WO2000065830A1 - Dispositif et procede de conversion d'image - Google Patents
Dispositif et procede de conversion d'image Download PDFInfo
- Publication number
- WO2000065830A1 WO2000065830A1 PCT/JP2000/002636 JP0002636W WO0065830A1 WO 2000065830 A1 WO2000065830 A1 WO 2000065830A1 JP 0002636 W JP0002636 W JP 0002636W WO 0065830 A1 WO0065830 A1 WO 0065830A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pixel
- field
- image signal
- input image
- mode
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0117—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving conversion of the spatial resolution of the incoming video signal
- H04N7/012—Conversion between an interlaced and a progressive signal
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0127—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level by changing the field or frame frequency of the incoming video signal, e.g. frame rate converter
- H04N7/0132—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level by changing the field or frame frequency of the incoming video signal, e.g. frame rate converter the field or frame frequency of the incoming video signal being multiplied by a positive integer, e.g. for flicker reduction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0135—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes
Definitions
- the present invention relates to an image conversion device and method suitable for use in a television receiver, a VTR, and the like.
- a field frequency of a television scanning method 50 Hz (PAL method, SEC AM method, etc.) or 60 Hz (NTSC method) has been widely adopted.
- the field frequency is relatively low, such as 50 Hz, there is a problem that a large area of the display image is flickered, that is, a so-called field flit is generated.
- FIG. 17 schematically shows an example and another example of a pixel structure obtained by the field double-speed processing, in which the vertical axis represents a vertical position v and the horizontal axis represents time t.
- an output image signal having a period of 1 to 2 of the field period of the input image signal is formed. Since the input image signal is interlaced, the vertical position (line position) of the input pixel indicated by a white circle is shifted by 12 between two fields that are temporally continuous.
- the pixel structure shown in FIG. 17A is obtained by doubling the speed so that the double-speed field of the output image signal has an instantaneous relationship.
- three pixels during input are supplied to the median filter in order to form output pixels of a field not existing in the input image signal.
- an output pixel (indicated by a star mark) is generated by a median fill from three input pixels surrounded by a dashed triangle line. Mede Ianfill outputs a pixel having an intermediate pixel value among the three input pixels.
- the median filter processing is shown for some output pixels, but all output pixels other than the output pixel that matches the input pixel position are processed by the media: ⁇ filter. Generated.
- the output image signal having the pixel structure shown in FIG. 17A is called an ABAB type.
- the pixel structure of the output image signal shown in FIG. 17B is double-speed so that a pixel of a new field exists at the same vertical position as the temporally previous field of the input image signal. It has been made. In Fig. 17B, the speed can be doubled by repeatedly outputting the previous field in time.
- the output image signal having the pixel structure shown in FIG. 17B is called an AAB B type.
- the ABAB type Since the AAB B type repeatedly outputs the same field, the resolution in the time direction is not improved by doubling the speed. Rather, in the case of an image taken by panning a video camera, a double image is generated and the image quality may be degraded.
- the ABAB type has advantages over the AAB B type in terms of spatio-temporal resolution (temporal and spatial resolution). However, in the ABAB type, three input pixels included in the previous and next fields are processed by median fill and the output pixels are interpolated. In the case of, degradation occurs in which the vertical line looks like a comb tooth. For example, when a screen in which numbers and characters are lined up like a stock information scrolls horizontally, there is a disadvantage that the numbers and characters are difficult to see.
- Another object of the present invention is to adapt the ABAB type to the pattern of the input image. It is an object of the present invention to provide an image conversion device and method capable of improving image quality by switching between AABB and AABB types.
- the invention according to claim 1 provides an input image signal in which a pixel position of a first field and a pixel position of a second field which are adjacent to each other are different from each other.
- an image conversion apparatus for converting into an output image signal having a field frequency of N times N is an integer of 2 or more
- At least all fields not present in the input image signal are sequentially set as a field of interest, and for each pixel of interest of the field of interest, based on a plurality of pixels of the input image signal determined based on the pixel of interest.
- a class determination unit that determines a class for the pixel of interest; a memory unit that stores prediction information obtained in advance;
- a prediction pixel selection unit that selects a plurality of pixels of the input image signal for each pixel of interest
- a pixel generation unit that generates each target pixel of the output image signal based on prediction information corresponding to the class determined by the class determination unit and the plurality of pixels selected by the prediction pixel selection unit;
- the pixel position of the output image signal changes in units of fields equal to N, and the pixel position of each field coincides with the pixel position of one of the first and second fields of the input image signal.
- the invention according to claim 13 is characterized in that an input image signal in which the pixel position of the first field and the pixel position of the second field adjacent to each other are different from each other is N times (N is the field frequency of the input image signal) , An integer greater than or equal to 2) to an output image signal having a field frequency of hand,
- At least all fields not present in the input image signal are sequentially set as a field of interest, and for each pixel of interest of the field of interest, based on a plurality of pixels of the input image signal determined based on the pixel of interest.
- the pixel position of the output image signal changes in units of fields equal to N, and the pixel position of each field is the same as the pixel position of one of the first and second fields of the input image signal.
- This is an image conversion method characterized by matching.
- a pixel value of an output field that does not exist in the input image signal is generated by the class classification adaptive processing using the image information of the preceding and succeeding input fields. Therefore, as compared with the conventional method of simply outputting the same field repeatedly, it is possible to solve the problem that the second output field has image information corresponding to the time and has no time resolution.
