JP2000092454A - Device and method for image information conversion - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the conversion performance of an image information conversion process, especially as regards to a moving part. SOLUTION: According to the output of an area segmentation circuit 4, a movement class determining circuit 5 detects a movement quantity by a block matching method, etc. Furthermore, the movement class determining circuit 5 determines a movement class according to the detected movement quantity. In this case, basically the movement class represents whether or not there is movement. Different from a conventional movement class, this class represents a more detailed movement. This movement class is supplied to a class code generating circuit to contribute to the generation of a final class, so that more detailed division as to a movement becomes possible and output image signal can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image information conversion apparatus and an image information conversion method having a function of generating an output image signal having a higher resolution from an input image signal.

[0002]

2. Description of the Related Art With the rise of audio-visual orientation, there has been a demand for the development of a television receiver so as to obtain a higher-resolution image, and in response to this demand, a so-called Hi-Vision has been developed. In Hi-Vision, the number of scanning lines is 1125, which is N
The number of scanning lines is twice or more the number of scanning lines 525 specified in the TSC system. In HDTV, the ratio of length to width is 9: 1.
6, which is a wide-angle screen compared to 3: 4 in the NTSC system. For this reason, it is possible to obtain a high-resolution screen with a sense of reality.

[0003] Even if an image signal of the NTSC system is supplied as it is to a television receiver capable of displaying a screen in the Hi-Vision system, the image cannot be displayed because the standard of the image signal is different. Therefore, conventionally, for example, the rate conversion of an image signal is performed using an image information conversion apparatus as shown in FIG. An NTSC image signal as an SD (Standard Definition) signal is input through an input terminal 100 and supplied to a horizontal interpolation filter 101. The horizontal interpolation filter 101 performs a horizontal interpolation process on the supplied signal. Further, the vertical interpolation filter 102 performs a vertical interpolation process on the output of the horizontal interpolation filter 101. In this way, a high-vision image signal is output as an HD (High Definition) signal.

A specific configuration of the horizontal interpolation filter 101 will be described with reference to FIG. The image signals of the NTSC system input via the input terminal 100 are divided into m + 1 multipliers 111 _m , 111 _m−1 , 111 _m−2 ,.
₀ is supplied respectively. Multipliers 111 _m−1 , 111
_m-2, ···, 111 ₀ performs a process of multiplying a predetermined coefficient to the signal supplied, adds the processing result 11
_{2 m-1, 112 m-} 2, ···, supplied to 112 _0. The outputs of the adders 112 _m−1 , 112 _m−2 ,..., 112 ₀ are the delay registers 113 _m−1 , 113 _m−2 ,.
..., it is supplied to the 113 _0. The delay register 11
_{3 m-1, 113 m-} 2, ···, 113 the output of _0, respectively, adders _{112 m-2, 112 m-} 3 ···, 112 0
Supplied to

[0005] Further, the output of the multiplier 111 _m is supplied to the delay register 113 _m, the output of the delay register 113 _m, it is supplied to the adder 112 _m-1. Here, delay register _{113 m-1, 113 m-} 2, ···, 113 0 is
This causes a delay time T in the supplied signal.

Therefore, the image signal of the NTSC system input through the input terminal 100 is delayed by the time T by the delay register 113 _m and supplied to the adder 112 _m-1 . The adder 112 _m-1 performs an addition process on the output of the delay register 113 _{m and} the output of the multiplier 111 _m-1 and outputs the result. The output of the adder 112 _m-1 is
A delay of time T is given by _m-1 and the adder 112 _m-2
Supplied to The adder 112 _m-2 is connected to the delay register 11
The output of 3 _{m−2 and} the output of multiplier 111 _m−2 are added and output.

Thereafter, the same processing is performed, and the adder 112 ₀ at the final stage adds the output of the delay register 113 _{0 and} the output of the multiplier 111 ₀ and outputs the result. Adder 11
The output of the 2 _0, the final output of the horizontal interpolation filter 101 (and thus, an image signal obtained by the horizontal direction interpolation processing) as supplied to the vertical interpolation filter 102 via the output terminal 120.

