JP4120945B2 - Image signal conversion apparatus and image signal conversion method - Google Patents

Description

  The present invention relates to a signal processing apparatus, a signal processing method, an image signal conversion apparatus, and an image signal conversion method, and is suitable for application to, for example, a television apparatus.

Conventionally, a television receiver receives a television broadcast signal transmitted from a television station, and displays an image based on the television broadcast signal on a screen.
In other words, a television receiver receives and selects a television broadcast signal transmitted from a television station with an antenna and a tuner, performs predetermined signal processing on the television broadcast signal, and then outputs an image and sound based thereon. To do.
In this way, the user can view a program or the like by the television broadcast signal transmitted from the television station.

By the way, such a television receiver has only a passive function to display a given image as it is, and when viewing an image, a user's request that “I want to repeat the current scene” is requested. I couldn't be satisfied.
As a method for realizing such a requirement, there is a method of reproducing an image recorded on a recording medium such as a VTR (Video Tape Recorder) or a video disk and displaying it on a monitor. When viewing and listening, there is a problem that the rewinding operation of the recording medium needs to be repeated, and the operation becomes complicated.

  In order to avoid this, a television receiver capable of viewing a scene to be reviewed by providing a storage unit such as a memory inside to store a television broadcast signal and reproducing the stored television broadcast signal. Has been proposed.

  By the way, in a television receiver having a storage unit inside as in such a configuration, when a television broadcast signal is stored as it is, the amount of information that can be stored is small, and only a still image can be displayed, which is requested by the user. Cannot be satisfied sufficiently.

In order to solve such a problem, a television signal receiver as disclosed in Japanese Patent Laid-Open No. 6-78229 has been proposed. By compressing and storing an input television broadcast signal, the amount of information that can be stored is increased, and a moving image can be replayed.
However, this type of television signal receiving apparatus reduces the image quality of the displayed image as the compression efficiency is increased in order to store the image signal for a long time, and the image quality of the display image is reduced. When efficiency was lowered, there was a problem that long-term memory was not possible.

  The present invention has been made in consideration of the above points, and a signal processing apparatus and signal capable of repeatedly displaying a replay image of a desired scene of an image based on an input image signal with a good image quality and for a long time. A processing method, and an image signal conversion apparatus and an image signal conversion method are proposed.

In order to solve such a problem, in the present invention, the first image signal which has been subjected to compression processing by thinning out pixels in the video signal and has deteriorated resolution is used as a unit block for the target pixel and a plurality of pixels around the target pixel. to correct dividing means for dividing the first divided image signals, an error occurring between the decoding teacher data decoded by extension our method corresponding to the compression processing after compression by the compression processing teacher data and the teacher data , After classifying into a plurality of classes classified according to the level distribution pattern of the image signal represented by the decoding teacher data , storage means for storing the prediction coefficient calculated for each class, and the first divided image signal first classification means for classification into one class of the first divided image signals classified according to the level distribution pattern more classes , By performing prediction computation process for generating a pixel value of pixels decimated by the first divided image signal to product-sum operation of the pixel value represented by the prediction coefficients and the first image signal corresponding to the class belonging A prediction calculation means for forming a second image signal having a resolution higher than that of the first image signal, and a region corresponding to the motion region in the second image signal from the region number indicating the motion region having motion in the video signal Is selected as an enlargement area to be enlarged, and the target pixel in the enlargement area and a plurality of pixels around the target pixel are divided into second divided image signals having unit blocks, and the level distribution pattern of the second divided image signal by classified according to, the second divided image signal classification into one class among the plurality of classes, prediction corresponding to the class of the second divided image signal belongs By performing prediction processing of generating a pixel value of a pixel to be interpolated by the number and the second divided image signal to product-sum operation of the pixel values representing, an enlarged image forming means for increasing the spatial resolution of the enlarged image signals I made it.

In the present invention, the first divided image signal in which the first image signal, which has been subjected to compression processing by decimation of pixels in the video signal and whose resolution has deteriorated, is the unit block of the target pixel and a plurality of pixels around the target pixel. A first class classification step for classifying the first divided image signal classified according to the level distribution pattern of the first divided image signal into one of a plurality of classes ; the level distribution pattern of the image signal representing the decoding teacher data to correct the error that occurs between the decoding teacher data decoded by extension our method corresponding to the compression processing after compression by the compression processing teacher data and the teacher data click the first divided image signals of the prediction coefficient calculated for each said class upon classified into a plurality of classes which are classified according belongs By performing prediction processing of generating a pixel value of pixels decimated by the product-sum operation of the pixel value represented by the prediction coefficients and the first image signal corresponding to the scan, resolution than the first image signal The prediction calculation step for forming the improved second image signal and the region number indicating the motion region having motion in the video signal are selected as the enlargement region to be enlarged from the region corresponding to the motion region in the second image signal. Then, by dividing the pixel of interest in the enlarged region and a plurality of pixels around the pixel of interest into a second divided image signal having unit blocks, and classifying according to the level distribution pattern of the second divided image signal, a second divided image signal classification into one class among the plurality of classes, prediction coefficients and a second divided image signals corresponding to the class in which the second divided image signal belongs By performing prediction processing of generating a pixel value of a pixel to be interpolated by the to the pixel value to the product-sum operation, and to provide an enlarged image forming step to increase the spatial resolution of the enlarged image signals.