- the class classification adaptive processing a class is detected based on a plurality of pixels of an input image signal, and a pixel value of an output image signal is created using an estimated prediction formula that is optimal for each class. This makes it possible to improve the resolution of the output image signal as compared with the case of performing the operation.
- AABB Eve and ABAB type can be selectively generated. Therefore, it is possible to select the AB AB type for a picture requiring a spatio-temporal resolution as the input image signal, and to select the A AB B type when viewing many horizontally moving telops. Become.
- FIG. 1 is a schematic diagram schematically showing a pixel structure used for describing a first embodiment of the present invention.
- FIG. 2 is a block diagram of a second embodiment of the present invention.
- FIG. 3 is a waveform chart for explaining field doubling.
- FIG. 4 is a schematic diagram for explaining a vertical synchronizing signal with respect to a field double speed output image signal.
- FIG. 5 is a schematic diagram showing an arrangement of pixels of an input image signal and pixels of an AABB type output image signal.
- FIG. 6 is a schematic diagram showing an arrangement of pixels of an input image signal and pixels of an ABAB type output image signal.
- FIG. 7 is a schematic diagram showing a tap structure when generating an AABB type output image signal.
- FIG. 8 is a schematic diagram showing an example of a prediction tap and a cluster tap in the AABB type.
- FIG. 9 is a schematic diagram illustrating an example of prediction taps and class taps in the AAB B type.
- FIG. 10 is a schematic diagram showing a tap structure when an ABAB type output image signal is generated.
- FIG. 11 is a schematic diagram showing an example of prediction taps and cluster taps in the ABAB type.
- Fig. 12 shows prediction taps and clusters in ABAB type. It is an approximate line figure showing other examples of a loop.
- FIG. 13 is a schematic diagram showing still another example of the prediction tap and the cluster tap in the ABAB type.
- FIG. 14 is a block diagram showing a configuration when learning a prediction coefficient.
- FIG. 15 is a schematic diagram for explaining the positional relationship between the pixels of the teacher image and the pixels of the student image when learning the prediction coefficients for the AABB type.
- FIG. 16 is a schematic diagram for explaining the positional relationship between the pixels of the teacher image and the pixels of the student image when learning the prediction coefficients for the ABBA type.
- FIG. 17 is a schematic diagram for explaining one example of the conventional field doubling process and another example.
- the field double speed processing device used in the first embodiment generates at least pixels of an output field that do not exist in an input image signal by a class classification adaptive process.
- Fig. 1 schematically shows the pixel structure obtained by the field double-speed processing, as in Fig. 17, with the vertical axis representing the vertical position v and the horizontal axis representing time t.
- the input image is indicated by a white circle and the output image is indicated by a black circle or star.
- FIG. 1A shows a conventional process similar to FIG. 17A, and is a diagram for comparison with the process according to the present invention. In other words, it indicates that double speed processing is performed by repeatedly outputting the same field as the field in the input image signal.
- FIG. 1B shows the process of doubling the speed according to the present invention.
- the input image signal is Pixel X 1 belonging to the temporally previous field of the field that does not exist inside, and having the same vertical position, pixels X 2 and X 3 above and below, and the temporally subsequent field And use pixels X 4 and X 5 at the top and bottom in the vertical direction to generate pixel y for the new field.
- This processing is a classification adaptive processing. Note that, as is clear from the description of the second embodiment described later, the input pixels used in the classification adaptive processing are not limited to xl to x5.
- the input pixels of the preceding and succeeding input fields are used, and the output pixels of the new field are generated by the classification adaptive processing.
- a new field with time resolution can be created. Therefore, it is possible to prevent a double image from being generated on a panned screen as in the conventional method of repeatedly outputting the same field.
- the input image signal is an interless signal having a field frequency of 60 Hz and the number of lines is 525, and the input image signal is converted to a field frequency of 120 Hz and the number of lines is increased by a field doubling process. It is designed to convert it into 5 2 5 output image signals.
- the pixel structure of the output image signal it is possible to switch between the AAB type and the ABAB type according to a user setting or in accordance with the input image signal.
- pixels are generated by the method described with reference to FIG. 1B.
- pixels are generated by the classification adaptive processing.
- the spatio-temporal resolution can be higher than the conventional method using the median filter.
- the classification adaptive processing has a storage unit that performs class division according to temporal and / or spatial characteristics of an image signal as an input signal, and stores a prediction coefficient value obtained by learning in advance for each class. This is a method to output the optimum estimated value by the calculation based on the prediction formula.
- the resolution can be increased more than that of the input image signal by the class classification adaptive processing.
- an input interface image signal (525 lines, Z 60 Hz) is supplied to a region cutout unit 1 and is required for class classification and prediction calculation.
- An area including a plurality of pixels is cut out.
- the output of the region cutout unit 1 is supplied to the class detection circuits 2 and 12, and the prediction tap selection circuits 3 and 13.
- the class detection circuits 2 and 12 detect a class corresponding to the characteristics of an input pixel near an output pixel to be created.
- the class detection circuits 2 and 12 may detect a motion class.
- the two configurations for generating pixels in parallel are provided to generate a pixel in a field existing in the input image signal and to generate a pixel in a field not present in the input image signal. .
- the classes detected by the class detection circuits 2 and 12, respectively, are supplied to the prediction tap selection circuits 3, 13 and the prediction coefficient memories 4, 14.
- the prediction coefficient set corresponding to the class is read from the prediction coefficient memories 4 and 14, and the read prediction coefficient set is supplied to the product-sum operation circuits 5 and 15.
- the prediction tap selection circuits 3 and 13 are configured to select a prediction tap to be used according to the class. When the prediction coefficient set of each class is obtained in advance by learning, both the prediction coefficient set and the information on the position of the prediction tap to be used are obtained.