The vertical interpolation filter 102 has a configuration similar to that of the horizontal interpolation filter 101, and performs vertical pixel interpolation processing on the output of the horizontal interpolation filter 101. Then, a high-vision image signal obtained as an output of the vertical interpolation filter 102 is supplied to the high-vision receiver. By the above image information conversion, N
An image corresponding to the image signal of the TSC system can be displayed on a high-vision receiver.

[0009]

However, since the above-described image information conversion merely performs horizontal and vertical interpolation on an NTSC image signal, the resolution of the original NTSC system image signal is low. No difference from image signal. In particular, when a normal moving image is to be converted, it is common practice to perform vertical interpolation as intra-field processing. In such processing, inter-field correlation of images is not used. Therefore, due to the conversion loss,
The resolution may be lower than that of the original NTSC image signal.

On the other hand, the present applicant has proposed an apparatus for performing a classification adaptive process as an image information conversion process (see Japanese Patent Application No. 6-205934). The class classification adaptive processing includes class division according to a three-dimensional (spatio-temporal) distribution of signal levels of an input SD signal, and includes a memory for storing prediction coefficient values obtained by learning in advance for each class. By performing an operation based on a predetermined prediction equation using the result of
This is a process for generating an optimum estimation value as a D pixel.

In the class classification adaptive processing, when an HD pixel is generated, an S pixel in the vicinity of the HD pixel to be generated is generated.
A class division process is performed using the D pixel data. And
For each class that can be detected by the class division process, the prediction coefficient value is obtained in advance by learning, so that the intra-frame correlation is used in the image still unit, and the intra-field correlation is used in the image motion unit to obtain a more true value. It is made to obtain a near HD pixel value.

As a specific example of such processing, a case where HD pixels y1 and y2 are generated as shown in FIG. 9 will be described. SD pixels m _{1 to} m ₅ and SD pixels n _{1 to} n
_{For 5} , the average value of the inter-frame difference between pixels located at the same spatial position is calculated, and the average value is subjected to threshold value processing to perform class classification, thereby mainly performing the class classification relating to the degree of motion. At the same time, the SD pixels k _{1 to} k ₅ as shown in FIG.
g) By processing, the class classification mainly related to the representation of the waveform in space is performed with a small number of bits.

Then, HD pixels y1 and y2 are generated for each class determined by the above-described two types of class classification by an operation according to the following linear linear equation (1).

[0014] _{y = w 1 × x 1 +} w 2 × x 2 + ··· + w n × x n (1) SD pixels x ₁ to be used for such calculation, x ₂ ··· x _n
FIG. 11 shows an example of the arrangement. Here, 17 SD pixels are used, so n = 17. The prediction coefficient values w _{1 to} w _n used in the equation (1) are obtained in advance by learning. In such processing, class conversion mainly representing the degree of motion and class classification mainly representing waveforms in space are performed separately and in a form suitable for each. There is a feature that performance can be realized. In other words, the class is mainly classified into classes from 0 to 3 (hereinafter, referred to as a motion class) according to the degree of movement, mainly by the class classification related to the degree of movement. here,
It is assumed that the smaller the numerical value is, the smaller the degree of the movement is, and the larger the numerical value is, the larger the degree of the movement is (that is, the more likely it is to move).

When the motion class is small, there is a strong correlation between data belonging to the field to which the HD pixel to be estimated belongs and data belonging to another field. Therefore, it is effective to use data of other fields when estimating HD pixels. On the other hand, when the motion class is large, the correlation between the data belonging to the field to which the HD pixel to be estimated belongs and the data belonging to another field is weak. In this case, it is not effective to use data of other fields when estimating HD pixels. In the above-described classification adaptive processing, it is possible to skillfully realize the switching of the processing using the intra-frame correlation / intra-field correlation as described above according to the motion class.

Incidentally, the motion class classification in the above-described class classification adaptive processing is performed based on the difference between frames, and basically expresses the presence or absence of motion. Therefore, especially when there is movement, information other than the current field is not sufficiently reflected in the conversion process,
In such a case, the conversion performance has been reduced.

[0017] For example, if the motion as shown in FIG. 12 occurs, SD pixels x ₆ in performing the production of HD pixel y2
It is considered effective to perform a process in which the value of plays an important role. However, in the conversion process as described above, it is rare for processing using the x ₆ is made.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and it is therefore an object of the present invention to provide an image information conversion apparatus and an image information conversion method capable of improving conversion performance especially in a moving part. It is to provide a conversion method.