As a result, the image quality of the enlarged image can be improved using the class classification adaptive processing.

According to the present invention, an image signal conversion apparatus capable of improving and restoring the resolution of an image signal that is reduced by enlarging an image by using class classification adaptive processing, and thus generating an image signal with a high resolution. In addition, an image signal conversion method can be realized.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

(1) Overall Configuration In FIG. 1, reference numeral 1 denotes a television signal receiving apparatus as a whole. The received television broadcast signal is compressed and encoded and temporarily stored while being updated every predetermined time, and the replay mode is selected. At this time, the replay image or the enlarged replay image is repeatedly displayed on the monitor until the replay mode is canceled by reading the stored television broadcast signal.

The television signal receiving apparatus 1 supplies a television signal S 1 received by the antenna 2 to the tuner 3. The tuner 3 demodulates and Y / C separates the television signal S1 to form an image signal S2, and sends it to paths RT1 and RT2, respectively. Here, the route RT1 is a route for sending a normal television broadcast signal, and the route RT2 is a route for forming and sending a replay image or an enlarged replay image.
The image signal S2 sent to the route RT1 is given to the switch 4. The analog / digital converter (A / D) 5 that receives the image signal S2 sent to the path RT2 quantizes the image signal S2 and converts it into an image signal S3 composed of a digital signal, which is then thinned out. 6 and motion detection circuit 7.

The thinning circuit 6 compresses the information amount of the image signal S3 to ¼ by subjecting the image signal S3 to pixel thinning processing in a spatio-temporal direction such as by the MUSE (Multiple Sub-Nyquist Sampling Encode) method. The thinned image signal S4 is formed and sent to the gradation compression circuit 8.
The gradation compression circuit 8 performs compression processing in the gradation direction, for example, by ADRC (Adaptive Dynamic Range Coding) coding, on the thinned image signal S4, so that the pixels of the thinned image signal S4 in the gradation direction are processed. The amount of information is compressed to 1/4.
Thus, by compressing the image signal S3 in the spatio-temporal direction and the gradation direction, compressed image data S5 is formed by compressing the image signal S3 to 1/16 of the original information amount, and the compressed image data S5 is stored. Supply to device 9.

In addition, the motion detection circuit 7 detects a motion region in the image based on the image signal S <b> 3 with a configuration described later, forms motion region data S <b> 6 representing the motion region, and supplies the motion region data S <b> 6 to the storage device 9.
The storage device 9 is composed of a hard disk, an MO (Magneto-optical) disk, an MD (Mini Disc), a large-capacity semiconductor memory, or the like, and updates the compressed image data S5 and the motion area data S6 every predetermined time. Remember.

The control unit 10 sends a switching control signal S8 corresponding to the selection content indicated by the mode selection data S7 given by the user's operation to the storage device 9, and performs switching control of writing / reading of the storage device 9.
For example, when the user selects a mode for displaying a replay image or an enlarged replay image, the storage device 9 stops updating and storing the compressed image data S5 and the motion area data S6 by the switching control signal S8, and the compressed image data S5. Are repeatedly read and transmitted to the decoding circuit 11 and the motion area data S6 to the delay circuit 12, respectively.

The decoding circuit 11 expands and decodes the compressed image data S5 read from the storage device 9 to form a restored image signal S9 with a configuration described later, and the restored image signal S9 is sent to the zoom processing circuit 13 and the switch 14. Send repeatedly.
The delay circuit 12 sends to the zoom processing circuit 13 motion region data S6 that has been delayed for a period during which the decoding circuit 11 decompresses and decodes the compressed image data S5 to form the restored image signal S9.
The zoom processing circuit 13 forms an enlarged image signal S10 based on the motion area data S6 from the restored image signal S9 with a configuration described later. The zoom processing circuit 13 repeatedly sends the enlarged image signal S10, which is an enlarged replay image obtained by enlarging the motion area in the replay image, to the switch 14.

The switch 14 is controlled to be switched by a control signal S8 sent from the control unit 10, and selectively switches between a restored image signal S9 given by the decoding circuit 11 and an enlarged image signal S10 given by the zoom processing circuit 13. To the digital / analog converter (D / A) 15.
That is, when the user selects a mode for displaying a replay image, the switch 14 remains normal and is connected to the restored image signal S9. On the other hand, when the user selects a mode for displaying an enlarged replay image, the connection of the switch 14 is switched to the enlarged image signal S10 side.
As described above, the switch 14 is controlled to be switched by the control signal S8 transmitted from the control unit 10 in accordance with the mode selected by the user, and selectively switches the transmitted restored image signal S9 or enlarged image signal S10 to be transmitted. .

The digital / analog converter 15 converts the restored image signal S9 or the enlarged image signal S10 into an analog signal and sends it to the switch 4.
Similarly to the switch 14, the switch 4 is controlled to be switched by a control signal S8 sent out by the control unit 10, and selectively switches between the input image signal S2 and the restored image signal S9 or the enlarged image signal S10. The signal is sent to the signal signal processing circuit 16.

That is, normally, when the user is viewing an image based on a television broadcast signal, the switch 4 is connected to the image signal S2 input via the route RT1. When the user selects a mode for displaying a replay image or an enlarged replay image, the switch 4 is connected to the restored image signal S9 or the enlarged image signal S10 input via the route RT2.
As described above, the switch 4 is controlled to be switched by the control signal S8 sent from the control unit 10 according to the mode selected by the user, and selects the input image signal S2 and the restored image signal S9 or the enlarged image signal S10. Switch to send.

