JPH04296176A - Blurring correction circuit - Google Patents

Abstract

PURPOSE:To reduce the storage capacity required for a memory. CONSTITUTION:A motion vector of a video signal between frames detected by a motion detection circuit 4 is sequentially written in latch circuits 5a, 5b, 5c and a motion vector written in the latch circuits 5a, 5b, 5c is fed to a motion vector estimating circuit 6, and the motion vector estimated by the motion vector estimating circuit 6 is fed to a write control circuit 7, in which the write of a memory 3 is controlled based on the estimated motion vector.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image stabilization circuit suitable for use in consumer video cameras and the like.

0002

2. Description of the Related Art For example, as consumer video cameras have become smaller and lighter, they are increasingly being used to take pictures with one hand. In this case, hand-held photography with one hand cannot avoid so-called camera shake, and images with such camera shake significantly impede good viewing.

[0003] Conventionally, therefore, devices have been proposed that correct camera shake by using a memory or the like to correct a photographed video signal. That is, FIG. 5 shows the configuration of the proposed conventional device. In this figure, 51 is an imaging section such as a CCD or an imaging tube. Here, the CC of this imaging unit 51
D or an image pickup tube or the like is provided with a sufficiently large imaging surface in consideration of the movement of camera shake, which will be described later. This imaging unit 5
The video signal photographed in step 1 is supplied to a memory 53 through an A/D converter 52.

Further, 54 is a counter, and this counter 54 counts arbitrary pixel clock signals and the like. The count value from this counter 54 is supplied to the memory 53. As a result, all pixels of the digitized video signal from the A/D converter 52 are written into the memory 53 using the count value from the counter 54 as a write address.

On the other hand, 55 is a motion detection circuit, and a digitized video signal from the A/D converter 52 is supplied to this motion detection circuit 55. For example, 1
The video signals for each frame are compared, and the motion vector of the video signal between these frames is detected. The motion vector of the video signal between frames detected by the motion detection circuit 55 is supplied to the readout control circuit 56. Reading of the memory 53 is then controlled based on the supplied motion vector, and the read video signal is taken out to the output terminal 57.

That is, in FIG. 6, the outer frame of the drawing indicates the range of the imaging surface provided on the CCD, imaging tube, etc. of the imaging section 51. On the other hand, all pixels of the video signal from this imaging surface are written in the memory 53, as indicated by diagonal hatching.

[0007] This memory 53 normally outputs a video signal within a range indicated by a broken line frame, for example. On the other hand, when a motion vector as shown by the arrow is detected by the motion detection circuit 55 in an arbitrary frame, reading from the memory 53 is controlled based on this motion vector. Then, in this arbitrary frame, reading from the memory 53 is performed so that a video signal within the range indicated by the dashed line frame is output.

Therefore, in this device, reading of the memory 53 is controlled so as to cancel the detected motion vector. As a result, a video signal with camera shake corrected can be taken out to the output terminal 57.

However, in this device, all pixels of the video signal from the imaging surface are written in the memory 53. Therefore, the memory 53 requires an extremely large storage capacity. Therefore, it is necessary to provide a large number of memory ICs, etc., making the configuration complicated and also making it difficult to control them.

0010

[Problem to be Solved by the Invention] The problem to be solved is that in order to write all the pixels of the video signal from the imaging surface into the memory in the conventional device, an extremely large storage capacity is required in the memory. It is something.

[0011]

[Means for Solving the Problems] The present invention provides a detection means (motion detection circuit 4) for detecting a motion vector between each predetermined period of a video signal photographed (imaging unit 1) and the previous predetermined period. The motion vector for the next predetermined period is estimated from one or more past motion vectors (circuit 6), and an arbitrary range in the captured video signal is stored in memory based on the estimated motion vector. (3) (control circuit 7
), and the video signal written in this memory is read out (control circuit 9) based on the motion vector detected by the detection means (arithmetic circuit 8) to obtain an output video signal (output terminal 10). ) is an image stabilization circuit designed to

0012

[Operation] According to this, the motion vector for the next predetermined period is estimated, and based on this estimated motion vector, only an arbitrary range of the photographed video signal is written to the memory. Storage capacity can be significantly reduced.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In FIG. 1, reference numeral 1 denotes an imaging section such as a CCD or an imaging tube. Here, the CCD or image pickup tube of the imaging unit 1 is provided with a sufficiently large imaging surface in consideration of the movement of camera shake, which will be described later. A video signal photographed by this imaging section 1 is supplied to a memory 3 through an A/D converter 2.

On the other hand, reference numeral 4 denotes a motion detection circuit, to which the digitized video signal from the A/D converter 2 is supplied. In this way, for example, video signals for each frame are compared, and a motion vector of the video signal between these frames is detected.