[0019]

According to a first aspect of the present invention, there is provided an image information converting apparatus for forming an output image signal having a different scanning line structure from an input image signal. A motion amount is calculated by using a first image cutout unit that cuts out the image data and image data cut out by the first image cutout unit, and a motion class representing a motion is determined based on the calculated motion amount, and is determined. Motion class detecting means for outputting information relating to a motion class, second image cutting means for cutting out image data at a predetermined position from an input image signal, and a pattern of a level distribution of image data cut out by the second image cutting means And determines the space class to which the image data belongs based on the detected pattern, and outputs information on the determined space class. A class code generator for combining the output of the space class detector, the output of the motion class detector, and the output of the space class detector to determine a class; Storage means for storing prediction coefficient data; third image cutting means for cutting out image data at a predetermined position from the input image signal; prediction coefficient data selected from the storage means in accordance with the output of the class code generation means; 3. An image information conversion device, comprising: an arithmetic processing unit for performing an arithmetic process for estimating an output image signal using the image data obtained by the image data selecting unit.

According to a ninth aspect of the present invention, there is provided an image information conversion method for forming an output image signal having a different scanning line structure from an input image signal, wherein the first image is obtained by cutting out image data at a predetermined position from the input image signal. A cutting step;
Calculating a motion amount using the image data cut out by the first image cutout step, determining a motion class representing the motion based on the calculated motion amount; A second image extraction step of extracting image data, and a space class for detecting a level distribution pattern of the image data extracted by the second image extraction step and determining a space class to which the image data belongs based on the detected pattern A detection step, a class code generation step of combining a result of the motion class detection step and a result of the space class detection step to determine a class, and prediction coefficient data determined in advance corresponding to the result of the class code generation step Storing an image data at a predetermined position from the input image signal. A third image extraction step of cutting out, according to the result of the class code generating step,
An arithmetic processing step of performing an arithmetic processing for estimating an output image signal using the prediction coefficient data selected from the storage step and the image data obtained by the third image data selecting step. Is an image information conversion method.

According to the above-described invention, a finer motion class can be calculated based on the motion amount detected by a method such as the block matching method.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below. Note that the SD signal /
As a combination with an HD signal as an output image signal, for example, a 525i signal (interlaced signal having 525 scanning lines) / 525p signal (progressive signal having 525 scanning lines), a 525i signal / 1050i signal (scanning) The present invention can be applied to a case where an interlace signal having 1050 lines is used.

FIG. 1 is a block diagram showing an overall configuration of an embodiment of the present invention. For example, a signal of the NTSC system is supplied as an SD image signal via the input terminal 1. The SD image signal is supplied to the region cutout units 4 and 7 and the delay circuit 12. The area cutout unit 4 cuts out pixels required for determining a motion class from the SD image signal, and supplies the cutout pixels to a motion class determination circuit 5. The motion class determination circuit 5 detects a motion amount by a method such as a block matching method.

FIG. 2 shows the positional relationship between SD pixels as input image signals and HD pixels as output image signals. Here, large dots indicate SD pixels, and small dots indicate HD pixels. When viewed in the same field, the HD pixel has two values, y1 existing at a position closer to the SD pixel and y2 existing at a position farther from the SD pixel.
There are types. In the following description, the mode for estimating the HD pixel y1 is described as mode 1, and the HD pixel y2 is estimated.
Is referred to as mode 2.

FIG. 3 shows an example of a pixel arrangement cut out by the area cutout unit 4 for detecting a motion class.
Shown in N for the HD pixel values y1 and y2 to be generated
A block of 5 × 5 (composed of a total of 25 pixels) (to be referred to as a block 1 in the following) located at two horizontal pixels relative to 1, n2, n3, n4, n5, and n1 to n5. Is cut out. Also, m1, m2, m3,
A block of 5 × 5 (consisting of a total of 25 pixels) (hereinafter, referred to as block 1), which is located at left and right two pixels in the horizontal direction with respect to m4, m5, and m1 to m5, is cut out. Further, block data having an offset of each of these 25 pixels within a predetermined search range is cut out.

Then, the motion class determination circuit 5 calculates a motion class mv_class according to the following equation (2).