- the prediction tap selection circuits 3 and 13 are provided with memories in which prediction tap position information is stored for each class. Have been.
- Predicted tap position information read from the memory corresponding to the class is supplied to a selector for switching the evening tap, and the selector selectively outputs a predicted tap.
- the prediction taps from the prediction tap selection circuits 3 and 13 are supplied to the product-sum operation circuits 5 and 15.
- the product-sum operation circuits 5 and 15 calculate the data of the output image signal using a linear estimation formula of the prediction tap (pixel of the input image signal) and the prediction coefficient set.
- the sum-of-products arithmetic circuits 5 and 15 simultaneously output the first and second pixel values of adjacent fields of the field double speed output image signal, respectively.
- the first pixel value from the product-sum operation circuit 5 is supplied to the field doubler 6, and the second pixel value from the product-sum operation circuit 15 is supplied to the field doubler 16.
- Each of the field doublers 6 and 16 has a field memory and performs a process of doubling the field frequency of each of the first and second fields generated by the class classification process.
- the outputs of the field doublers 6 and 16 are input to a selector 7 that can be switched for each field.
- the selector 7 alternately selects the outputs of the field doublers 6 and 16 to generate an output image signal (525 lines / 120 Hz output signal).
- the output image signal is supplied to, for example, a CRT display.
- the synchronizing system of the CRT display is configured so that an output image signal can be displayed.
- a broadcast signal or a playback signal of a playback device such as a VTR is supplied. That is, this example can be incorporated in a television receiver.
- a selection signal generator 8 is provided.
- the selection signal generating section 8 generates selection signals SL1 and SL2 for instructing whether to generate an AABB type or an ABAB type as an output image signal.
- the high and low levels of the selection signals SL 1 and SL 2 AABB type and ABAB type are associated with each other.
- the selection signal generator 8 generates selection signals SL1 and SL2 by manually operating switches and the like while viewing the image presented by the output image signal. Not only manual operation but also selection signals SL 1 and SL 2 may be generated in response to an instruction of a control signal sent together with the input image signal. Further, the selection signals SL1 and SL2 may be automatically generated according to the characteristics of the input image signal itself.
- the selection signal S L 1 controls the extraction method of the area extraction unit 1.
- the selection signal S L2 controls the setting of the class taps and prediction taps in the class detection circuits 2 and 12 and the prediction tap selection circuits 3 and 13.
- the prediction coefficient tables of the prediction coefficient memories 4 and 14 are switched by the selection signal SL2.
- the selection signal SL2 is supplied to the synchronizing signal generator 9, and a vertical synchronizing signal suitable for each type of output image signal is generated.
- the vertical deflection of the CRT monitor (not shown) is controlled by the vertical synchronization signal.
- FIG. 3 shows the operation of the field doublers 6 and 16 with time on the horizontal axis and the addresses of the field memories in the field doublers 6 and 16 on the vertical axis.
- the data of the first field and the data of the second field are simultaneously generated by the multiply-accumulation circuits 5, 15.
- FIG. 3A shows the address when the image signal of the first field generated by the product-sum operation circuit 5 is written into the field memory
- FIG. 3C shows the second address generated by the product-sum operation circuit 15. This is the address when writing the image signal of the field No. into the field memory.
- the numbers 1 A, 1 B, 2 A, 2 B,... Added to each field indicate the two fields (A and B) that make up one frame. Also, ignoring the delay caused by the class classification adaptive processing, the input image signal and the product-sum operation circuits 5, 15 Numbers are assigned to the output image signals.
- the output image signal of the product-sum operation circuit 5 shown in FIG. 3A is supplied to the field doubler 6.
- one field (for example, 1A in FIG. 3A) of the output of the multiply-accumulate operation circuit 5 is written into the field memory, and is twice as fast as the write speed so as not to overtake this write address.
- One field (for example, 1A in Fig. 3B) is read from the field memory at the speed. Reading of one field is repeated twice. Therefore, as shown in FIG. 3B, an output having a doubled field frequency is generated from the field doubler 6.
- the other field doubler 16 also generates an output signal (FIG. 3D) in which the field frequency of the operation signal from the product-sum operation circuit 15 is doubled, similarly to the field doubler 6.
- the output signals of these field doublers 6 and 16 are supplied to a selector 7, and the selector 7 alternately selects each output signal for each field.
- an output image signal shown in FIG. 3E is obtained from the selector 7.
- the output image signal is obtained by alternately positioning the image signal of the first field generated by the product-sum operation circuit 5 and the image signal of the second field generated by the product-sum video circuit 15 in the time axis direction. This is an image signal whose field frequency is doubled.
- the configuration of the field doubling speed of the class classification adaptive processing shown in FIG. 2 corresponds to one component of the color image signal, and is divided into three components of a luminance signal and two color difference signals.
- the configuration shown in FIG. 2 may be provided.
- the present invention can be applied not only to a component color image signal but also to a composite color image signal.
- FIG. 4 illustrates the vertical sync signal generated by the sync signal generator 9.
- FIG. 4A When the output image signal is of the ABAB type, raster scanning is performed in an interless mode, as shown in FIG. 4A. That is, scanning is performed from the first line of the first field to the second field in the middle of the 263rd line, and scanning is performed up to the 525th line.
- the line (dashed line) of the second field exists at the center position between the lines above and below the first field.
- a vertical synchronizing signal having a period of 262.5 lines is output from the synchronizing signal generator 9.
