JPH08340534A - Video signal processor - Google Patents

Abstract

PURPOSE: To attain high image quality for part of an image even in a coding system in which a prescribed code quantity is allocated in the unit of segments. CONSTITUTION: A SCT circuit 12 applies discrete cosine transformation to a digital video signal from an input terminal 10 in the unit of 8×8 picture elements. A block processing circuit 14 collects four adjacent 8×8 transformation coefficient data into one block on an image and a segment processing circuit 16 five blocks separated on the image into one segment. A quantization circuit 18 quantizes DCT coefficient data in the unit of segments. A setting partial address generating circuit 20 provides an output of an address denoting a specific area on the image whose image quality is desired to be high, the circuit 18 selects a smaller quantization step of the specific area on the image than a quantization step of other parts according to an output of the circuit 20. A variable length coding circuit 22 applies variable length coding to the output of the quantization circuit 18.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal processing device for data-compressing a digital video signal.

[0002]

2. Description of the Related Art A digital video transmission system for compressing the data amount of a digital video signal and transmitting at a reduced bit rate (for example, recording on a digital recording medium) is well known.

FIG. 8 shows a schematic block diagram of an encoding processing unit in a conventional video transmission system. Input terminal 1
The digital video signal sampled and quantized at the sampling frequency fs is input to the digital signal 10. The discrete cosine transform (DCT) circuit 112 performs orthogonal transform on the digital video signal from the input terminal 110, for example, discrete cosine transform in units of 8 × 8 pixel blocks. The 8 × 8 coefficient data output from the DCT circuit 112 is applied to the blocking circuit 114. Blocking circuit 114
Is a number of adjacent 8x8 (eg 4) on the screen
The coefficient data is collected in one block, and the segmenting circuit 116 distributes some (eg, five) on the screen.
The blocks of are picked up according to a certain rule and are combined into one segment.

FIG. 9 shows the relationship between blocks formed by the blocking circuit 114 and segments formed by the segmenting circuit 116. The blocking circuit 114 includes four 8 × 8 DCT coefficient data 132, 134, 1 arranged horizontally on the screen 130.
36 and 138 are combined into one block 140. The segmenting circuit 116 is divided into five blocks 14 by the blocking circuit 114 and dispersed on the screen.
0, 142, 144, 146, 148 are combined into one segment.

The DCT coefficient data grouped into segments by the segmentation circuit 116 is quantized by the quantization circuit 118.
Is quantized according to a predetermined quantization step. The quantization step can be set for each component of the DCT coefficient. The variable length coding circuit 120 includes a quantization circuit 118.
Variable length coded, and the output is recorded on a magnetic tape, a magnetic disk, an optical disk, or the like.

When the image data encoded in this way is recorded on a magnetic tape or the like, it is desirable that the code amount for each screen is constant. Under such a demand, the quantization circuit 118 and the variable length coding circuit 120 are set so that the code amount for each segment is constant. Since one segment consists of five blocks, the quantization step of the five blocks in the quantization circuit 118 is controlled so that the code amount of the segment after the variable length coding process becomes constant.

A method of controlling the quantization step will be described with reference to FIG. First, one of the quantization step selection numbers 1 to 4 is selected for each block in the segment.
The larger the quantization step selection number is, the larger the quantization step is (that is, the size of the step is increased, the number of steps is decreased, and as a result, the image becomes rough). According to the selection of the quantization step selection number, the quantization step applied to each of the four 8x8 DCT coefficients in the block is determined. 4 8x8 DC
The quantization step for each of the T coefficients is 8 × 8
The value varies depending on the content of the DCT coefficient, in other words, the content of the 8 × 8 pixel image converted into the DCT coefficient. That is, the applied quantization step depends on the quantization step selection number and the DCT transform coefficient value. If the values of the four 8x8 DCT coefficients are exactly the same, then the quantization steps will be the same. Furthermore, the quantization step may be set to a different value for each frequency component in the 8 × 8 DCT coefficient.

Next, a quantization step selection number is set for each block in the segment, all DCT coefficients in the segment are quantized, and then variable length coding is performed. When the code amount of the segment after the variable length coding does not fall within a predetermined fixed amount, the quantization step selection number set for each block in the segment is incremented to coarsen the quantization step. The segment is quantized and variable-length coded by a new quantization step, and the code amount of the segment is examined. By repeating this process, the code amount of the segment is set to a certain value or less.

