JPH06217294A - Image data compressing device and image data transmitting method - Google Patents

Abstract

PURPOSE:To efficiently compress dynamic image data. CONSTITUTION:A motion vector detecting circuit 3 detects the motion vector of dynamic image, data and a mean value arithmetic circuit 4 calculates the mean value of data along the motion vector. A comparing circuit 5 compares the mean value with image data stored in a memory 2 and an extracting circuit 6 extracts a range whose error with the mean value is small as a dynamic image area and an area whose error is large as a still picture area. An encoder 7 performs encoding by methods suitable for the respective areas.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression apparatus and an image data transmission method suitable for use in compressing and transmitting moving image data, for example.

[0002]

2. Description of the Related Art The MPEG system is known as a system for compressing and transmitting moving image data. In this method, a motion vector of an image between frames is detected, a predicted image is generated using this motion vector, difference data between the predicted image and a reference image is generated, and this is DCT-converted and transmitted. I am trying. As a result, the amount of data can be compressed.

[0003]

[Problems to be Solved by the Invention] However, MPE
The G method can compress the data of an image having a relatively small motion, but the correlation between frames is low for a moving image, so that sufficient prediction can be performed. However, there is a problem that the compression efficiency becomes poor.

The present invention has been made in view of such a situation, and is to enable data to be effectively compressed even in a moving image.

[0005]

An image data compression apparatus according to a first aspect of the present invention is a motion vector detecting circuit 3 as a detecting means for detecting a motion of an image, and a motion vector detecting circuit 3.
Along the direction corresponding to the motion vector detected by
An average value calculation circuit 4 as a calculation means for calculating the average value of the image and a comparison circuit as a comparison means for comparing the image and the average value calculated by the average value calculation circuit 4 and outputting a signal corresponding to the error. 5 and an extraction circuit 6 as an extraction means for extracting a range in which the error of the image is relatively small.

The comparison circuit 5 can calculate the variance as an error from the image and the average value.

The image data transmission method according to claim 3 is
The motion of the image is detected, the average value of the image is calculated according to the detected motion, the error between the image and the average value is calculated, and the range in which the error between the average value of the image is relatively small is set as the moving image area. In addition to the extraction, a relatively large range is extracted as a still image area, and the data of the moving image area and the data of the still image area are distinguished and transmitted.

This movement can be detected by utilizing affine transformation.

[0009]

In the image data compression apparatus according to the first aspect and the image data transmission method according to the third aspect, for example, the error between the average value calculated along the direction corresponding to the motion vector and the image is compared. A relatively small range is extracted as a moving image region, and a relatively large range is extracted as a still image region. Therefore, a preferable coding method can be used for each area, and more efficient data compression can be performed.

[0010]

1 is a block diagram showing the configuration of an embodiment of an image data compression apparatus of the present invention. To the A / D converter 1, moving image data to be encoded (transmitted) is input. The A / D converter 1 A / D-converts this data, outputs it to the memory 2, and stores it. The moving image data stored in the memory 2 is read at a predetermined timing and supplied to the motion vector detection circuit 3, the average value calculation circuit 4, the comparison circuit 5, and the extraction circuit 6, respectively.

The motion vector detection circuit 3 detects the motion vector of the data stored in the memory 2. This motion vector is supplied to the average value calculation circuit 4 and the comparison circuit 5. The average value calculation circuit 4 calculates the average value of the image data stored in the memory 2 according to the motion vector detected by the motion vector detection circuit 3. Comparison circuit 5
Calculates the error between the average value calculated by the average value calculation circuit 4 and the image data stored in the memory 2, and outputs the calculation result to the extraction circuit 6.

The extraction circuit 6 extracts a range having a small error as a moving image area corresponding to the comparison result of the comparison circuit 5. Then, the data on the extracted range is supplied to the memory 2, the data excluding the extracted range (masked) is read from the memory 2, and the same operation is executed by each circuit.

The above processing is repeated to divide the moving image area and the still image area. And encoder 7
Encodes the data of the moving image area and the data of the still image area with an encoding method suitable for each,
It is transmitted to a circuit (transmission path) not shown.

Next, the operation will be described in detail with reference to the flow chart of FIG. First, in step S1, the movement of the data stored in the memory 2 is detected. Then, in step S2, the average value of the image data is calculated along the movement detected in step S1.

Now, for example, as shown in FIG.
It is assumed that 5 frames of image data of F _{1 to} F ₅ are stored in. The image of each of these frames is an image in which the object A and the object B are moving. These images correspond to A ₁ in each frame F _{1 to} F ₅ of FIG.
Through A ₅ and B ₁ through B ₅ . Object A
Are sequentially moving in the order of A _{1 to} A ₅ , and the object B is B
_It moves in order from _{1 to} B ₅ .

The data in the x-axis direction at a predetermined coordinate position (y = y _j ) on the y-axis of this image data is extracted and shown in FIG. In the figure, the vertical axis i
Indicates the magnitude of brightness. From the figure, it can be seen that the objects A and B are sequentially moving.

Further, a plan view of this figure is as shown in FIG. In the figure, the horizontal axis is the x-axis and the vertical axis is the time axis. In the figure, the range indicated by the shaded narrow hatching corresponds to the movement of the object A, and the range indicated by the coarser hatching corresponds to the movement of the object B.