The television signal processing circuit 16 processes the image signal S 2, the restored image signal S 9 or the enlarged image signal S 10 input via the switch 4 so that the image can be displayed on the monitor 17, and sends this to the monitor 17.
The monitor 17 displays an image based on the image signal S2 supplied via the signal processing circuit 16, a replay image based on the restored image signal S9, or an enlarged replay image based on the enlarged image signal S10.

  As described above, the television signal receiving apparatus 1 switches the replay image or the enlarged replay image obtained by enlarging the motion region in the replay image to the normal television image and repeatedly displays the replay image or the replay image according to the user's mode switching operation. Has been made.

(2) Configuration of Motion Detection Circuit In FIG. 2, in which parts corresponding to those in FIG. 1 are assigned the same reference numerals, 7 denotes a motion detection circuit as a whole, and a motion vector is obtained by a technique based on representative point blocking. Has been made.
For example, as shown in FIG. 3, an image is divided into 12 macroblocks, and motion vectors are detected in each block.

That is, information such as the luminance of the representative point p is stored for each field of the image while being updated for each representative point p arranged at predetermined intervals vertically and horizontally in each macro block.
Further, in the field after one field, a search block having a size including the moving distance where the representative point is expected to move is cut out. Here, the block is cut out in a size of 31 pixels in the vertical direction and 31 pixels in the horizontal direction.

In each search block thus cut out, the absolute difference between the information such as the luminance of each pixel in the field at a certain point in time and the information such as the luminance of the representative point p in the same search block in the field immediately before that point in time Find the value. Here, information such as the luminance of the representative point p of each block in the previous field is read out and used. By obtaining the absolute value of the difference in this way, 31 × 31 tables are created.
Thereafter, the absolute values of the obtained differences are added (that is, integrated) in each block. Since the integration result has a mortar shape, the position of the minimum value is obtained from this, and a vector from the representative point p to the position of the minimum value is subtracted. In this way, an averaged motion vector for each macro block is obtained.

Next, the motion vectors obtained in this way are sequentially compared with adjacent blocks (that is, vector distance calculation) and matched within a certain error e (that is, a conditional expression of | vector V _n −vector V _{n + 1} ≦ e |). Satisfy) vector. Further, those that satisfy the condition in the vicinity of the center of the image are merged, and this macro block number is used as the motion region number.

In practice, the motion detection circuit 7 first inputs the image signal S3 to the memory 20 and the cut-out circuit 21, respectively, as shown in FIG. The memory 20 stores the information of the representative point p extracted from the image signal S3 for each field of the image and then sends it to the difference circuit 22 as the information signal S11.
Further, the cut-out circuit 21 cuts the search block by dividing the image signal S3 after one field of the field from which the information signal S11 is extracted, and sends the information in each cut-out search block to the difference circuit 22 as the information signal S12.

The difference circuit 22 calculates the absolute value of the difference between the information signal S11 and the information signal S12, and supplies the calculation result to the calculation unit 23 as table data S13.
The calculation unit 23 is composed of integration circuits 23A to 23L that are allocated to and associated with one of the macroblocks, respectively, and extract only the table data in the corresponding macroblock from the table data S13 and perform an integration calculation. The calculation results are sent to the detection circuit 24 as vector data S14A to S14L.

  The detection circuit 24 obtains a motion vector of each macro block based on the vector data S14A to S14L, selects a motion vector satisfying the above-mentioned conditions from the motion vector, sets the macro block having the motion vector as a motion region, The area number of the area is sent as motion area data S6 to the storage device 11 (FIG. 1). As described above, the motion detection circuit 7 detects the motion region of the image signal S3 for each field, forms motion region data S6 indicating the region number of the motion region, and sends the motion region data S6 to the storage device 9. It is made to do.

(3) Configuration of Decoding Circuit In FIG. 4, in which the same reference numerals are assigned to the corresponding parts as in FIG. 1, 11 indicates the decoding circuit as a whole, and compression processing is performed in the space-time direction and the gradation direction. By decoding the compressed image data S5 using the class classification adaptive processing, a high-resolution restored image signal S9 is generated.

The decoding circuit 11 inputs the compressed image data S5 read from the storage device 9 (FIG. 1) to the blocking circuit 25 and the decoding circuit (decode) 26.
The blocking circuit 25 blocks the compressed image data S5 using the pixel of interest and a plurality of pixel codes around the pixel of interest as one unit, thereby forming a blocking signal S15 and sending it to the class generation circuit 27.
The class generation circuit 27 classifies the blocked signal S15 and generates an index indicating the classified class. For example, only the MSB (Most Significant Bit) is extracted from the pixel code in the block signal S15, and the block signal S15 is subjected to pattern classification by arranging 1/0 of the MSB in order, and the block signal S15 is classified according to the thus classified pattern. The class to which the blocking signal S15 belongs is determined, and an index signal S16 composed of an index representing the class is generated and transmitted.

On the other hand, the decoding circuit 26 sends the image data S17 expanded in the gradation direction by performing the ADRC decoding process on the compressed image data S5 to the blocking circuit 28.
The blocking circuit 28 extracts a plurality of peripheral pixel values centered on the target pixel from the image data S17, blocks them, and supplies them as a blocking signal S18 to the arithmetic unit 29.