By the way, when the target value (upper limit) of the code amount of the segment is h, the code amounts of the five blocks in the segment do not necessarily have to be the same amount, that is, h / 5. It may be more or less than h / 5 depending on the content of the 8 × 8 DCT coefficient. In other words, it is sufficient if the amount becomes a certain amount as a segment.

By the above processing, the code amount of the segment becomes a value less than or equal to the target value and close to the target value.
The code amount on the screen is also less than or equal to the target value for one screen.

[0011]

However, in the conventional example,
Since the quantization step of each block in the segment is determined by the contents of the four 8 × 8 DCT coefficients forming each block, in other words, the picture pattern of the image, more codes cannot be assigned to a part of the screen. It was That is, in the conventional example, it is impossible to perform data compression while keeping the code amount of one screen constant (or less) and giving priority to the image quality of a specific portion or an arbitrarily selected portion in the screen.

It is an object of the present invention to provide a video signal processing device for quantizing a specific portion or an arbitrarily selected portion in a screen with a finer quantization step than the other portions.

[0013]

A video signal processing device according to the present invention is a video signal processing device for sampling and coding a screen of a video signal at a fixed density, and a coding rate of a predetermined portion in the screen. Is set to be lower than the coding rate of the portion other than the predetermined portion.

The video signal processing apparatus according to the present invention also divides the screen data of the digital video signal subjected to the orthogonal transformation into a plurality of blocks each having a constant data amount, and stores the blocks in different places on the screen. N the block
A video signal processing device for collecting data of the n blocks so that the code amount of the segments is set to a predetermined code amount by collecting the individual pieces to form one segment. Among the blocks, the quantization step for the data of the blocks included in the predetermined range within the screen is set more finely than the quantization step for the data of the other blocks.

The video signal processing device according to the present invention also includes an orthogonal transformation means for orthogonally transforming image information, and a predetermined number of transform coefficient data distributed on the screen, which is output from the orthogonal transformation means, into one segment. And a variable length coding means for variable-length coding the output of the quantizing means, the variable-length coding means for quantizing the segment data output from the rearranging means. A video signal processing device in which the quantizing step of the quantizing means is controlled so that the code quantity for each segment after coding by the coding means is equal to or less than a predetermined code quantity, wherein the quantizing means is a screen. It is characterized in that the data belonging to the predetermined area above is set with a finer quantization step than the data belonging to the area other than the predetermined area.

The predetermined range can be arbitrarily selected.

[0017]

By the above means, it becomes possible to quantize a preset specific portion or an arbitrarily selected portion on the screen with a finer quantization step than the other portions. As a result, a specific portion or an arbitrarily selected portion can be image-compressed so as to have higher image quality than other portions.

[0018]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a schematic block diagram of an embodiment of the present invention. The input terminal 10 receives the quantized digital video signal sampled at the sampling frequency fs. Discrete cosine transform (DCT) circuit 12
Performs orthogonal transform of the digital video signal from the input terminal 10, for example, discrete cosine transform in units of blocks of 8 × 8 pixels. The blocking circuit 14 and the segmenting circuit 16 are the blocking circuit 114 and the segmenting circuit 11.
Operate in the same manner as 6 to collect some (for example, four) 8 × 8 coefficient data adjacent to each other on the screen into one block and distribute them on the screen (for example, five)
The blocks of are picked up according to a certain rule and are combined into one segment.

The quantizing circuit 18 quantizes the DCT coefficient data collected by the segmenting circuit 16 in segment units. At that time, according to the address of the setting portion supplied from the setting portion address generation circuit 20, the output of the circuit 16 is controlled so that the setting portion on the screen, that is, the specific portion has high image quality and the other portions have low image quality. Quantize. Details of the operation of the quantization circuit 18 will be described later. The variable length coding circuit 22 variable length codes the output of the quantization circuit 18, and the output is recorded on a magnetic tape, a magnetic disk, an optical disk, or the like.

In the present embodiment, in a specific area set on the screen, the quantization step is reduced to obtain a high image quality, and in other areas, the quantization step is increased,
The image quality should be lower than that of the specific area. In this way, the code amount of each segment is controlled to a constant value (below) while providing the image quality difference between the specific area and the other areas.

The basic principle of this embodiment will be described with reference to FIG. 24, 26, 28, 30, 32 are blocks formed by the blocking circuit 14, and the segmenting circuit 16 puts these blocks 24 to 32 into one segment. Further, the target value of the code amount of one segment is 100. Here, if the entire screen has a uniform pattern, D included in blocks 24 to 32
It is assumed that all CT coefficients have the same value. At this time, in data compression by quantization and variable length coding in the conventional example,
The code amount after compression of each of the blocks 24 to 32 is the same for all 20.