The motion vector of the video signal between frames detected by the motion detection circuit 4 is sequentially transmitted to the latch circuits 5a and 5.
b, 5c. These latch circuits 5a, 5b
, 5c are supplied to a motion vector estimation circuit 6, and the motion vector estimated by this motion vector estimation circuit 6 is supplied to a write control circuit 7. Then, writing to the memory 3 is controlled based on this estimated motion vector.

Further, this estimated motion vector and the motion vector of the interframe video signal detected by the motion detection circuit 4 described above are supplied to the arithmetic circuit 8 . In this way, the difference between the estimated motion vector and the actual motion vector is calculated. A vector of this difference is supplied to the read control circuit 9. Then, reading from the memory 3 is controlled based on the supplied difference vector. This read video signal is taken out to the output terminal 10.

That is, in FIG. 2, the outer frame of the drawing indicates the range of the imaging surface provided on the CCD or imaging tube of the imaging section 1. Here, the motion vector estimation circuit 6 estimates a motion vector as indicated by a broken line arrow, for example. As a result, as shown with diagonal hatching, the memory 3 stores a range that is larger by a predetermined width than the range of the video signal read out based on the estimated motion vector, taking into account the difference described later. A video signal is written.

On the other hand, when the motion detection circuit 4 detects a motion vector as indicated by the solid line arrow, a vector of difference as indicated by the dashed-dot line arrow is obtained from this motion vector and the above-mentioned estimated motion vector. Calculated. Reading from the memory 3 is then controlled based on this vector of differences. As a result, in this frame, the memory 3 is read out so that the video signal in the range indicated by the dashed-dotted line is output.

Therefore, in this device, for the detected motion vector, the estimated motion vector is stored in the memory 3.
The reading of the memory 3 is controlled using the differential vector, and the reading of the memory 3 is controlled so as to cancel out the detected motion vector as a whole. As a result, a video signal with camera shake corrected can be taken out to the output terminal 10.

[0020] Thus, according to the above-mentioned device, the motion vector for the next predetermined period is estimated (motion vector estimation circuit 6), and only an arbitrary range in the photographed video signal is estimated based on the estimated motion vector. Since the information is written to the memory (3), the storage capacity required for the memory can be significantly reduced.

That is, in the above-mentioned apparatus, screen shake due to hand shake is smooth to some extent in the time direction. Therefore, the next motion vector can be estimated relatively accurately from the past motion vector. Therefore, there is no need to consider very large differences in the video signals stored in the memory 3, and as is clear from the figure, the required storage capacity can be significantly reduced.

In the above-described apparatus, the motion vector can be estimated by approximating the temporal change in the motion vector using a quadratic function, as shown in FIG. 3, for example. Here, as a quadratic function, M(t)=At2 +Bt+C (A, B, and C are constants), and M at times t0, t1, and t2, respectively.
0, M1, M2, so at time t3 M
(t3) = 3M2 -3M1 +M0 (=M'3
).

Furthermore, FIG. 4 shows a circuit for realizing this. In this figure, among the values M0, M1, and M2 from latch circuits 5a, 5b, and 5c, M1 and M2 are tripled by multipliers 41 and 42, and added to and subtracted from M0 by subtracter 43 and adder 44, respectively. Then, 3M2 -3M1 +M0 =M'3 is obtained.

[0024]Although the above explanation is given only for one dimension, in reality, processing is performed for two dimensions in the horizontal and vertical directions. Also past 1, 2, or 4
Two or more motion vectors may be latched and estimated using a function other than a quadratic function, such as a linear function, a trigonometric function, or an average value.

[0025]

According to the present invention, the motion vector for the next predetermined period is estimated, and only an arbitrary range of the photographed video signal is written to the memory based on the estimated motion vector. It has become possible to significantly reduce the storage capacity required.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram of an example of an image stabilization circuit according to the present invention.

FIG. 2 is a diagram for explaining the same.

FIG. 3 is a diagram for explaining motion vector estimation.

FIG. 4 is a configuration diagram of an example of the specific circuit.

FIG. 5 is a configuration diagram of a conventional circuit.

FIG. 6 is a diagram for explaining the same.

[Explanation of symbols]

1 Imaging unit 2 A/D converter 3 Memory 4 Motion detection circuit 5a, 5b, 5c latch circuit 6 Motion vector estimation circuit 7 Write control circuit 8 Arithmetic circuit 9 Read control circuit 10 Output terminal

Claims (1)

[Claims] 1. A detection means for detecting a motion vector between each predetermined period of a photographed video signal and the previous predetermined period, and detecting a motion vector from one or more past motion vectors to the next predetermined period. estimating a motion vector, writing an arbitrary range in the photographed video signal into a memory based on the estimated motion vector, and detecting the video signal written into the memory by the detection means. An image stabilization circuit that obtains an output video signal by reading out the motion vector based on the detected motion vector.