The arithmetic unit 29 includes product-sum arithmetic circuits 29A to 29D and coefficient memories M1 to M4. The product-sum operation circuit 29A and the coefficient memory M1 perform a prediction operation for restoring the resolution that has been lowered by the compression in the gradation direction, and the product-sum operation circuits 29B to 29D and the coefficient memories M2 to M4 perform the thinning process. Prediction calculation is performed to interpolate the lost pixels.
For this reason, different prediction coefficients are stored in the coefficient memory M1 and the coefficient memories M2 to M4, respectively.

  That is, the index signal S16 supplied from the class generation circuit 27 is input to the coefficient memories M1 to M4. The coefficient memories M1 to M4 are ROM (Read Only Memory), and the ROM stores prediction coefficients obtained in advance by a method described later. The coefficient memories M1 to M4 use the index signal S16 as an address signal to read out the stored prediction coefficients and send them to the product-sum operation circuits 29A to 29D, respectively.

The product-sum operation circuit 29A uses the prediction coefficient supplied from the coefficient memory M1 to perform a prediction operation for restoring the resolution in the gradation direction of the pixel indicated in the block signal S18. Further, the product-sum operation circuits 29B to 29D perform pixels using the pixels indicated in the block signal S18 and the prediction coefficients given from the coefficient memories M2 to M4, thereby restoring the pixels in the thinned portion. Generate a value.
The calculation result thus obtained is sent to the multiplexing circuit 30.

The multiplexing circuit 30 generates the restored image signal S9 by multiplexing the pixel values based on the calculation result in the calculation unit 29 so as to be arranged in the television time series direction, and the switch 16 and the zoom processing circuit 13 (FIG. 1). ).
Thus, when decoding the compressed image data S5 using the class classification adaptive process, the decoding circuit 11 performs arithmetic processing using different prediction coefficients for the thinned pixels and the pixels that have not been thinned. Therefore, by interpolating the pixels thinned out in the spatio-temporal direction to increase the resolution and to improve the resolution in the level direction, a replay image having an overall improved resolution is formed.

(4) Configuration of Zoom Processing Circuit In FIG. 5, in which parts corresponding to those in FIG. 1 are assigned the same reference numerals, 13 denotes a zoom processing circuit as a whole, and an enlarged image signal S10 in which the motion region of the restored image signal S9 is enlarged. Has been made to form.

The zoom processing circuit 13 sets an area including the entire moving area indicated by the moving area data S6 and having an aspect ratio of 4: 3 within a range slightly larger than the size of the moving area. Using the area set in this way as an enlarged area, the enlargement ratio is calculated from the ratio of the size of the enlarged area and the monitor screen.
The zoom processing circuit 13 reads out pixels in the enlargement area set from the restored image signal S9 and performs enlargement processing with the calculated enlargement ratio. Here, by using the class classification adaptive process for the enlargement process, the resolution is improved by interpolating pixels that are insufficient during the enlargement process.
Thus, a full-size enlarged replay image is formed.

In practice, as shown in FIG. 5, the zoom processing circuit 13 first inputs the motion region data S6 to the region setting circuit 31, and also inputs the restored image signal S9 to the enlarged image generation unit 32.
The region setting circuit 31 sets a region to be enlarged based on the macro block number of the motion region indicated in the motion region data S6. The set area information is given to the enlargement ratio calculation circuit 33.
The enlargement ratio calculation circuit 33 calculates the enlargement ratio of the image from the ratio between the area set by the area setting circuit 31 and the size of the monitor screen, and gives the calculation result to the enlarged image generation unit 32.

  Based on the enlargement ratio given from the enlargement ratio calculation circuit 33, the enlargement image generation unit 32 performs an enlargement process using the class classification adaptive process on the pixels in the enlargement area read out from the restored image signal S9, so that FIG. An enlarged image signal S10 is generated by enlarging the motion region as indicated by a broken line inside.

Here, the configuration of the enlarged image generating unit 32 is shown in FIG. In FIG. 7, in which parts corresponding to those in FIG. 5 are denoted by the same reference numerals, 13 indicates a zoom processing circuit as a whole, and a restored image signal S9 and an enlargement ratio calculation circuit are added to the enlarged image generator 32 indicated by a broken line in the drawing. The enlargement ratio determined at 33 is supplied.
The enlarged image generating unit 32 inputs the restored image signal S9 to the blocking circuit 34 and the delay circuits 35 and 36, respectively.

The blocking circuit 34 blocks the restored image signal S9 with the target pixel and a plurality of pixel codes around the target pixel as one unit, and sends each block to the class generation circuit 37 as a blocking signal S19.
The class generation circuit 37 performs pattern classification of the block signal S19 by performing ADRC encoding processing on the block signal S19, and determines a class to which the block signal S19 belongs according to the thus classified pattern. Then, an index signal S20 composed of an index representing the class is generated and sent to the calculation unit 38.

The restored image signal S9 delayed by a predetermined time via the delay circuit 35 is sent to the block circuit 39.
The blocking circuit 39 extracts a plurality of peripheral pixel values centered on the target pixel from the image signal S9, blocks them, and sends them to the arithmetic unit 38 as a blocking signal S21.