It is assumed that the information is left as a high-definition image as high as possible in the part of the preset area 34 in the screen, and the other part may be the information to the extent that the pattern can be seen.
The set partial address generation circuit 20 supplies the address indicating the specific area 34 to the quantization circuit 18. The quantizing circuit 18 includes a block (block 3 in FIG. 2) included in the specific region 34 when quantizing the DCT transform coefficient data in segment units supplied from the segmenting circuit 16.
For 0), the DCT coefficient data is quantized in a fine quantization step, and the block not included in the specific region (see FIG. 2).
Then, the blocks 24, 26, 28, 32) are quantized with a coarse quantization step. Assuming that the target value of the code amount of the blocks in the specific area after the variable length coding is 40, in the example shown in FIG. 2, the code amounts of the blocks 24 to 28 and 32 are 15 Must.

FIG. 3 shows the flow of quantization and variable length coding processing in this embodiment. Of the five blocks of the segment, the block (block 30 in FIG. 2) included in the specific area 34 is first quantized (S1). At this time, the smallest quantization step among several preset quantization steps is used. For example, if the quantization steps prepared in advance are four types of 1, 2, 4, and 8, 1 is used.

Next, the blocks (blocks 24, 26, 28, 32 in FIG. 2) that do not belong to the specific area 34 are also quantized (S2). For the quantization step here, 1 is used first.

When the quantization of all blocks of the segment is completed, variable length coding is performed (S3). The code amount of the segment after the variable length coding is checked (S4), and if it is less than or equal to a predetermined target value, the process ends as it is, and if it exceeds the target value, the quantization step is increased (S5), and the specific region Blocks not belonging to 34 (in FIG. 2, blocks 24, 26, 2
8, 32) are quantized again (S2). Again, the segment is subjected to variable length coding to check whether the code amount is less than or equal to the target value (S4). In this way, the quantization steps of blocks that do not belong to the specific area 34 are sequentially increased until the code amount of the segment becomes equal to or less than the target value.

Although the quantization step of the block which does not belong to the specific area is successively increased, the quantization step is changed by the binary search method so that the code amount of the segment is less than or equal to the target value and It is also possible to detect the quantization step closest to Further, the quantization of blocks that do not belong to the specific area may be performed simultaneously in a plurality of quantization steps. Then, the most appropriate quantization step can be quickly determined.

In the conventional example, the DCT included in each block
The quantization step of each block is set to a different value only depending on the difference in the component of the coefficient.
Even if the CT coefficient values are the same, quantization is performed at different quantization steps depending on whether or not the CT coefficient value is included in a specific area in the screen. That is, in order to make the code amount of the segment constant,
In the same segment, a block included in a specific area is quantized by a fine quantization step, and a block not included in the specific area is quantized by a coarse quantization step. In this way, after the variable length coding, the specific area 3
A larger code amount is assigned to the block included in 0.

The specific area 34 has (x1, y1) and (x
2, y2), the setting portion address generating circuit 20 outputs (x1, y1) and (x
2, y2), or the quantizing circuit 18
An identification signal indicating whether or not the block to be quantized in (1) is included in the specific region is supplied to the quantization circuit 18.

In this way, in this embodiment, the data in the image in the specific area 30 can be compressed so that the image quality is higher than that in the area outside the specific area 30. It is obvious that the specific area 30 can be set at any position and size on the screen. For example, a subject tracking system may be incorporated and the specific region 30 may be automatically moved on the screen so that the subject or a part thereof has high image quality as the subject moves on the screen.

Further, as an example of giving priority to the image quality of an arbitrarily selected portion on the screen, for example, when a subject is photographed by a video camera or the like having a line-of-sight input device incorporated therein, a photographer can be displayed on the finder screen. There is a case where the part to be gazed at is captured and recorded in high definition.

The principle of line-of-sight detection for detecting the gazing point on the finder will be briefly described. FIG. 4 is a plan view showing the positional relationship between the eyeball and the optical system for detecting the line of sight, and FIG. 5 is a side view thereof. 40 is an eyeball, 42 is an iris, and 44 is a cornea.
Reference numerals 46a and 46b denote infrared light emitting diodes for irradiating the eyeball 40 with infrared rays, and 48 denotes an image forming lens for forming an image of the light rays emitted from the light emitting diodes 46a and 46b and reflected by the eyeball 40 on the image sensor 50. The light emitting diodes 46a and 46b are arranged substantially equidistant from each other in the x direction (horizontal direction) with respect to the optical axis of the imaging lens 48, and are arranged slightly below in the y direction (vertical direction).

