10 input from the motion vector determination means 4,
The image screen is moved in parallel from 29 to 30 in the direction 21 for correcting the motion, and the screen is enlarged or enlarged by interpolation or the like, and image signals 27 and 28 of one screen size with little shaking are output. .

As described above, according to the present embodiment, when determining the reliability of the vector of each detection area, not only the state of the correlation value and the value of the vector in each detection area at that time, but also the past of the detection area. Since the determination is performed based on the determination result of the reliability of the vector and the value of the vector, first, as described in the first embodiment, a moving object having a size equal to or larger than the detection area enters the screen. Even in this case, the correction can be made if the motion vector of the video camera is detected from other areas.

Further, in the second embodiment, even when a moving object that occupies most or the whole of the entire screen passes through the screen, the correction is interrupted and the screen is corrected in a direction different from the direction intended by the photographer. Does not occur. FIG. 9 and FIG. 10 are diagrams for explaining this. FIG. 9 shows the state of the correlation value between the screen and each area when the moving object 13 having the same size as the screen passes through the screen, and FIG. 10 shows the motion vector 10 detected at that time. It shows. First, when a moving object 13 is present in a certain detection area 6 as shown in FIGS. 9 (a) and 9 (b), the area 6 has no reliability as in the case of FIG. A judgment is made (second judgment). Next, as shown in FIG. 9 (c), when the moving object 13 also enters the other area 12, that area is also determined to be unreliable, and all areas are determined to be unreliable. . In this case, a value calculated from the motion vectors in the previous field and the field immediately before the field is output as shown in the above equation. Since the above equation has a second-order attenuation characteristic, the output vector is 10 (c) in FIG.
As shown in the part of (d), it attenuates smoothly. Therefore, the shake correction at this time is also smoothly stopped. Next, as shown in FIG. 9 (d), when the moving object 13 occupies the entire area, a motion vector 15 obtained by adding the motion of the video camera and the motion of the moving object is detected from all the detection areas 6 and 12. ,
In addition, it is determined that there is reliability from the correlation value states 7 and 14 (first determination). However, since these vectors 9 and 16 generally have large values in one direction as shown at 15 in FIG. 7, they do not coincide with the output vector 10, so the second and third judgments are unreliable. Will be determined. Next, as shown in FIG. 10 (e), after the moving object has passed through a certain area 6, a motion vector 8 due to the shaking of the video camera is detected from the detection area 6 again, and the motion vector 8 is reliable from the state of the correlation value. It is determined that there is sex (first determination).
Generally, the motion vector due to the shaking of the video camera oscillates up and down around the 0 vector as shown in 8 of FIG. 10, so that it coincides with the output vector 10 at a certain time 11, and at this time, the second and third Is determined to be reliable, and thereafter, the vector of this area 6 is output as the motion vector 10, and the suspended correction is resumed. Therefore, the motion vector detected at the time points (a) to (e) in FIG. 9 is the motion vector 8 due to the shaking of the video camera when there is an area where no moving object is present as shown in FIG. Detected and shake correction can be correct,
Further, when a moving object is present in all areas, the correction is smoothly interrupted, and after the moving object passes, from the time 11 when the motion vector 8 due to the shaking of the video camera coincides with the output vector 10 within a certain range. The correction will be restarted. Therefore, compared with the shake correction device using the conventional motion vector detection device, the section where correct correction can be performed is longer,
Further, no malfunction occurs when a moving object is present in all areas.

In the first embodiment, the method of determining the reliability of the vector in each region has been described in detail. However, this is not necessarily the same, and only the state of the correlation value and the value of the vector at that time are used in the determination. Instead, it is easy to obtain the same effect as long as the previous state of the correlation value, the value of the vector, and the determination result are used together.

For example, substantially the same effect can be obtained by using a method as shown in the flowchart of FIG. Referring to FIG. 5, it is assumed here that the reliability determination result of the area is not reset even if the field changes. Then, it is assumed that all the initial values of the reliability determination result of each area are set to have reliability or no reliability. As a procedure, first, data of a vector value and a correlation value of each area are input.
Next, the state of the correlation value is determined, and if the condition does not meet the condition, the determination result is determined to be unreliable (determination 1). However, the reverse is not performed, that is, even if the state of the correlation value meets the condition, if the judgment result before this judgment is unreliable,
The judgment result is not changed. Next, the motion vector of the entire screen is determined from the vector of the area determined to be reliable at this time. The determination method may be the same as in the second embodiment. This motion vector is output and the next decision 2
Used for In judgment 2, the state of the correlation value satisfies the condition,
In addition, a region where the output motion vector matches the region vector is determined to be reliable. However, the reverse is not performed. This determination result remains without being reset even when the next field is entered. Using such a method,
When a moving object enters the screen, the operation is substantially the same as in the first and second embodiments.

Further, for example, in a region determined to be unreliable from the state of the correlation value at a certain point in time, an effect can be expected even if the determination result without reliability is maintained for a certain period of time.