Mv_class = (v _x + s _x ) × (s _y × 2 + 1) + (v _y + s _y ) (2) where v _x is a horizontal motion amount and v _y is a vertical motion amount. . Further, s _x is the absolute value of the horizontal search range, s _y is the absolute value of the vertical search range.

For example, when the search range of the motion amount is (± 8, ±
8), and when the detected motion amount is (+3, -2), the motion class mv_cls is calculated as follows.

Mv_cls = (3 + 8) × (8 × 2 +
1) + ((− 2) +8) = 193 The motion detection method may be performed by a method other than the block matching method, for example, a gradient method, a phase correlation method, or the like.

The motion class determined in this way is supplied to the area cutout unit 2 and the class code generation circuit 6 in FIG. The area cutout circuit 2 is further supplied with the output of the delay circuit 12, that is, the delayed SD image signal. The area cutout circuit 2 cuts out pixels required for determining the space class from the SD image signal,
The extracted pixels are supplied to the ADRC circuit 3. The delay circuit 12 determines the timing at which the motion class is supplied to the region cutout circuit 2 and the timing at which the motion class is input via the input terminal 1 due to the time required for the operation of the motion class determination circuit 5.
A deviation that occurs between the timing at which the D image signal is supplied to the area cutout circuit 2 is compensated.

FIG. 4 shows an example of the arrangement of pixels cut out by the area cut-out unit 2. Here, the HD to be generated
K1, k2, k3, k4, k for pixel values y1, y2
5 is cut out for the determination of the space class.

The ADRC circuit 3 converts the output of the area extracting circuit 2 from, for example, 8-bit SD data for the main purpose of classifying in the space, that is, patterning the supplied SD image signal with a small number of bits in the space waveform in the space. 2
An operation of compressing the data into bit SD data is performed. ADRC
Is an adaptive requantization method originally developed for high-efficiency coding for VTRs (Video Tape Recoders) .However, since local patterns at the signal level can be expressed efficiently with a short word length, In one embodiment of the invention, ADRC is used for code generation for spatial class classification. The ADRC calculates the maximum value MAX and the minimum value by the following equation (3), where DR is the dynamic range of the space class tap, n is the bit allocation, L is the data level of the pixel of the space class tap, and Q is the requantization code. Re-quantization is performed by equally dividing the data with the MIN by the designated bit length.

DR = MAX−MIN + 1 Q = {(L−MIN + 0.5) × 2 / DR} (3) where {} indicates a truncation process.

In one embodiment of the present invention, the ADRC
The circuit 3 compresses the SD data of 5 pixels each separated by the area cutout circuit 2 into 2 bits. Hereinafter, the compressed SD data will be referred to as q ₁ to q, respectively.
Notation ₅ These pattern compression data are supplied to the class code generation circuit 6.

On the other hand, the area cutout circuit 7 cuts out the pixels used for the estimation calculation from the supplied SD image signal.
FIG. 5 shows an example of a pixel arrangement to be cut out. In one example the cut out 17 pixels x ₁ ~x _17. The output of the region extracting circuit 7 is supplied to the estimation operation circuit 1 through the delay circuit 8.
0 is supplied. The delay circuit 8 delays the output of the area cutout circuit 7 with a delay necessary for adjusting the timing of processing in the estimation calculation circuit 10.

On the other hand, the class code generation circuit 6
An operation according to the following equation (4) is performed based on the space class supplied from the C circuit 3 and the motion class supplied from the motion class determination circuit 5. With this operation, the class to which each belongs is detected for each block. And
The class code class indicating the detected class is supplied to the memory 9.

[0037]

(Equation 1)

Here, the values of n and p are, for example, n = 5, p
= 2.

The memory 9 stores coefficient data for calculating HD pixels corresponding to SD pixels for each class. Such coefficient data is information for converting an SD image signal into an HD image signal having a higher resolution. In one embodiment of the present invention, coefficient data is prepared independently for mode 1 and mode 2. Memory 9
Outputs, using the class code class supplied from the class code generation circuit 6 as a read address, _wi (class) which is coefficient data for the class. The coefficient data output from the memory 9 is supplied to the estimation operation circuit 10. A method of creating the coefficient data stored in the memory 9 by learning will be described later.