The calculation unit 38 includes a plurality of product-sum calculation circuits and coefficient memories, and each coefficient memory has prediction coefficients for predicting and generating pixels that are insufficient during the enlargement process based on surrounding pixel values. These are stored in advance for each of various enlargement ratios.
Each coefficient memory reads a prediction coefficient using the expansion ratio given from the index signal S20 and the enlargement ratio calculation circuit 33 as an address value, and gives the prediction coefficient to the corresponding product-sum operation circuit.
Each product-sum calculation circuit generates a pixel value to be interpolated by calculating the given prediction coefficient and the blocked signal S21 and supplies the pixel value to the multiplexing circuit 40.
The coefficient memory described above is a ROM (Read Only Memory) as in the case of the decoding circuit 13, and stores a prediction coefficient obtained in advance by a method described later.

The multiplexing circuit 40 generates the interpolation pixel data S22 by multiplexing each pixel value calculated by the calculation unit 38 in the television signal direction, and sends it to the synthesis circuit 41.
The synthesizing circuit 41 extracts pixels in the area to be enlarged from the restored image signal S9 supplied via the delay circuit 36, and adds the interpolated pixel data S22 supplied from the multiplexing circuit 40 to the missing pixels in the enlargement process. The enlarged image signal S10 is generated by interpolating the data and is sent to the switch 14 (FIG. 1).
As described above, the zoom processing circuit 13 applies the restored image signal S9 and the motion region data S6 indicating the motion region in the image to the motion region by enlarging the restored image signal S9 using the class classification adaptive processing. An enlarged image signal S10 is generated by enlarging and interpolating pixels that are insufficient during enlargement to increase the resolution.

(5) Configuration of Prediction Coefficient Generating Circuit In FIG. 8, in which the same reference numerals are assigned to corresponding parts to FIGS. 1 and 4, 42 indicates the prediction coefficient generating circuit as a whole, and the pixel by the thinning circuit 6 (FIG. 1) A prediction coefficient for restoring the resolution in the gradation direction of the pixels that have not been thinned out during the thinning process is created. Incidentally, the compression encoding process performed in the thinning circuit 6 (FIG. 1) performs 1/4 pixel thinning, and 3 pixels are thinned out from a block having 4 pixels as one unit to leave one pixel. Yes.
The prediction coefficient generation circuit 42 inputs a sample image as teacher data, compresses and encodes the sample image, and then decodes and restores the image. The image quality that has been reduced due to the correlation between the restored image and the original sample image is obtained. An optimum prediction coefficient is obtained for restoration by arithmetic processing.

The prediction coefficient creation circuit 42 inputs the image signal S3 to the gradation compression circuit 8 and the blocking circuit 43, respectively.
The gradation compression circuit 8 performs compression processing in the gradation direction, for example, by ADRC encoding processing on the image signal S3, and blocks the obtained compressed image data S23 into a block circuit 25 and a decoding circuit (decode) 26. To send.
On the other hand, the blocking circuit 43 blocks the image signal S3 with the target pixel and a plurality of pixels centered on the target pixel as one unit, and sends this to the coefficient calculation circuit 44 as a blocking signal S24.

The blocking circuit 25 blocks the compressed image data S23 using the pixel of interest and a plurality of pixel codes around the pixel of interest as a unit, thereby forming a blocking signal S15 and sending it to the class generation circuit 27.
The class generation circuit 27 classifies the blocked signal S15 and generates an index indicating the classified class. For example, only the MSB (Most Significant Bit) is extracted from the pixel code in the block signal S15, and the block signal S15 is subjected to pattern classification by arranging 1/0 of the MSB in order, and the block signal S15 is classified according to the thus classified pattern. The class to which the blocked signal S15 belongs is determined, and an index signal S16 composed of an index representing the class is generated and sent to the coefficient calculation circuit 44.

On the other hand, the decoding circuit 26 sends image data S17 expanded in the gradation direction by performing ADRC decoding processing to the compressed image data S23 to the blocking circuit 28.
The blocking circuit 28 extracts and blocks a plurality of peripheral pixel values centered on the target pixel from the image data S17, and supplies them to the coefficient calculation circuit 44 as a blocking signal S18.
The coefficient calculation circuit 44 calculates a prediction coefficient representing the correlation between the blocked signal S24 and the blocked signal S18 for each index code indicated by the index signal S16.

In the ADRC encoding process, the data is rounded by rounding off the fractional part in the calculation process when compressing and decoding the input image signal in the gradation direction. An error occurs between the values.
As described above, the prediction coefficient creation circuit 42 corrects the error by a prediction calculation and reduces it by performing a learning operation to obtain a correlation between the sample image and the image decoded after the sample image is compressed. An optimal prediction coefficient that can restore the image quality can be calculated.
The prediction coefficient obtained in this way is output as coefficient data S25 and written in the memory M1 (FIG. 4).

  Further, in FIG. 9 in which the same reference numerals are assigned to the corresponding parts to those in FIGS. 1, 4 and 8, 45 denotes a prediction coefficient generation circuit as a whole, and in the pixel thinning process by the thinning circuit 6 (FIG. 1). A prediction coefficient for generating pixel values of the thinned pixels is generated. The prediction coefficient creation circuit 45 is configured in substantially the same manner as the prediction coefficient creation circuit 42 except that the thinning circuit 6 that performs thinning processing is provided in the previous stage of the gradation compression circuit 8.