- the operation is performed at the physical position of the last scan.
- a scan of line 262 from line 1 to line 262 is performed, and in the second field, line 1 to line 262 of the previous field are identical. Is scanned, and further, the scanning of 262.5 lines up to the first half of the 2263rd line is performed.
- the third field scanning starts from the latter half of the 263rd line, and 263 lines up to the first half of the 263rd line are scanned.
- scanning is performed from the latter half of line 263, and scanning of line 262.5 up to line 262 is performed. The fourth raster scan returns to the first scan.
- the synchronization signal generator 9 In order to perform one raster scan shown in FIG. 4C, the synchronization signal generator 9 generates a vertical synchronization signal in the AABB type as shown in FIG. 4D. Thus, the synchronization signal generator 9 The vertical synchronization signal is generated according to the type indicated by the selection signal SL2 from the selection signal generator 8. The vertical synchronization of the CRT monitor is controlled by this vertical synchronization signal, and the output image signal of the field double speed is displayed.
- the classification adaptive processing according to the second embodiment of the present invention will be described in more detail.
- Pixels in the fields existing in the input image signal are generated by the classification process. That is, the pixels in the field that do not exist in the input image signal are generated by the class detection circuit 2, the prediction tap selection circuit 3, the prediction coefficient memory 4, and the product-sum operation circuit 5. Pixels in the field that are not present in the input image signal are generated by the class detection circuit 12, the prediction tap selection circuit 13, the prediction coefficient memory 14, and the product-sum operation circuit 15. Then, the AABB type output image signal having the pixel arrangement shown in FIG. 5 is generated by the field doublers 6 and 16 and the selector 7.
- FIG. 6 schematically shows a pixel position when an ABAB type output image signal is generated when the vertical axis is the vertical direction and the horizontal axis is the time direction.
- Pixels in the fields existing in the input image signal are generated by the class detection circuit 2, the prediction tap selection circuit 3, the prediction coefficient memory 4, and the product-sum operation circuit 5.
- Pixels in the fields not present in the input image signal are generated by the class detection circuit 12, the prediction tap selection circuit 13, the prediction coefficient memory 14, and the product-sum operation circuit 15. Then, the field doublers 6 and 16 and the selector 7 generate an ABB type output image signal having the pixel arrangement shown in FIG.
- FIG. 7 shows the vertical and temporal tap structures.
- Mode 1 is a mode in which pixels in one of the odd and even fields present in the input image signal are This is a process of generating a pixel at the same position in the direction as a pixel of the first field.
- Mode 2 is a process of generating a pixel in a field that does not exist in the input image signal at the same position in the vertical direction as a pixel in one field as a pixel in the second field.
- Mode 3 is a process to generate a pixel as a third field pixel at the same position in the vertical direction as a pixel in the other field of the odd field and the even field existing in the input image signal. is there.
- Mode 4 is a process of generating a pixel in a field that does not exist in the input image signal, at the same position in the vertical direction as a pixel in the other field, as a pixel in the fourth field.
- the input pixels in the fields before and after the newly generated field are class-tapped or predicted as shown by the triangles in Fig. 7. Used as a tap. That is, it is included in the field (referred to as one field) before the new field to be generated, is included in the pixel at the same position as the pixel to be generated, and is included in the field (0 field) after the new field. Pixels that are vertically above and below the pixel to be used are used as evening pixels.
- FIG. 8A is an example of a prediction tap in mode 1 and mode 3
- FIG. 8B is an example of a class tap in mode 1 and mode 3.
- 8A and 8B show the arrangement of pixels with the vertical axis being the vertical direction and the horizontal axis being the horizontal direction.
- Fig. 8A the pixels to be generated (black circles) y included in one field and the six input pixels x vertically above and below.
- ⁇ x 5, 0 is included in the field, four input pixels x 6 ⁇ x of the pixel y to be cane product vertically adjacent left and right positions, the prediction tap by the sum 1 0 of the input pixel with the structure
- FIG. 8B is an example of class taps in mode 2 and mode 4.
- 9A and 9B show the pixel arrangement with the vertical axis being the vertical direction and the horizontal axis being the horizontal direction.
- the five input pixels x that are included in the —1 field and that are the same in the vertical direction as the pixel to be generated (black circle) y.
- ⁇ x 4, 0 contained in a field by a total of 1 one input pixel of the six input picture element x 5 ⁇ x 1 0 of the upper and lower positions in the vertical direction to the pixel y to be generate A prediction tap is configured.
- the spatial position of the generated pixel y and the input pixel X2 is the same.
- five input pixels X By to X 4, class tap mode 2 and mode 4 is formed. X.
- X and X 2 are pixels included in one field and located vertically above and below the pixel y to be generated.
- X 3 and X 4 are included in the 0 field, the pixel positions of the upper and lower in vertical direction with respect to the pixel y to be cane product.
- FIG. 10 shows the vertical and temporal sunset structures.
- Mode 1 is for one of the fields present in the input image signal. This is a process of generating a pixel at the same position in the vertical direction as the pixels in the inside. mode
- Reference numeral 2 denotes a pixel in a field which does not exist in the input image signal, and generates a pixel at a position shifted by one or two lines in the vertical direction from a pixel in one field.
- Mode 3 is a process for generating a pixel at a position shifted by 12 lines in the vertical direction from a pixel in the other field existing in the input image signal.
- Mode 4 is a process of generating a pixel in a field that does not exist in the input image signal at the same position in the vertical direction as a pixel in the other field.
- mode 2 and mode 4 the input pixels of the fields before and after the new field including the pixel to be generated are used as taps as shown by the triangles in FIG.