The estimating operation circuit 10 performs an operation based on the output of the area extracting circuit 7 supplied via the delay circuit 8 and the coefficient data supplied from the memory 9 to obtain the HD pixel corresponding to the input SD pixel. Generate

More specifically, for example, 17 SD pixels x _{1 to} x cut out by the area cutout circuit 7
_17, the coefficient data w _i supplied from the memory 9 (cla
ss), an HD pixel corresponding to the SD pixel is calculated by performing an operation according to the following equation (5).
However, as described above, the coefficient data is set to the mode 1 and the mode 2
Are prepared independently of the above, the coefficient data for block 1 is used when mode 1 is performed, and the coefficient data for block 2 is used when mode 2 is performed.

Hd ′ = w ₁ × x ₁ + w ₂ × x ₂ + w ₃ × x ₃ + ... + w ₁₇ × x ₁₇ (5) The HD pixel data created in this way is supplied to the horizontal interpolation filter 11. Is done. As the horizontal interpolation filter 11, a conventionally used filter can be used. That is, the number of pixels in the horizontal direction is doubled by performing an interpolation process on the supplied signal. The output of the horizontal interpolation filter 11 is output via an output terminal 12. Such an output is supplied to, for example, an HD television receiver or an HD video tape recorder.

Next, creation of coefficient data stored in the memory 9 will be described with reference to FIG. In order to obtain such coefficient data by learning, an S corresponding to a known HD image signal and having 1/4 the number of pixels of the HD image signal is used.
First, a D image signal is formed. Specifically, the thinning-out process is performed by the vertical thinning filter 22 so that the frequency in the vertical direction in the field becomes に な る. Further, a thinning process is performed by the horizontal thinning filter 23 so that the horizontal frequency in the field becomes に な る. As an output of the horizontal thinning filter 23, an SD image signal having 1/4 the number of pixels of the HD image signal supplied via the input terminal 21 is obtained. The output of the horizontal thinning filter 23 is
Region cutout circuits 24 and 26, and delay circuit 34
Supplied to

The area cutout circuit 26 cuts out the data necessary for the motion class classification from the supplied SD image signal and sends the cutout data to the motion class determination circuit 27, similarly to the area cutout circuit 4 in FIG. Supply. The motion class determination circuit 27 shown in FIG.
In the same manner as described above, the motion class is determined, and the determined motion class is
And 8.

The output of the delay circuit 34, that is, the delayed SD image signal is further supplied to the area extracting circuit 24. The area cutout circuit 24 cuts out pixels required for determining the space class from the SD image signal,
The extracted pixels are supplied to the ADRC circuit 25. In addition,
The delay circuit 34 determines whether the motion class is determined by the time required for the operation of the motion class determination circuit 27 or the like.
4 and the horizontal thinning filter 23
Is provided to cope with a deviation from the timing at which the SD image signal output by the.

The ADRC circuit 25 detects a one-dimensional or two-dimensional level distribution pattern of the supplied SD image signal, and outputs the output of the area extracting circuit 25 from, for example, 8-bit SD data to 2-bit SD data. An operation of compressing to SD data is performed. Then, the pattern compression data generated by the calculation is
To supply. The class code generation circuit 28 is similar to the class code generation circuit 6 in FIG. That is,
The class code generation circuit 28 performs an operation based on the supplied pattern compression data (space class) and the motion class supplied from the motion class determination circuit 27 to detect the class to which each block belongs, and indicates the class. The class code is supplied to the normal equation adding circuit 31.

On the other hand, the area cutout circuit 29 is the same as the area cutout circuit 8 in FIG. 1, and performs an estimation calculation for obtaining a pixel to be generated from the SD image signal output from the horizontal thinning filter 23. Cut out the SD pixels to be used. The output of the area extracting circuit 29 is supplied to a normal equation adding circuit 31 via a delay circuit 30. Note that the delay circuit 30 delays the output of the area cutout circuit 29 to adjust the processing timing in the normal equation addition circuit 31.

Here, for explanation of the normal equation adding circuit 31, learning of a prediction formula used when converting a plurality of SD pixels into HD pixels and signal conversion using the prediction formula will be described. explain. The following description is for a general case where prediction is performed using n pixels. n
The number of SD pixel levels _{x 1, x 2, x 3} , ···, x n
, And re-quantized data obtained by performing p-bit ADRC on them, respectively, are q ₁ , q ₂ , q ₃ ,.
Notated as _qn . Further, the n SD pixel levels are x ₁
The class class of the area including ~ _xn is defined by the above equation (4).