The prediction coefficient creating circuit 45 inputs the image signal S3 serving as teacher data to the thinning circuit 6 and the blocking circuit 43.
The thinning circuit 6 performs a pixel thinning process in the spatio-temporal direction on the image signal S3, and sends the thinned image signal S4 with the information amount compressed to ¼ to the gradation compression circuit 8.
Thereafter, similarly to the prediction coefficient generating circuit 42, compression processing in the gradation direction, blocking processing, decoding processing, and class generation processing are performed, and the index signal S16 and the blocking signals S18 and S24 are sent to the coefficient calculation circuit 44. .

The coefficient calculation circuit 44 calculates a prediction coefficient representing the correlation between the blocked signal S18 and the blocked signal S24 for each index code indicated by the index signal S16. At this time, since the thinning process performed by the thinning circuit 6 is 1/4 thinning, the coefficient calculating circuit 44 calculates prediction coefficients for three thinned pixels.
Each coefficient calculated in this way is output as coefficient data S26 and written in the memories M2 to M4 (FIG. 4).

(6) Operation and Effect of Embodiment In the above configuration, it is assumed that the user encounters a scene that he / she wants to see repeatedly while viewing the display image. At this time, when the user selects the replay mode by operating the operation panel or the remote control device (not shown) on the front surface of the casing of the television signal receiving device 1, the mode selection signal S7 is given to the control unit 10. It is done.
The control signal S8 sent from the control unit 10 according to the selection content indicated by the mode selection signal S7 stops the update storage of the storage device 9, and the compressed image data S5 and the motion area data stored in the storage device 9 S6 is repeatedly read.

The control signal S8 is also given to the switches 4 and 14, and the switches 4 and 14 are switched.
When the selection content indicated by the mode selection data S7 is for displaying a replay image, the control signal S8 normally switches the switch 4 connected to the route RT1 side to the route RT2 side, and usually restores the restored image signal. The switch 14 connected to the S9 side is not switched.
In this way, the restored image signal S9 restored by the decoding circuit 11 is given to the monitor 17 via the switch 14, the digital / analog converter 15, the switch 4 and the television signal processing circuit 16 in this order, and the replay image is sent to the monitor 17. Is displayed repeatedly.

When the selection content indicated by the mode selection data S7 is for displaying an enlarged replay image, the control signal S8 normally switches the switch 4 connected to the path RT1 side to the path RT2 side, The switch 14 connected to the restored image signal S9 side is switched to the enlarged image signal S10 side.
As a result, the enlarged image signal S10 is given to the monitor 17 via the switch 14, the digital / analog converter 15, the switch 4, and the television signal processing circuit 16 in order, and an enlarged replay image with an enlarged motion region is given to the monitor 17. Display repeatedly.
Thus, the user can display a replay image and an enlarged replay image of a desired scene on the monitor 17.

In the zoom processing circuit 13, an enlarged image signal S10 is formed based on the compressed image data S5 and the motion area data S6 read from the storage device 9. The enlarged image signal S10 is formed by enlarging the motion region in the image.
Normally, the part of interest in the image is an area where a moving object is present, so that the part of interest can be displayed as an image with a favorable size by enlarging the moving area in this way. it can.

In addition, the storage device 9 sequentially overwrites and updates the given compressed image data S5 and motion region data S6 at regular intervals until the replay mode is selected. As a result, the latest image data within a predetermined time can always be stored.
Incidentally, the time during which image data can be stored in the storage device 9 is determined by the capacity of the medium constituting the storage device 9 and the bit rate after compression. For example, when the storage capacity of the storage device 9 is 2 M [bits] and the information amount of the compressed image is compressed to 1 M [bps], the storage device 9 can store image data for 2 seconds. .

The compressed image data S5 is converted into a restored image signal S9 that is a replay image of a normal size and an enlarged image signal S10 that is an enlarged replay image by using a classification adaptation process.
Class classification adaptive processing classifies a given image signal by patterning it, and predicts image data that is not included in the input signal using the predicted value obtained by learning in advance for each class. Thus, the image signal conversion method generates an output image signal having a higher resolution than the input image signal.

The class classification adaptive processing used in the decoding circuit 11 and the zoom processing circuit 13 interpolates the pixels lost due to the thinning-out processing at the time of compression encoding and increases the resolution reduced by the compression in the gradation direction. In addition, the resolution can be improved by interpolating pixels that are insufficient when generating an enlarged replay image.
In this way, a replay image and an enlarged replay image with good image quality can be generated.

  In addition, by using class classification adaptive processing, the reduced image quality of the image data can be restored, so that a compression encoding method with a high compression rate that uses multiple compression processes can be used regardless of the degradation in image quality. Can do. As a result, the amount of information of the input image signal can be greatly compressed, and long-time image data can be recorded.

According to the above configuration, the input image signal that has been compression-encoded by the high-compression compression encoding method and temporarily stored is repeatedly read out when the replay mode is selected, and the degraded image quality is detected using the class classification adaptive processing. The restored replay image was generated.
Thus, the television signal receiver 1 and the television signal receiving method capable of repeatedly displaying a replay image of a desired scene of an image based on the received television signal with good image quality and capable of displaying it for a long time can be realized.