- mode 2 — The pixels included in the 1 field and located vertically above and below the pixel to be generated, and the pixels included in the 0 field and located at the same vertical position as the pixel to be generated Is used as a tap.
- mode 4 the pixels that are included in one field and are at the same vertical position as the pixel to be generated, and the pixels that are included in the 0 field and are vertically above and below the pixel to be generated are Is used as a tap.
- FIG. 11A is an example of a prediction tap in mode 2
- FIG. 11B is an example of a class tap in mode 2.
- Fig. 11 A And Fig. 11B shows the pixel arrangement with the vertical axis being the vertical direction and the horizontal axis being the horizontal direction.
- ⁇ x 5, 0 contained in a field, in the same position in the pixel y direction perpendicular to you'll generate five input pixels x 6 ⁇ x,.
- the prediction tap is composed of a total of 11 input pixels.
- FIG. 11 B five input pixels X.
- motor - class tap de 2 is constructed.
- X. And X are included in the —1 field and are pixels vertically above and below the pixel y to be generated.
- X 2 is 0 included in the field is a pixel of the same position as the pixel y to be cane product.
- X 3 and X 4 are included in the 0 field, a pixel position in the vertical in a direction perpendicular to the pixel to be cane product.
- FIG. 12B is an example of a prediction tap in mode 3
- FIG. 12B is an example of a class tap in mode 3.
- FIGS. 12A and 12B show the arrangement of pixels with the vertical axis being the vertical direction and the horizontal axis being the horizontal direction. As shown in FIG. 12A, two input pixels x included in one field and located vertically above and below the pixel y to be generated.
- a prediction tap is composed of a total of 10 input pixels including the two input pixels x 8 and ⁇ 9 at the upper and lower positions.
- X. And X are — contained in one field, and vertically above and below the pixel y to be generated.
- Pixel. X 2 and X 3 are included in the 0 field, a pixel position in the vertical in a direction perpendicular to the pixel y to generated so.
- FIG. 13A is an example of a prediction tap in mode 4
- FIG. 13B is an example of a cluster tap in mode 4.
- FIGS. 13A and 13B show the pixel arrangement with the vertical axis being the vertical direction and the horizontal axis being the horizontal direction.
- Fig. 13A five input pixels x included in one field and located at the same vertical position as the pixel y to be generated.
- ⁇ ⁇ 4 and 6 input pixels x 5 to x 10 that are included in the 0 field and are located vertically above and below the pixel y to be generated.11 Predicted by 1 input pixel A tap is configured.
- X. X4 constitutes a mode 4 class tap.
- X. I a pixel that is included in one field and is at the same vertical position as the pixel to be generated.
- X and X 2 are pixels included in one field and located vertically above and below the pixel to be generated.
- X 3 and X 4 are included in the 0 Fi one field, a pixel position in the vertical in a direction perpendicular to the pixel y to be cane product.
- the prediction tap and class tap in mode 1 are processing to recreate the fields existing in the input signal, and are the same as those in mode 1 of the ABA B type shown in Fig. 8A. Is good.
- the selection signal SL2 is supplied to the class detection circuits 2 and 12 to switch the class tap between the AABB type and the ABAB type described above.
- the selection signal SL2 is supplied to the prediction tap selection circuits 3 and 13 to switch the prediction tap according to the type. Is done. Switching between class tap and prediction tap modes
- the region extracting circuit 1 simultaneously outputs all input pixels that may be used as a class tap and a prediction tap.
- the class detection circuits 2 and 12 are adapted to select a cluster tap according to the mode and the position of the pixel of interest.
- the class detection circuits 2 and 12 detect characteristics of the class tap, for example, a level distribution. In this case, in order to prevent the number of classes from becoming enormous, processing is performed to compress the input data of 8 bits per pixel into data of a smaller number of bits.
- data of an input pixel of a class tap is compressed by ADR C (Adaptive Dynamic Range Coding).
- ADR C Adaptive Dynamic Range Coding
- a compression method such as DPCM (prediction coding) and VQ (vector quantization) may be used in addition to ADRC.
- ADRC is an adaptive requantization method developed for high-efficiency coding for VTRs (Video Tape Recoders) .However, since local features at the signal level can be expressed efficiently with a short word length, In this example, ADRC is used to generate code for classification.
- ADR C the dynamic range of the cluster tap is DR
- the bit allocation is n
- the data level of the pixel of the class tap is L
- the requantization code is Q. Re-quantization is performed by dividing equally between the value MIN and the specified bit length.
- a class may be detected by integrating a motion class with a motion class. In this case, depending on the motion class, The cluster top may be switched.
- the multiply-accumulation circuits 5 and 15 include the prediction taps (pixel values) selected by the prediction tap selection circuits 3 and 13 and the prediction coefficient sets read from the prediction coefficient memories 4 and 14, respectively. Pixel values are generated by a linear linear combination with The pixel value may be generated not only by the linear expression but also by a higher-order estimation expression of second order or higher.
- the prediction coefficient memories 4 and 14 store a prediction coefficient table used in the AABB type and a prediction coefficient table used in the ABAB type. Each table has a plurality of prediction coefficient sets corresponding to the modes and the classes determined by the class detection circuits 2 and 12. Each table (prediction coefficient) is obtained in advance by a learning process described later.
- the product-sum operation circuit 5 includes a prediction tap (pixel value) xl, X 2, ⁇ ⁇ ⁇ ⁇ ⁇ from the prediction tap selection circuit 3 or 13, and a prediction coefficient set w 2 , wi
- pixel values in the case of mode 1 and mode 3 are calculated, for example.