FIG. 6 shows the image sensor 5 in (a).
A schematic diagram of an eyeball image projected on 0 is shown, and an output intensity diagram on a predetermined line is shown in (b).

First, considering the horizontal plane, in FIG.
The infrared light emitted from the light emitting diode 46b is emitted by the eyeball 4
Illuminate 0 cornea 44. At this time, the corneal reflection image d (virtual image) formed by the infrared light reflected on the surface of the cornea 44.
Is imaged at a position d ′ on the image sensor 50 by the imaging lens 48. Similarly, the infrared light emitted from the light emitting diode 46 a also illuminates the cornea 44 of the eyeball 40.
At this time, the cornea reflection image e (virtual image) formed by the infrared light reflected on the surface of the cornea 44 is imaged at the position e ′ on the image sensor 50 by the imaging lens 48. Iris 4
The two ends a and b are located in the imaging lens 48 and are imaged at the positions a ′ and b ′ on the image sensor 50.

Since the iris 42 is ring-shaped, it becomes a ring on the image sensor 50, but if the rotation angle θ of the optical axis of the eyeball 40 with respect to the optical axis of the imaging lens 48 is so small as to be negligible, a simple least squares method is used. Due to the center of the pupil 52 (x
xc of (c, yx) can be calculated.

On the other hand, the x coordinate of the center of curvature O of the cornea 44 is x
If o, the rotation angle θx of the optical axis of the eyeball 40 with respect to the front direction is D × sin θx = xc−xo (1). However, D is the center of curvature O of the cornea 44 and the iris 42.
And the distance.

Further, when xo is calculated in consideration of a predetermined correction value δx at the midpoint k of the corneal reflection image d and the same e, xk = (xd + xe) / 2 xo = (xd + xe) / 2 + δx (2) . xk is the x coordinate value of the midpoint k. Note that δx
Is a numerical value obtained geometrically from the installation position of the line-of-sight detection device, the eyeball distance, etc., and the calculation method is omitted.

By substituting the expression (1) into the expression (2), θx can be obtained as in the following expression. That is, θx = arcsin {(xc-((xd + xe) / 2 + δ
x)) / D} This is converted on the image sensor 50. image·
When the coordinates of each feature point projected on the sensor 50 are described by adding ', they are as follows. That is, θx = arcsin {(xc ′-((xd ′ + xe ′) / 2+
δx ′)) / D / β} where β is a magnification determined by the distance sze of the eyeball 40 with respect to the imaging lens 48, and is actually a function of the interval | xd′−xe ′ | of corneal reflection images. Desired.

On the vertical plane, the arrangement is as shown in FIG. The corneal reflection images due to the irradiation light of the light emitting diodes 46a and 46b are at the same position on the vertical plane, and this is designated as i. The method of calculating the rotation angle θy of the eyeball 40 is almost the same as the rotation angle θx of the horizontal plane, but only the formula (2) is different. That is, assuming that the y coordinate of the center of curvature O of the cornea 44 is yo, yo = yi + δy (5) where δy is a numerical value geometrically determined from the position of the visual axis detection device, the eyeball distance, and the like, and the calculation method is Omit it. The vertical rotation angle θy is given by the following equation. That is, θy = arcsin {(yc ′-(yi ′ + δy ′)) / D / β} (6)

The positional coordinates (xn, yn) on the viewfinder screen of the video camera satisfy the relationship of the following equation on the horizontal plane and the vertical plane, respectively, when the constant m determined by the viewfinder optical system is used. That is, xn = m × arcsin {(xc ′ 1 ((xd ′ + xe ′) / 2 + δx ′)) / D / β} (7) yn = m × arcsin {(yc ′ 1 ((yi ′ + δy ′))) / D / β} (8)

As is apparent from FIG. 6A, the edge of the pupil 52 can be detected by utilizing the falling edge (xb ') and the rising edge (xa') of the image sensor output waveform. Further, the corneal reflection image has a sharp rising edge (xe ′ and xd ′).