Then, for example, calculating the average value along with the movement of the object A is performed in step S1 as shown in FIG.
A line L ₁ parallel to the line M along the motion vector detected by
It means to calculate the average value of the pixel data on L ₂ .

This average value A _ia is the luminance level of the point (x, y) at i (x, y, t), and pixel data cut at a predetermined position on the y axis is At _{, i} (t = 0, 1, 2, ... n,
i = 0, 1, 2, ... M) (n is the number of frames to be compressed, m is the total number of pixels in the original image), and the motion vector is v
Let (t) be expressed by the following equation. A _ia = (1 / n _in ) Σ (A _{t, i} + v (t))

Here, n _in is the number of effective pixels on the straight line L along the motion vector, and is equal to n if the point i does not protrude from the screen and does not pass through the masked range. When the motion vector can take a real number, the average value can be calculated after correction with an appropriate filter.

Next, in step S3, the average value calculated in step S2 is compared with the original image data. For example, the data included in the moving range of the object A in FIG. 6 is only the image of the same object that changes with time, so that the luminance i is almost constant regardless of the change in time t. It should be. On the other hand, since the background image data is an image unrelated to the object A, the background image data has a value that is significantly different from the average value obtained along the movement. Therefore, data having a small difference from the average value can be considered as a range of moving image data, and data having a large error can be considered as data of a still image area (background image).

The squared error matrix E (dispersion) shown in the following equation can be used for the evaluation of this error. When E _i = Σ (At _{, i} −A _ia ) ² E _i is smaller than a predetermined threshold value, it is determined to be a moving image, and when it is large, it is determined to be a still image.

Therefore, in step S4, a portion having a small error is extracted as a moving image area. As a result, for example, the data in the range shown with a narrow shadow in FIG.
Extracted (masked) from the original image.

Further, in step S5, the time axis (see FIG.
The variance of the pixel data is calculated along the vertical axis in, for example, along the straight lines L ₃ and L ₄ . At this time, the range extracted as the moving image area in step S4 is excluded (masked) from the calculated pixel data. When there is no moving part in the masked image data (only the background image), this variance is smaller than a preset threshold value. On the other hand, if there are still moving parts,
The value of this variance is greater than the threshold.

In this way, when there is still moving image data, the process returns from step S6 to step S1 and the same processing is repeated. Then, when it is determined in step S6 that the variance value is smaller than the preset reference value, the process proceeds to step S7, and encoding is performed for each data of the extracted area. In the present case, as shown in FIG. 7, a range a corresponding to the movement trajectory of the object A, a range b corresponding to the movement trajectory of the object B, and a background C.
The data is divided into three areas of the range c corresponding to. The encoder 7 encodes the data of the moving image area (a, b) by a predetermined method capable of encoding it most efficiently, and encodes the data of the still image area most efficiently. It is encoded by a predetermined method that enables

In the above embodiment, the motion vector is used to detect the motion between frames.
It is also possible to detect motion by affine transformation.
For example, as shown in FIG. 8A, a triangle on the xy plane is calculated by the following equation,
As shown in (b), affine transformation can be performed on the XY plane.

[0027]

[Equation 1]

For the motion vector detection, a block matching method, a method of detecting a zero crossing and obtaining a motion vector on the zero crossing, a method of using an optical flow and the like can be considered. Further, the evaluation of the error can be performed by weighting the error by the frequency in consideration of the comparison characteristic.

[0029]

As described above, according to the image data compression apparatus of the first aspect and the image data transmission method of the third aspect, the motion of the image is detected and the image is divided into the moving image area and the still image area. Then, since each of them is separately encoded and transmitted, the data can be efficiently compressed.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of an image data compression apparatus of the present invention.

FIG. 2 is a flowchart illustrating the operation of the embodiment of FIG.

FIG. 3 is a diagram illustrating movement of an image.

FIG. 4 is a diagram illustrating data at predetermined coordinate points on the y-axis in FIG.

5 is a plan view of the data of FIG.

FIG. 6 is a diagram illustrating calculation of an average value corresponding to a motion vector.

FIG. 7 is a diagram illustrating a range divided into a moving image area and a still image area.

FIG. 8 is a diagram illustrating affine transformation.

[Explanation of symbols]

 1 A / D converter 2 Memory 3 Motion vector detection circuit 4 Average value operation circuit 5 Comparison circuit 6 Extraction circuit 7 Encoder

Claims (4)

[Claims] 1. Comparing a detection means for detecting a motion of an image, a calculation means for calculating an average value of the image along the motion detected by the detection means, and an image and an average value calculated by the calculation means. Then
An image data compression apparatus comprising: a comparison unit that outputs a signal corresponding to the error, and an extraction unit that extracts a range in which the error of the image is relatively small. 2. The comparing means calculates a variance from the image and an average value as the error.
The image data compression device described in 1. 3. A range in which a motion of an image is detected, an average value of the images is calculated along with the detected motion, an error between the images is calculated, and an error between the average value of the images is relatively small. Is extracted as a moving image area, a relatively large range is extracted as a still image area, and the data of the moving image area and the still image area are separately distinguished and transmitted. 4. The image data transmission method according to claim 3, wherein the movement is detected by using an affine transformation.