- the product-sum operation circuit 15 calculates the pixel values in the case of the mode 2 and the mode 4 in the same manner.
- the prediction coefficient set is determined in advance for each class by learning, and then stored in the prediction coefficient memories 4 and 14. A calculation is performed based on the input prediction taps and the read prediction coefficient set, and output data corresponding to the input data is formed and output, thereby simply interpolating the input data. Unlike processing, it can output high-quality field double-speed image signals.
- the decimation unit 31 converts the number of pixels in the vertical direction from a progressive image signal having a field frequency of 120 Hz (field double speed) to 1 2 in the vertical direction, and the field frequency 6 It forms an image signal (student image) at 0 Hz.
- the number of pixels in the vertical direction is set to 1 Z2 by the thinning unit 41 to form an image signal (teacher image) having a field frequency of 120 Hz.
- the teacher image is of AABB type or AB AB type.
- the thinning unit 41 forms one of the AABB type and ABAB type teacher images by changing the method of thinning each field in the vertical direction.
- the teacher image from the thinning unit 41 and the student image from the thinning unit 31 are used as a learning pair.
- FIG. 15 shows a pixel structure at the time of learning a prediction coefficient for forming an AABB type output image signal.
- FIG. 16 shows a pixel structure when learning a prediction coefficient for forming an output image signal of an ABAB evening.
- the vertical axis is the vertical direction, and the horizontal axis is the time direction.
- black circles indicate pixels of the progressive image having a field frequency of 120 Hz.
- the thinning-out section 41 thins out 1 to 2 pixels in the vertical direction so that pixels at the same position in the vertical direction in continuous fields are used as pixels of the teacher image.
- 1Z2 thinning in the vertical direction is performed in the thinning unit 41 such that the vertical position has a shift of one line of the progressive image in continuous fields.
- the field circumference Time thinning is performed so that the wave number is 60 Hz, and the vertical position is shifted by one line of the progressive image between temporally continuous fields after the time thinning.
- One-two thinning is performed. The student image formed in this way corresponds to the input image at the time of pixel value generation.
- the student image signal from the thinning section 31 is supplied to the predicted tap area cutout section 32 and the class tap area cutout section 33.
- Class tap The class tap from the region cutout section 33 is supplied to the class detection circuits 34 and 35.
- the prediction tap area cutout unit 32 outputs a prediction tap for creating a pixel in each mode.
- the class detection circuits 34 and 35 compress the cluster map data set for each mode by ADRC, similarly to the class detection circuits 2 and 12 in the pixel generation device shown in FIG. Generate information.
- the class detection circuits 34, 35 independently detect the classes for Modes 1 and 3 and Modes 2 and 4, respectively.
- the prediction tap from the prediction tap area cutout section 32 is supplied to normal equation addition circuits 36 and 37.
- learning of a conversion formula from a plurality of input pixels to output pixels and signal conversion using the prediction formula will be described.
- the prediction coefficient set w,, ⁇ ⁇ ⁇ sets the linear estimation equation of n tap according to w n. before this is shown in equation (3) below.
- Wi is undetermined coefficient.
- y Wi xi + w 2 x 2 + ⁇ ⁇ ⁇ + W «x n (3)
- Learning is performed on multiple signal data for each class.
- the following equation (4) is set according to the equation (3).
- Equation (7) can be rewritten into Equation (10) using a matrix.
- Each of the regular equation addition circuits 36 and 37 in FIG. 14 includes the class information supplied from the class detection circuits 34 and 35 and the prediction tap supplied from the prediction tap area cutout unit 32.
- the normal equation is added using the pixels of the teacher image.
- the normal equation adding circuits 36 and 37 output the normal equation data to the prediction coefficient determination unit 38.
- the prediction coefficient determination unit 38 solves for W i using a general matrix solution such as a sweeping method of a normal equation, and calculates a prediction coefficient set.
- the prediction coefficient determination unit 38 writes the calculated prediction coefficient set in the prediction coefficient memories 39, 40.
- the prediction coefficient memories 39, 40 can estimate the closest true value for estimating the target pixel y of the field double-speed signal for each class.
- the prediction coefficient is stored. Forecast
- the prediction coefficients stored in the measurement coefficient memories 39 and 40 are loaded into the prediction coefficient memories 4 and 14 in the above-described image conversion device.
- the prediction coefficient determination unit 38 obtains more prediction coefficient sets for each class. From the obtained prediction coefficient sets, the prediction coefficient sets to be used are selected in ascending order of absolute value. The selected prediction coefficient set is stored in the addresses corresponding to the classes of the memories 39 and 40, respectively. Therefore, a prediction tap is selected for each class, and the selected position information of the prediction tap is stored in a memory (not shown) for each class. By such a prediction tap selection process, it is possible to select a prediction tap suitable for each class. With the above processing, the learning of the prediction coefficients for creating the image of the field frequency of 120 Hz from the interlaced image signal of the field frequency of 60 Hz is completed by the linear estimation formula.
- the type of an output image signal whose field is doubled according to the pattern of the input image signal can be selected. So you can see high quality playback images.
- the AABB type is generated as the output image signal while performing the field doubling speed
- a field that does not exist in the input image signal is generated using the image information of the preceding and succeeding input fields.
- the generated field becomes image information corresponding to the time. Therefore, it is possible to solve the problem of not having the time resolution of the process of simply repeating the same field as the input field.
- the class classification adaptive processing is used, the spatio-temporal resolution can be improved.