In addition, the motion area data representing the motion area in the image is temporarily stored together with the compression-encoded input image signal, and when the enlargement replay mode is selected, the storage is repeatedly read out, and the image quality that decreases during the enlargement process is reduced. Improved enlarged replay images were generated using the classification adaptation process.
Thus, the television signal receiving apparatus 1 and the television signal receiving method capable of repeatedly displaying an enlarged replay image in which a motion area of a desired scene of an image based on the received television signal is enlarged with good image quality can be realized.

(7) Other Embodiments In the above-described embodiments, the case where a television broadcast signal is received as an input image signal has been described. However, the present invention is not limited to this, and a recording medium such as a VTR or a video disk is used. The present invention can also be applied when inputting an image signal obtained by reproduction as an input image signal. In this case, if the image signal is recorded in the digital format on the recording medium, the motion detection circuit comprising the compression coding means comprising the thinning circuit 6 and the motion detection circuit 7 directly without going through the analog / digital converter 5. The image signal can be input to the means.

  Further, in the above-described embodiment, a case where a normal television image based on the input image signal S1 and a replay image based on the restored image signal S9 or an enlarged replay image based on the enlarged image signal S10 are selectively switched and displayed. As described above, the present invention is not limited to this, and both images may be displayed in the PinP (Picture in Picture) format by the configuration shown in FIG.

  That is, in FIG. 10, in which parts corresponding to those in FIG. 1 are assigned the same reference numerals, 60 denotes a television signal receiver as a whole, and a thinning circuit that performs thinning processing on the input restored image signal S9 or enlarged image signal S10 61, except for the addition of a memory 62 for storing the thinned image signal and a digital / analog converter (D / A) 63 for converting the image signal sent from the memory 62 into an analog signal. It is configured in the same manner as the television signal receiver 1 shown.

The thinning circuit 61 performs a thinning process on the restored image signal S9 or the enlarged image signal S10 sent through the switch 14 by the user's selection operation, thereby reducing the image to fit within a PinP (Picture in Picture) format child screen. Processed and stored in the memory 62 as the reduced image signal S32.
The memory 62 is a frame memory, which superimposes the image signal S3 supplied from the analog / digital converter (A / D) 5 and the reduced image signal S32 supplied from the thinning circuit 61 on the address, and displays them on the parent screen. An image signal S33 composed of a screen with a screen added is generated and output and displayed on the monitor 17 via the digital / analog converter (D / A) 63, the switch 4 and the signal processing circuit 16 in this order.

  With this configuration, the television signal receiving device 60 can display a replay image of a desired scene or an enlarged replay image in which the motion area of the image is enlarged at the same time as a normal television image.

  In the above-described embodiment, the class generation circuit 27 and the prediction coefficient generation circuits 42 and 45 take out only the MSB from the pixel code in the block signal, and arrange the 1/0 of the MSB in order. Although the case where signal pattern classification is performed has been described, the present invention is not limited to this. For example, DCT (Discreat Cosine Transform) encoding, DPCM (differential encoding), vector quantization, subband encoding, wavelet transform, etc. Using this compression method, class classification processing by pattern classification may be performed.

  Further, in the above-described embodiment, the case where ADRC encoding is used as the index code generating means used in the class classification adaptive processing in the zoom processing circuit 13 has been described. However, the present invention is not limited to this. For example, class classification processing by pattern classification may be performed using compression methods such as DCT (Discreat Cosine Transform) coding, DPCM (differential coding), vector quantization, subband coding, and wavelet transform. .

  In the above-described embodiment, the compression for compressing the input image signal includes the thinning means for thinning out pixels in the spatio-temporal direction by the MUSE method and the gradation direction compression means for compressing in the gradation direction by ADRC encoding. Although the case of using as an encoding method has been described, the present invention is not limited to this, and another compression encoding method may be used instead of the compression encoding method.

  In the above-described embodiments, the case where the ROM is used as the coefficient memory storing the prediction coefficient has been described. However, the present invention is not limited to this, and instead of the ROM, a RAM (Random Access Memory) or an SRAM (Static RAM). Etc. may be used.

  The present invention can be widely used for applications in which video is temporarily recorded and reproduced.

1 is a block diagram illustrating a television signal receiver according to an embodiment of the present invention. It is a block diagram which shows the structure of a motion detection circuit. It is a basic diagram provided for description of detection of a motion region. It is a block diagram which shows the structure of a decoding circuit. It is a block diagram which shows the structure of a zoom processing circuit. It is a basic diagram with which it uses for description of the image expansion of a setting area | region. It is a block diagram which shows the structure of a zoom processing circuit. It is a block diagram which shows the structure of a prediction coefficient production circuit. It is a block diagram which shows the structure of a prediction coefficient production circuit. It is a block diagram which shows the structure of the television signal receiver which added the PinP function.

Explanation of symbols

1, 60 ... Television signal receiver, 2 ... Antenna, 3 ... Tuner, 4, 14 ... Switch, 5 ... Analog / digital converter, 6, 61 ... Decimation circuit, 7 ... Motion detection Circuit, 8 ... gradation compression circuit, 9 ... storage device, 10 ... control unit, 11, 26 ... decoding circuit, 12, 35, 36 ... delay circuit, 13 ... zoom processing circuit, 15, 63 …… Digital / analog converter, 16 …… TV signal processing circuit, 17 …… Monitor, 20, 62 …… Memory, 21 …… Cut out circuit, 22 …… Differential circuit, 23, 29, 38 …… Calculation , 23A to 23L... Integration circuit, 24... Detection circuit, 25, 28, 34, 39, 43... Blocking circuit, 27 and 37. Class generation circuit, 29A to 29D. 30, 40 ... Multiplexing circuit, DESCRIPTION OF SYMBOLS 1 ... Area | region setting circuit, 32 ... Enlarged image production | generation part, 33 ... Enlargement ratio calculation circuit, 41 ... Synthesis | combination circuit, 42, 45 ... Prediction coefficient creation circuit, 44 ... Coefficient calculation circuit, M1-M4 ... ... coefficient memory.