Then, an n-tap linear estimation expression (the following expression (6)) is set using coefficients w ₁ ,..., W _n determined by learning for each class. Here, the coefficients w ₁ ,..., W _n are determined by learning, and are undetermined coefficients before learning. y = w ₁ × x ₁ + w ₂ × x ₂ + ‥‥ + w _n × x _n (6) Learning is performed on a plurality of image signal data (called training data) for each class. When the total number of training data is expressed as m, the following equation (7) is set according to the equation (1).

Y = w ₁ × x _k1 + w ₂ × x _k2 + ‥‥ + w _n × x _kn (7) (k = 1,2, ‥‥, m) When m> n, the prediction coefficients w ₁ , ‥ Since ‥ and w _n are not uniquely determined, the element e _k of the error vector e is defined by the following equation (8), and the prediction coefficient is set so as to minimize the error vector e defined by the equation (9). To be determined. That is, the prediction coefficient is uniquely determined by the so-called least square method.

E _k = y _k − {w ₁ × x _k1 + w ₂ × x _k2 + Δ + w _n × x _kn } (8) (k = 1, 2, Δm)

[0052]

(Equation 2)

As a practical calculation method for obtaining the prediction coefficient minimizing e ² in Equation (9), partial differentiation of e ² with the prediction coefficient w _i (i = 1, 2 ‥‥) is performed (Equation 9). (10)) Each prediction coefficient w _i may be determined so that the partial differential value becomes 0 for each value of _i .

[0054]

(Equation 3)

A specific procedure for determining each prediction coefficient w _i from equation (10) will be described. When X _ji and Y _i are defined as in Expressions (11) and (12), Expression (10) becomes Expression (1).
3) can be written in the form of the determinant.

[0056]

(Equation 4)

[0057]

(Equation 5)

[0058]

(Equation 6)

Equation (13) is generally called a normal equation. The normal equation adding circuit 27 includes a class code supplied from the class code generating circuit 28 and SD pixel data x ₁ , x ₂ , supplied from the area extracting circuit 30.
Based on x ₃ ,..., x _n and the HD pixel level y input via the input terminal 21 as training data, normal equations are added, and equations (11), (1)
The values of X _ji and Y _i according to 2) are calculated.

After the input of all the training data is completed, the normal equation adding circuit 31 supplies the normal equation data to the prediction coefficient determining circuit 32. The prediction coefficient determination circuit 32 calculates a prediction coefficient w _i by performing a calculation process for solving a normal equation according to a general matrix solution method such as a sweeping-out method based on the supplied normal equation data. Then, the calculated prediction coefficient w _i is supplied to the memory 33. Memory 33 stores the prediction coefficient w _i to be supplied to each class.

By the learning as described above, the memory 33 stores a prediction coefficient that enables statistically closest estimation to the true value when estimating the HD pixel data y for each class. The coefficient data stored in the memory 33 is loaded into the memory 9 in FIG. 1 and used in the image information conversion processing as described above.

As described above, in the present invention, the amount of motion is used in the motion class classification (see equation (4)), so that pixels belonging to fields other than the current field are replaced with pixels in the current field. It can be used effectively for pixel estimation.

[0063]

As described above, according to the present invention, when performing image information conversion processing, the motion amount of image data cut out from an input image signal is calculated by, for example, a block matching method, and the motion amount is calculated based on the calculated motion amount. Is determined, and the final class is determined by combining the motion class and a separately determined space class. Further, it has a memory for storing prediction coefficient data determined in advance corresponding to the final class thus obtained.

By determining the motion class based on the amount of motion in this way, a more detailed motion class classification can be performed as compared with the motion class classification based on the detection of the frame difference conventionally proposed. .

For this reason, in the arithmetic processing for generating pixels in the image information conversion processing, the pixel data included in the fields other than the current field can be effectively used, thereby improving the conversion performance of the image information conversion processing. Can be done.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of the overall configuration of an embodiment of the present invention.

FIG. 2 is a schematic diagram illustrating an example of image information conversion performed according to an embodiment of the present invention.

FIG. 3 is a schematic diagram illustrating an example of a pixel arrangement for detecting a motion class according to an embodiment of the present invention.