FIG. 7 shows a second embodiment of the present invention which utilizes line-of-sight detection.
The schematic block diagram of an Example is shown. Quantization circuit 54
In accordance with the output of the line-of-sight detection device 56, the vicinity of the point of interest of the photographer is specified as a specific area, and similarly to the quantizing circuit 18, it is quantized by a small quantization step, and the area other than the specified area is quantized by a coarse quantization step. Quantize. The line-of-sight detection device 56 supplies the quantization circuit 54 with a coordinate signal on the viewfinder screen that indicates the point of interest of the photographer.

In this way, the portion of the photographer's arbitrary attention can be data-compressed so as to have higher image quality than the other portions. That is, it is possible to perform data compression in which the image quality of an arbitrarily selected portion on the screen is prioritized.

[0044]

As can be easily understood from the above description, according to the present invention, in the image transmission system in which the data of each part in the screen is interleaved and the data is compressed, the specific area in the screen is The data can be selectively compressed so that the image quality is higher than that of other parts. The specific area can be set arbitrarily. As a result, the image quality of the desired portion can be prioritized over the image quality of other portions.

[Brief description of drawings]

FIG. 1 is a schematic block diagram of a first embodiment of the present invention.

FIG. 2 is an example of correspondence between blocks and segments and code amounts in the present embodiment.

FIG. 3 is an operation flowchart of the present embodiment.

FIG. 4 is a layout view in the horizontal plane of the gaze detection principle.

FIG. 5 is an explanatory diagram of the principle of line-of-sight detection in a vertical plane.

FIG. 6 is an eyeball image projected on the image sensor 50 and an output intensity diagram on a predetermined line.

FIG. 7 is a schematic configuration block diagram of a second embodiment of the present invention using eye-gaze detection.

FIG. 8 is a schematic block diagram of a conventional image compression device.

FIG. 9 is a relationship diagram between blocks and segments.

FIG. 10 is an explanatory diagram of quantization in the conventional example.

[Explanation of symbols]

 10: Input terminal 12: Discrete cosine transform circuit 14: Blocking circuit 16: Segmentation circuit 18: Quantization circuit 20: Setting partial address generation circuit 22: Variable length coding circuit 24, 26, 28, 30, 32: Block 34: specific region 40: eye 42: iris 44: cornea 46a, 46b: infrared light emitting diode 48: imaging lens 50: image sensor 52: pupil 54: quantization circuit 56: line-of-sight detection device 110: input terminal 112: discrete Cosine transform circuit 114: Blocking circuit 116: Segmenting circuit 118: Quantizing circuit 120: Variable length coding circuit 132, 134, 136, 138: DCT coefficient data 140, 142, 144, 146, 148: Block

Claims (6)

[Claims] 1. A video signal processing device for sampling and coding a screen of a video signal at a fixed density, wherein a coding rate of a predetermined portion in the screen is lower than a coding rate of a portion other than the predetermined portion. A video signal processing device characterized by setting. 2. The screen data of the digital video signal subjected to the orthogonal transformation is divided into a plurality of blocks each having a fixed data amount, and n blocks at different positions on the screen are collected to form one block. A video signal processing device that configures one segment and compresses the data of the n blocks so that the code amount of the segment fits in a predetermined code amount, wherein n of the blocks in the segment are: A video signal processing device, wherein a quantization step for data of blocks included in a predetermined range in the screen is set more finely than a quantization step for data of other blocks. 3. The video signal processing device according to claim 2, wherein the predetermined range is arbitrarily selectable. 4. An orthogonal transformation means for orthogonally transforming image information, a rearrangement means for collecting a predetermined number of transform coefficient data distributed on the screen, which is output from the orthogonal transformation means, into one segment, and the rearrangement. The quantization means for quantizing the segment data output from the means, and the variable length coding means for variable length coding the output of the quantizing means. A video signal processing device in which the quantization step of the quantization means is controlled so that the code amount for each segment is equal to or less than a predetermined code amount, wherein the quantization means is for data belonging to a predetermined area on the screen. A video signal processing device, wherein a quantization step finer than that of data belonging to an area other than the predetermined area is set. 5. The re-arrangement means is a conversion means for converting the conversion coefficient data output from the orthogonal conversion means into a predetermined number of adjacent conversion coefficient data on the screen, and the block conversion means. The video signal processing device according to claim 5, further comprising: segmenting means for collecting a predetermined number of block data distributed on the screen into the segments, the block data being output from the means. 6. The video signal processing device according to claim 4, wherein the predetermined range is arbitrarily selectable.