In the first embodiment, the determination result is a binary value with or without reliability. However, the present invention can also be applied to a case where the reliability is represented by a weight by considering an intermediate value. In addition, although the match between the vector of each region and the output vector is checked in the determination 3, the match may be checked with a certain range.
This range may be changed according to camera work such as zooming and panning.

Further, in the second embodiment, when it is determined that all the areas have no reliability, as shown in the above equation, the following equation is used for the secondary attenuation.
Although the vector value predicted from the previous motion vector is output, this may be a primary attenuation equation or another equation. Alternatively, a zero vector may be output. In the first and second embodiments, the number of detection areas in the screen is four, but this may be other than this.

Next, a motion vector detecting device according to a third embodiment of the present invention will be described. The configuration of the motion vector detecting device of the third embodiment is almost the same as that of the conventional example, and can be described with reference to FIG. 13 which is a block diagram of the conventional example. However, the difference from the conventional example is that, as shown in FIG.
17 and 12 and 18 are provided so as to be shifted in the horizontal direction so as not to be arranged uniformly in the vertical direction. The reliability determining means 3 determines reliability only from the state of the correlation value at that time, as in the conventional example. Further, the motion vector determining means 4 outputs an intermediate value of the vector from the area determined to be reliable as a motion vector of the screen.

As described above, according to the present embodiment, when a plurality of detection areas are provided in a screen, they are shifted in the horizontal direction so as not to be uniformly aligned in the vertical direction, so that a long movement in the vertical direction as in the conventional example is achieved. When an object enters, it is possible to prevent erroneous vectors from being simultaneously detected from vertically arranged areas.

This will be described with reference to FIG. Fig. 11 (a)-
(E) {(c) is not shown in order to correspond to FIG. 7, and {} shows the state of the correlation values of the detection areas 6, 12, 17, and 18, respectively.
It is shown in 7,14,22,23. The images in (a) to (e) in this figure are the same as those in (a) to (e) in FIG. 7, respectively. In the conventional example, it is impossible to detect the motion vector in the state (b). On the other hand, in this embodiment, when a moving object enters the area 6 in the state (a), the area 6 is determined to be unreliable in the state of the correlation value 7 at that time. Since the motion vector 8 is detected in the other regions 12, 17, and 18 due to the camera shake, the motion vector determining means 4 outputs the motion vector 8 due to the camera shake. Next, in the state of (b), since the moving object occupies most of the area 6, the vector 15 to which the motion of the moving object is added is added.
Is detected, and the state of the correlation value 7 is determined to be reliable. However, at this time, the area 17 is determined to be unreliable based on the correlation value 22, and the motion vectors 8 are detected in the areas 12 and 18, and are determined to be reliable. Therefore, the motion vector determining means 4 takes the intermediate value of each of the three vectors of the three areas determined to be reliable, and outputs the motion vector 8 due to the camera shake. (D) ~
In the state (e), it is determined that there is no reliability from the state of the correlation value in the area where the moving object is applied, and in the other area, the motion vector 8 due to the camera shake is detected. Outputs a motion vector 8 due to camera shake. Therefore (a) ~
The motion vector 8 due to the camera shake is output over the state (e). From this, it can be seen that the present embodiment has a wider range in which the motion vector 8 due to camera shake can be detected for a moving object than the conventional example.
Further, it is of course possible to apply the first invention to this.

Next, a motion vector detecting device according to a fourth embodiment of the present invention will be described. The configuration of the motion vector detecting device of the fourth embodiment is almost the same as that of the motion vector detecting device of the first embodiment and the motion vector detecting device used in the second embodiment. This can be explained with reference to FIG. 2 which is a block diagram. However, the difference is that, as shown in FIG. 4, a plurality of detection areas 6 and 17 and 12 and 18 in the screen are shifted in the horizontal direction so as not to be aligned in the vertical direction.

As described above, according to the present embodiment, when a plurality of detection areas are provided on a screen, the detection areas are shifted in the horizontal direction so as not to be uniformly aligned in the vertical direction. When a long moving object enters in the vertical direction as in the example, it is possible to prevent erroneous vectors from being simultaneously detected from vertically arranged areas. Therefore, the following effects are newly produced as compared with the first and second embodiments. This will be described with reference to FIG. First, first, second
In the embodiment of FIG. 12 (b), the movement due to the camera shake without interruption is performed until a moving object having a size approximately equal to or slightly larger than the space 26 between the left and right regions passes. Vector 8 could be detected. But this interval 26
Is large, the motion vector cannot be detected in the case of an image having a pattern feature only at the center of the screen as shown in FIG. 12 (a).

In contrast, in the present embodiment, as shown in FIG.
The motion vector 8 due to camera shake can be detected without interruption until a moving object having a size equal to or slightly larger than the interval 24 between the rightmost region and the leftmost region passes.

At the same time, the interval 25 where there is no region when viewed from the vertical direction can be made narrower than the interval 26, so that the motion vector can be detected even in the case shown in FIG. 12 (c). From this, it can be seen that the range in which the motion vector 8 due to camera shake can be detected is wider than in this embodiment.