FIG. 4 is a schematic diagram illustrating an example of a pixel arrangement for detecting a space class in one embodiment of the present invention.

FIG. 5 is a schematic diagram illustrating an example of an arrangement of pixels used for estimating pixels in an output image signal according to an embodiment of the present invention.

FIG. 6 is a block diagram illustrating an example of a configuration of a prediction coefficient calculation processing system according to an embodiment of the present invention.

FIG. 7 is a block diagram for explaining a conventional image information conversion process.

FIG. 8 is a block diagram for explaining the conventional image information conversion processing in more detail.

FIG. 9 is a schematic diagram illustrating an example of a pixel arrangement for detecting a motion class in a conventional image information conversion process.

FIG. 10 is a schematic diagram illustrating an example of a pixel arrangement for detecting a space class in conventional image information conversion processing.

FIG. 11 is a schematic diagram illustrating an example of an arrangement of pixels used for estimating pixels in an output image signal in a conventional image information conversion process.

FIG. 12 is a schematic diagram for explaining a problem in a conventional image information conversion process.

[Explanation of symbols]

 5 ... Motion class determination circuit

Claims (9)

[Claims] 1. An image information conversion apparatus for forming an output image signal having a different scanning line structure from an input image signal, wherein: first image cutout means for cutting out image data at a predetermined position from the input image signal; A motion calculating a motion amount using the image data clipped by the first image clipping unit, determining a motion class representing the motion based on the calculated motion amount, and outputting information relating to the determined motion class; Class detection means, second image cutout means for cutting out image data at a predetermined position from the input image signal, and a level distribution pattern of image data cut out by the second image cutout means is detected and detected. The spatial class to which the image data belongs is determined based on the pattern, and the spatial class that outputs information related to the determined spatial class. Class detection means; class code generation means for combining the output of the motion class detection means with the output of the space class detection means to determine a class; Storage means for storing the predicted coefficient data, third image cutout means for cutting out image data at a predetermined position from the input image signal, and a prediction coefficient selected from the storage means in accordance with an output of the class code generation means An image information conversion device comprising: an arithmetic processing unit that performs an arithmetic process for estimating an output image signal using data and image data obtained by the third image data selecting unit. 2. The image processing apparatus according to claim 1, wherein the prediction coefficient is selected by a true pixel value in a predetermined image signal having the same signal format as the output image signal and the third image data selection unit. An image information conversion apparatus characterized in that a difference between the calculated linear linear combination and image data is determined to be the minimum. 3. The motion class detecting unit according to claim 1, wherein the motion class detecting unit detects a motion amount based on an output of the first image clipping unit, and determines the motion class using the detected motion amount. An image information conversion device characterized by performing an operation for performing the above operation. 4. The image information conversion device according to claim 3, wherein the motion amount is detected by a block matching method. 5. The image information conversion device according to claim 3, wherein the motion amount is detected by a gradient method. 6. The image information conversion device according to claim 3, wherein the motion amount is detected by a phase correlation method. 7. The apparatus according to claim 1, wherein the input image signal is an interlaced image signal having 525 scanning lines, and the output image signal is a progressive image signal having 525 scanning lines. Image information conversion device. 8. The method according to claim 1, wherein the input image signal is an interlace image signal having 525 scanning lines, and the output image signal is a progressive image signal having 1050 scanning lines. Image information conversion device. 9. An image information conversion method in which an output image signal having a different scanning line structure is formed from an input image signal, a first image cutout step of cutting out image data at a predetermined position from the input image signal, A motion class detecting step of calculating a motion amount using the image data cut out in the first image cutting step, and determining a motion class representing a motion based on the calculated motion amount; A second image cutout step of cutting out the image data of the position, a pattern of a level distribution of the image data cut out by the second image cutout step is detected, and a space class to which the image data belongs is determined based on the detected pattern. The spatial class detection step to be determined; the result of the motion class detection step; A class code generation step of combining the result of the inter-class detection step to determine a class; a storage step of storing prediction coefficient data determined in advance corresponding to the result of the class code generation step; A third image cutout step of cutting out image data at a predetermined position from the image data; prediction coefficient data selected from the storage step according to a result of the class code generation step;
An image processing method for estimating an output image signal using the image data obtained in the image data selecting step.