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
  • H04N7/0135—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes
  • H04N7/0137—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes dependent on presence/absence of motion, e.g. of motion zones
• • 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
  • H04N7/0117—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
  • H04N7/0122—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 the input and the output signals having different aspect ratios
• • 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
  • H04N7/0125—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level one of the standards being a high definition standard
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N9/00—Details of colour television systems
  • H04N9/64—Circuits for processing colour signals
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N5/00—Details of television systems
  • H04N5/14—Picture signal circuitry for video frequency region
  • H04N5/147—Scene change detection

Definitions

• the present invention relates to video line interpolation, field rate converting and de-interlacing methods and apparatus for converting an interlaced standard video image into a progressive HDTV video image.
• An HDTV Up Converter is an interesting device for broadcasting at HDTV resolution or format an existing standard interlaced video signal into an HDTV video signal.
• An HDTV Up Converter can be essentially decomposed in three main parts: a field rate converter, a de-interlacer and a picture resizer. The combination of this three modules provide at the output a 60.00 Hz field (interlaced) or frame (progressive) rate having a 16:9 aspect ratio video signal.
• the major part in an HDTV Up Converter is the de- interlacer unit which converts the incoming interlaced video signal into a progressive video output. This progressive output permits in turn an easy vertical resizing.
• the present invention focuses on the first two modules of the HDTV up converter, thus providing improved methods and devices for the field rate conversion and for the de- interlacing a video image.
• the median filter which is excellent for preserving the edges, can give some unnatural results.
• the interpolations are edge-based and purely spatial which result in similar defaults caused essentially by the edge detection technique.
• the staircase effect becomes noticeable for nearly horizontal and long edge in the picture.
• the present de-interlacer structure is similar to the one disclosed in US Patent
• the present invention provides an improved video frame frequency converter for converting the lower frequency of an interlaced video signal into a higher frequency of an intermediate video signal which is then de-interlaced by the second device of the invention, namely the de-interlacer.
• the apparatus comprises line interpolation means for producing at least one of temporally and spatially interpolated video signals using information from at least one field of the interlaced video input signal.
• the interpolated signals are suitable for a)- still part of picture, b)-moving horizontal edges and c)-various directions along edge.
• the apparatus comprises various detection means for the above-mentioned conditions.
• the edge direction detection means are robust for nearly horizontal edges in the present of noise or in high frequency picture part.
• the proposed method for edge direction detection is composed in two steps: a)-interpolation in all considered directions and b)-selection of the "best" direction among these interpolation results.
• the considered directions are chosen in function of the knowledge of human visual system.
• the apparatus comprises also a resizing means to convert part of 4:3 aspect ratio progressive picture to 16:9 HDTV format which is not presented in the following text since it is well known in the art.
• an apparatus for producing, for example, a 60.00 Hz field (frame) rate HDTV format video signal from a 59.94 Hz interlaced video input signal.
• the apparatus comprises picture deletion and insertion means for deleting appropriately a lower number of incoming frames and inserting a higher number of new frames in order to produce the exact field rate of 60.00 Hz.
• the two new pictures are preferably not repeated frame but rather interpolated from the incoming signal.
• the proposed "deleting one- inserting two" technique reduces substantially the motion discontinuity.
• the apparatus comprises also various detector means for determining subjectively favorable conditions in picture deletion/insertion. In a 50.00 Hz system, picture frame rate converter is not necessary.
• An object of an aspect in the present invention is to provide a video de- interlacer in which the degradation of diagonal or nearly horizontal edges, moving or fixed, can be reduced.
• An object of an aspect in the present invention is to provide a line interpolation apparatus and method in which the edge direction, nearly horizontal or not, is robust in the presence of noise or in high frequency picture parts.
• Another object of an aspect of the invention is to provide a line interpolation apparatus or method in which high resolution in still picture parts is fully preserved.
• an apparatus for producing an output HDTV format video component signal from an input interlaced video component signal is provided.
• the apparatus For 60Hz system, the apparatus comprises a 59.94Hz to 60.00Hz field rate converter, a line doubler and a format resizer. For 50Hz system, the apparatus comprises a line doubler and a format resizer.
• the 59.94 to 60.00Hz proposed converter comprises two major features for a)- frame insertion condition detection, b)- frame interpolation and frame insertion.
• the insertion condition detection in a time frame of 1000 incoming frames is the first happened event of a)-nearly-static picture, b)-scene change, c)-reduced motion activity picture and d)- end of time frame.
• the first three conditions are based on a proposed measure of motion activity indice, which is simply the mean of absolute frame difference value.
• the line doubling technique is a combination of three main interpolations: a)- temporal interpolation for picture still parts, b)-vertical interpolation for vertically moving horizontal edges and c)-steered spatio-temporal interpolation along edges.
• the third proposed interpolation could be divided further by two categories: a)-purely directional along well-detected edges and b)-vertical and directional for weakly detected edges. The last type of interpolation is a compromise for nearly horizontal weak edges.
• the associated detectors for the interpolation are a)-motion detection using four-field motion information, b)-vertical motion detection for horizontal edge and c)- edge detection.