- the AABB type and the ABAB type can be selected as the pixel structure as the output image signal, it is possible to perform the field double speed processing adapted to the pattern of the input image signal. Therefore, in the case of a screen in which characters are scrolled horizontally, for example, a stock information presentation screen, the characters are clearly displayed by selecting the AABB type, and otherwise, by selecting the ABAB type. Images with good spatio-temporal resolution can be displayed.
- the present invention is not limited to the above-described embodiments and the like, and various modifications and applications are possible without departing from the gist of the present invention.
- the description has been given of the double-speed processing of a field frequency of 60 Hz and 120 Hz, the present invention can be similarly applied to the double-speed processing of 50 Hz to 100 Hz.
- the field frequency of the input image signal is not limited to 50 Hz or 60 Hz.
- the processing of the field doubling has been described.
- the processing of doubling the number of pixels in the horizontal and Z or vertical directions is simultaneously performed. May be.
- the field frequency conversion may be performed at an arbitrary ratio without being limited to the double speed.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computer Graphics (AREA)
- Television Systems (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US09/720,348 US6646684B1 (en) | 1999-04-23 | 2000-04-21 | Image conversion device and method |
EP00917418A EP1111918B1 (en) | 1999-04-23 | 2000-04-21 | Image conversion device and method |
DE60041114T DE60041114D1 (de) | 1999-04-23 | 2000-04-21 | Bildumwandlungsvorrichtung und -verfahren |
JP2000614655A JP4470323B2 (ja) | 1999-04-23 | 2000-04-21 | 画像変換装置および方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP11680699 | 1999-04-23 | ||
JP11/116806 | 1999-04-23 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2000065830A1 true WO2000065830A1 (fr) | 2000-11-02 |
Family
ID=14696131
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2000/002636 WO2000065830A1 (fr) | 1999-04-23 | 2000-04-21 | Dispositif et procede de conversion d'image |
Country Status (6)
Country | Link |
---|---|
US (1) | US6646684B1 (ja) |
EP (1) | EP1111918B1 (ja) |
JP (1) | JP4470323B2 (ja) |
KR (1) | KR100730499B1 (ja) |
DE (1) | DE60041114D1 (ja) |
WO (1) | WO2000065830A1 (ja) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1686801B1 (en) * | 2000-02-10 | 2009-12-16 | Sony Corporation | Image processing device and method and recording medium |
JP4691812B2 (ja) * | 2001-03-29 | 2011-06-01 | ソニー株式会社 | 係数データの生成装置および生成方法、それを使用した情報信号の処理装置および処理方法 |
EP1326436B1 (en) * | 2001-12-28 | 2013-02-13 | Sony Corporation | Displaying information |
US8780093B2 (en) * | 2009-03-25 | 2014-07-15 | Himax Technologies Limited | Method for transmitting image data through RSDS transmission interfaces |
US20110025697A1 (en) * | 2009-07-28 | 2011-02-03 | Ying-Lieh Chen | Method for transmitting image data through rsds transmission interfaces |
EP2544145B1 (en) * | 2011-07-06 | 2018-09-12 | Brandenburgische Technische Universität Cottbus-Senftenberg | Method, arrangement, computer programm and computer-readable storage medium for scaling two-dimensional structures |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH02112392A (ja) * | 1988-10-20 | 1990-04-25 | Sony Corp | 映像信号処理装置 |
JPH02127884A (ja) * | 1988-11-08 | 1990-05-16 | Sony Corp | テレビジョン受像機 |
JPH02198286A (ja) * | 1988-10-13 | 1990-08-06 | Sony Corp | 映像信号受像機 |
JPH06217264A (ja) * | 1992-11-04 | 1994-08-05 | Philips Electron Nv | 画像信号変換回路 |
Family Cites Families (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2623040B1 (fr) * | 1987-11-09 | 1990-02-09 | France Etat | Procede et dispositif de traitement de signaux d'image a balayage de trame entrelace |
JPH02237280A (ja) * | 1989-03-10 | 1990-09-19 | Hitachi Ltd | 標準/高品位テレビジョン受信装置 |
US5461420A (en) * | 1992-09-18 | 1995-10-24 | Sony Corporation | Apparatus for coding and decoding a digital video signal derived from a motion picture film source |
KR100360206B1 (ko) * | 1992-12-10 | 2003-02-11 | 소니 가부시끼 가이샤 | 화상신호변환장치 |
JP3362463B2 (ja) | 1993-07-09 | 2003-01-07 | ソニー株式会社 | フレーム補間装置 |
JP4190576B2 (ja) * | 1994-08-31 | 2008-12-03 | ソニー株式会社 | 撮像信号処理装置及び撮像信号処理方法、並びに撮像装置 |
JP3794505B2 (ja) * | 1995-03-22 | 2006-07-05 | ソニー株式会社 | 信号変換装置及び信号変換方法 |
US5852470A (en) * | 1995-05-31 | 1998-12-22 | Sony Corporation | Signal converting apparatus and signal converting method |