Claims (4)

Division means for dividing the first image signal, which has been subjected to compression processing by thinning out pixels in the video signal and whose resolution has been degraded, into a first divided image signal having a pixel of interest and a plurality of pixels around the pixel of interest as unit blocks; ,
Level of teacher data and to correct the errors occurring between the decoding teacher data decoded by extension our method corresponding to the compression processing after the teacher data compressed by the compression processing, an image signal from which the decoding teacher data represents After classifying into a plurality of classes classified according to the distribution pattern, storage means for storing the prediction coefficient calculated for each class,
First class classification means for classifying the first divided image signal classified according to the level distribution pattern of the first divided image signal into one of the plurality of classes ;
Prediction calculation processing is performed to generate pixel values of the thinned pixels by performing a product-sum operation on the prediction coefficient corresponding to the class to which the first divided image signal belongs and the pixel value represented by the first image signal. A predictive calculation means for forming a second image signal having a resolution higher than that of the first image signal;
An area corresponding to the motion area in the second image signal is selected as an enlargement area to be enlarged from an area number indicating a movement area having motion in the video signal, and a pixel of interest in the enlargement area and the periphery of the pixel of interest Are divided into second divided image signals having a plurality of pixels as unit blocks, and are classified according to the level distribution pattern of the second divided image signals, whereby the second divided image signals are divided into a plurality of classes. The pixel value of the pixel to be interpolated by classifying into one of the classes and performing a product-sum operation on the prediction coefficient corresponding to the class to which the second divided image signal belongs and the pixel value represented by the second divided image signal. An image signal conversion apparatus comprising: an enlarged image forming unit that increases a spatial resolution of the enlarged image signal by performing a prediction calculation process to be generated . In the storage means,
As the above prediction coefficient,
A coefficient necessary for generating pixels thinned out in the space-time direction, the first coefficient obtained by learning;
A coefficient necessary for increasing the resolution reduced by the compression processing in the gradation direction, and the second prediction coefficient obtained by learning are stored;
The dividing means is
Dividing the first image signal compressed in the spatio-temporal direction and the gradation direction into divided image signals;
The prediction calculation means is
The pixel values of the thinned pixels are generated by multiplying the prediction coefficient corresponding to the class to which the first divided image signal belongs and the first image signal, and the resolution in the gradation direction is improved. The image signal converter according to claim 1, wherein The above prediction coefficient is
The compressed image data is formed by compressing the teacher data by a predetermined compression processing method, the decoded image data is formed by decoding the compressed image data by the compression processing by the compression processing method,
Based on the compressed image data, the pixel of interest and a plurality of pixels around the pixel of interest are blocked as unit blocks to obtain a first block signal,
Pattern classification for each first block, determine the class to which the block belongs,
Based on the decoded image data, the pixel of interest and a plurality of pixels around the pixel of interest are blocked as unit blocks to obtain a second block signal,
Based on the teacher data, a pixel of interest and a plurality of pixels around the pixel of interest are unit-blocked to obtain a third block signal,
2. The image signal conversion apparatus according to claim 1, wherein each of the determined classes is created by an operation using the second block signal and the third block signal corresponding to the second block signal. A division step of dividing the first image signal, which has been compressed and reduced in resolution by thinning out pixels in the video signal , into a first divided image signal having a target pixel and a plurality of pixels around the target pixel as unit blocks; ,
A first class classification step of classifying the first divided image signal classified according to the level distribution pattern of the first divided image signal into one of the plurality of classes ;
Level distribution of training data and the image signal representing the above decoding teacher data to correct the error that occurs between the decoding teacher data decoded by extension our method corresponding to the compression processing after the teacher data compressed by the compression processing The prediction coefficient and the first image signal corresponding to the class to which the first divided image signal belongs among the prediction coefficients calculated for each class after classification into a plurality of classes classified according to patterns. Prediction that forms a second image signal with improved resolution over the first image signal by performing a prediction calculation process that generates a pixel value of a thinned pixel by performing a product-sum operation on the pixel values to be represented A computation step;
An area corresponding to the motion area in the second image signal is selected as an enlargement area to be enlarged from an area number indicating a movement area having motion in the video signal, and a pixel of interest in the enlargement area and the periphery of the pixel of interest Are divided into second divided image signals having a plurality of pixels as unit blocks, and are classified according to the level distribution pattern of the second divided image signals, whereby the second divided image signals are divided into a plurality of classes. The pixel value of the pixel to be interpolated by classifying into one of the classes and performing a product-sum operation on the prediction coefficient corresponding to the class to which the second divided image signal belongs and the pixel value represented by the second divided image signal. An image signal conversion method comprising: an enlarged image forming step for increasing the spatial resolution of the enlarged image signal by performing a prediction calculation process to be generated Law.

Images