In the third and fourth embodiments, the respective detection areas in the screen are four areas each having an interval. However, this does not have an interval as shown in FIGS. And may overlap each other. Also, as shown in FIG. 6 (c), the number of regions may be shifted from the regions arranged in the horizontal direction, and this is effective when a moving object from the vertical direction passes. 6 (d), (e), (f)
As shown in FIG. Further, the motion vector detecting device according to the third or fourth embodiment can be applied to the shake correcting device, just as the motion vector detecting device according to the first embodiment is applied to the shake correcting device according to the second embodiment.

Advantageous Effects of the Invention According to the present invention, even when a moving object of the same degree or more enters each region, a vector that cannot be detected by the conventional device is selected from the detection vectors of each region. Only the motion vector due to the shaking of the video camera can be detected as the motion vector of the screen. Also, when a moving object of the same size or larger than the screen comes in, the detection of the vector with the movement of the moving object was erroneously detected in the conventional one, but the detection is automatically performed. The detection of the erroneous vector which is interrupted can be prevented. The shake correction device of the present invention is not suitable for practical use. Even if the possible range is large and exceeds it, malfunctions are greatly reduced. Further, the present invention can be easily realized with almost no increase in circuits and the like as compared with the conventional one, and therefore has a large practical effect.

[Brief description of the drawings]

FIG. 1 is a flowchart showing the operation of the vector reliability determining means and the motion vector determining means of the motion vector detecting devices of the first and second embodiments of the present invention, and FIG. 2 is a first embodiment of the present invention. FIG. 3 is a block diagram of a motion vector detecting device of an example image, FIG. 3 is a block diagram of a shake correcting device of a second embodiment of the present invention, and FIG. 4 is a diagram of an image of the third and fourth embodiments of the present invention. FIG. 5 is a diagram showing the arrangement of each detection area in the screen of the motion vector detection device, FIG. 5 is a flowchart showing an operation example of the reliability judgment means and the motion vector determination means of claim 1 of the present invention, and FIG. FIG. 7 is a diagram showing an example of the arrangement of each detection area in the screen of the image motion vector detecting device according to the second embodiment of the present invention. FIGS. 7 to 12 are lines for explaining the operation of the motion vector detecting device of the embodiment. FIG. 13 is a diagram showing the movement of a conventional image. FIG. 14 is a block diagram of a shake correction device to which a motion vector detection device is applied, FIG. 14 is a flowchart for explaining the operation of a conventional motion vector detection device, and FIG.
FIG. 16 is a diagram for explaining the operation of the conventional motion vector detecting device, and FIG. 17 is a diagram for explaining the motion correcting operation. 1 ... Image signal input terminal, 2 ... Correlation value calculation and vector detection means in each area, 3 ... Reliability determination means in each area, 4
…… Motion vector determination means, 5 …… Shake correction means, 6,12,1
7,18 …… Vector detection means, 13 …… Moving object, 7,14,22,23…
… Correlation value of each area, 8… Motion vector due to video camera shake, 15… Vector obtained by adding motion of moving object to motion vector due to video camera shake, 9,16 …… Detection vector of each area, 24,25,26 …… Interval of each detection area, 27,2
8: Correction screen, 29, 30 ... Readout position of image information.

Continuation of the front page (56) References JP-A-61-269475 (JP, A) JP-A-61-201584 (JP, A) JP-A-62-25590 (JP, A) JP-A-1-125074 (JP, A) , A)

Claims (4)

(57) [Claims] A plurality of detection areas are provided on a screen, and for each of the detection areas, means for calculating a correlation value between a current field and a previous field at a predetermined shift, and a motion for each detection area based on the correlation value. Means for obtaining a vector, and determining the reliability of each of the detection areas, wherein at least the result of determining the reliability of each frame is recorded, and reliability is determined from the state of the correlation value in each frame. Is changed from unreliable to reliable based on the state of the correlation value in each frame and the degree of coincidence of the detected motion vector with the motion vector of the entire screen. Means for determining the reliability of the vector of the motion of each detection area, and the motion of the entire screen using the motion vector of each area based on the reliability determination. Image motion vector detecting device characterized by comprising means for determining the vector. 2. A method according to claim 1, wherein a plurality of detection areas are provided in the screen so as to be shifted in the horizontal direction so as not to be arranged in the vertical direction, or are provided so as to be shifted in the vertical direction so as not to be arranged in the horizontal direction. Means for obtaining a correlation value at a predetermined shift in the detection area, means for obtaining a motion vector for each detection area from the correlation value, and means for determining the reliability of the motion vector of each detection area at this time. And a means for determining a motion vector of the entire screen by using a motion vector of each region based on the reliability determination there. 3. The image according to claim 1, wherein the detection areas are provided so as to be shifted in the horizontal direction so as not to be arranged in the vertical direction, or are provided so as to be shifted in the vertical direction so as not to be arranged in the horizontal direction. Motion vector detection device. 4. An image motion vector detecting device according to claim 1, wherein at least one field of image signal storage means, and a motion vector detected from the image motion vector detecting device. An image shake correction apparatus, further comprising means for controlling a reading position of the image signal storage means.