JP3669530B2 (ja) * | 1995-06-30 | 2005-07-06 | ソニー株式会社 | 画像信号変換装置及び画像信号変換方法 |
US5946044A (en) * | 1995-06-30 | 1999-08-31 | Sony Corporation | Image signal converting method and image signal converting apparatus |
EP0843475B1 (en) * | 1996-05-30 | 2004-12-08 | Sony Corporation | Picture information converting apparatus and method |
AU719477B2 (en) * | 1996-12-11 | 2000-05-11 | Sony Corporation | Signal converting apparatus and method |
JP3787823B2 (ja) * | 1997-07-31 | 2006-06-21 | ソニー株式会社 | 画像処理装置および画像処理方法 |
JP4093621B2 (ja) * | 1997-12-25 | 2008-06-04 | ソニー株式会社 | 画像変換装置および画像変換方法、並びに学習装置および学習方法 |
JP4158232B2 (ja) * | 1998-07-23 | 2008-10-01 | ソニー株式会社 | 画像情報変換装置および画像表示装置 |
JP4147632B2 (ja) * | 1998-08-24 | 2008-09-10 | ソニー株式会社 | 画像情報変換装置、画像情報変換方法、およびテレビジョン受像機 |
JP4140091B2 (ja) * | 1998-09-14 | 2008-08-27 | ソニー株式会社 | 画像情報変換装置および画像情報変換方法 |
JP4135229B2 (ja) * | 1998-09-29 | 2008-08-20 | ソニー株式会社 | 映像信号の変換装置および変換方法、並びにそれを使用した画像表示装置およびテレビ受信機 |
WO2000021301A1 (en) * | 1998-10-05 | 2000-04-13 | Sony Corporation | Image transform device and method, learning device and method, and recording medium |
US6307560B1 (en) * | 1999-02-12 | 2001-10-23 | Sony Corporation | Classified adaptive spatio-temporal format conversion method and apparatus |
US6377307B1 (en) * | 1999-05-27 | 2002-04-23 | Pioneer Corporation | Line interpolation apparatus and video signal conversion device |
US20040128391A1 (en) * | 2002-12-31 | 2004-07-01 | Robert Patzer | Method and system for managing a validity period in association with a presence attribute |
-
2000
- 2000-04-21 US US09/720,348 patent/US6646684B1/en not_active Expired - Fee Related
- 2000-04-21 EP EP00917418A patent/EP1111918B1/en not_active Expired - Lifetime
- 2000-04-21 KR KR1020007014644A patent/KR100730499B1/ko not_active IP Right Cessation
- 2000-04-21 JP JP2000614655A patent/JP4470323B2/ja not_active Expired - Fee Related
- 2000-04-21 DE DE60041114T patent/DE60041114D1/de not_active Expired - Lifetime
- 2000-04-21 WO PCT/JP2000/002636 patent/WO2000065830A1/ja active IP Right Grant
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH02198286A (ja) * | 1988-10-13 | 1990-08-06 | Sony Corp | 映像信号受像機 |
JPH02112392A (ja) * | 1988-10-20 | 1990-04-25 | Sony Corp | 映像信号処理装置 |
JPH02127884A (ja) * | 1988-11-08 | 1990-05-16 | Sony Corp | テレビジョン受像機 |
JPH06217264A (ja) * | 1992-11-04 | 1994-08-05 | Philips Electron Nv | 画像信号変換回路 |
Non-Patent Citations (1)
Title |
---|
See also references of EP1111918A4 * |
Also Published As
Publication number | Publication date |
---|---|
EP1111918A1 (en) | 2001-06-27 |
US6646684B1 (en) | 2003-11-11 |
EP1111918A4 (en) | 2005-09-14 |
DE60041114D1 (de) | 2009-01-29 |
EP1111918B1 (en) | 2008-12-17 |
KR100730499B1 (ko) | 2007-06-22 |
KR20010034918A (ko) | 2001-04-25 |
JP4470323B2 (ja) | 2010-06-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4158232B2 (ja) | 画像情報変換装置および画像表示装置 | |
JP2533393B2 (ja) | Ntsc―hdコンバ―タ | |
KR100367432B1 (ko) | 영상 표시 장치 및 영상 표시 방법 | |
JP4131048B2 (ja) | 画像情報変換装置、変換方法および画像表示装置 | |
JP3596520B2 (ja) | 画像信号処理装置及び方法 | |
JP4470323B2 (ja) | 画像変換装置および方法 | |
WO1997001936A1 (en) | Method of generating high-resolution video | |
JP2005192230A (ja) | 表示装置 | |
JP4470324B2 (ja) | 画像信号変換装置および方法 | |
JP4078719B2 (ja) | 画像情報変換装置、変換方法および画像表示装置 | |
JP4847688B2 (ja) | インターレース−プログレッシブ変換器及びその変換方法 | |
JP2000092455A (ja) | 画像情報変換装置および画像情報変換方法 | |
EP1606954B1 (en) | Arrangement for generating a 3d video signal | |
JP4099555B2 (ja) | 画像情報変換装置 | |
JP2003198980A (ja) | 映像表示装置 | |
WO1997001929A1 (en) | Circuit for interpolating scan lines of a video signal and method of using same | |
JP4174891B2 (ja) | 画像情報変換装置および方法 | |
JP4154768B2 (ja) | 画像情報変換装置および画像表示装置 | |
JP4123587B2 (ja) | 画像情報処理装置および方法 | |
JP3800638B2 (ja) | 画像情報変換装置および方法 | |
JP4310847B2 (ja) | 画像情報変換装置および変換方法 | |
JP4186582B2 (ja) | 画像信号の処理装置および処理方法 | |
JP3480011B2 (ja) | 画像情報変換装置 | |
JP2007336239A (ja) | デジタル放送受信装置及びデジタル放送信号の保存・再生方法 | |
JPH0937175A (ja) | 画像表示装置及び画像表示方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): JP KR US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 09720348 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020007014644 Country of ref document: KR |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2000917418 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1020007014644 Country of ref document: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 2000917418 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1020007014644 Country of ref document: KR |