• the proposed edge detection is extended for following nine edges 90o, 45o, 30o, 7o, 4o, -4o, -7o, -30o and -45o suitable with the human visual system.
• the edge detection is done in two steps a)- interpolate the image in all given directions and b)-select the direction with minimum variation in the result.
• Various consolidation schemes are proposed in order to get a robust decision even in the present of noise or in the high frequency picture parts. Compromise decision for weak and nearly horizontal edges is also proposed.
• a method and an apparatus for converting an incoming interlaced video signal from a lower field frequency, such as 59.94 Hz to a higher field frequency, such as 60.00 Hz, as needed by the HDTV video format is realized by adding supplementary video fields at each sequence of a predetermined number of fields, such as 1000 fields, for increasing the field frequency.
• a predetermined number of fields such as 1000 fields
• two fields (one frame) may be deleted from the sequence of fields, but it serves to create four other interpolated video fields which are added to the sequence of fields, thus increasing the number of fields by two fields (one frame).
• This process increases the field frequency from 59.984 Hz to 60.00 Hz.
• the same technique may be applied to a progressive video signal, and in this case the apparatus will use frames instead of fields for deleting and adding.
• the preferred embodiment of the present invention also relates to the precise moment when the field (or frame) inserting process is performed.
• the best moment for doing the process is when the video image is either still or moves very rapidly.
• a motion detector detects the motion indice which is continuously processed in order to detect the best moment for adding a field to the images.
• a count detector allows keeping track of the number of fields and commands the insertion of the additional field each 1000 fields, even if the best inserting conditions did not occur until reaching 1000 fields, so that the frequency conversion is performed constantly, in every sequence of 1000 fields.
• a video frame frequency converter for converting a standard video signal having a first frame frequency into an intermediate video signal having a higher second frame frequency
• said video frame frequency converter comprising: a converter input for receiving said standard video signal; a frame insertion detector means for analyzing said standard video signal and for detecting a best moment for adding at least one new frame in a sequence of existing frames of said standard video signal; means for generating an accelerated video signal from said standard video signal, said accelerated video signal having said higher second frame frequency; insertion means for inserting at least one frame into said sequence of existing frames of said accelerated video signal for increasing a number of frames of said sequence having a predetermined duration, said inserting means outputting said intermediate video signal having said higher frame frequency; and a converter output for providing said intermediate video signal.
• Another object of the invention is to provide a method for converting a standard video signal having a first frame frequency into an intermediate video signal having a higher second frame frequency, said method comprising the steps of: accelerating said standard video signal from said lower frame frequency to said higher frame frequency thus producing an accelerated video signal; analyzing said standard video signal for detecting a best moment for adding at least one frame into a sequence of existing frames, and for producing an insert frame control signal; and upon control of said insert frame control signal, adding said at least one frame in said sequence of existing frames, thus producing said intermediate video signal.
• Still another object of the invention is to provide a video frame frequency converter for converting a standard interlaced video signal having a first frame frequency into an intermediate interlaced video signal having a higher second frame frequency
• said video frame frequency converter comprising: a converter input for receiving said standard interlaced video signal; means for generating an accelerated interlaced video signal from said standard interlaced video signal, said accelerated video signal having said higher second frame frequency; frame interpolator means for creating at least two new interpolated fields by interpolation of existing adjacent fields of said standard interlaced video signal; and insertion means for inserting said at least two new interpolated fields into said sequence of existing frames and outputting said sequence of frames including said interpolated fields at said higher second frequency as said intermediate interlaced video signal.
• an improved edge direction detector to be used in a video de-interlacer for detecting at least one best direction from a set of pre-defined directions for producing a spatial direction control signal used for performing a spatio-temporal interpolation in said best direction
• said edge direction detector comprising: directional interpolator means for performing an interpolation for each one of said pre-defined directions using a past, a present and a future video field signal received at its input, said directional interpolator means outputting an interpolated signal comprising interpolated signals, for each pixel, for each of said pre-defined directions; and edge direction selector means for selecting said at least one best direction for interpolating using said interpolated signal received from said directional interpolator means.
• an up-converter apparatus for converting a standard interlaced video signal into a progressive HDTV video signal
• said up-converter apparatus comprising: a video frame frequency converter for converting said standard interlaced video signal having a first frame frequency into an intermediate interlaced video signal having a higher second frame frequency, said video frame frequency converter comprising a converter input for receiving said standard interlaced video signal; means for generating an accelerated interlaced video signal from said standard interlaced video signal, said accelerated video signal having said higher second frame frequency; frame interpolator means for creating at least two new interpolated fields by interpolation of existing adjacent fields of said standard interlaced video signal; insertion means for inserting said at least two new interpolated fields into said sequence of existing frames and outputting said sequence of frames including said interpolated fields at said higher second frequency as said intermediate interlaced video signal; and a frame insertion detector means for analyzing said standard video signal and for detecting a best moment for adding at least one new frame in said sequence
• Figure 1 is a general block diagram illustrating the main parts of an HDTV up converter, in which Figure la illustrates the preferred embodiment of the invention for a 60.00 Hz system and Figure lb refers to a 50.00 Hz system;
• Figure 2 is a detailed functional block diagram illustrating the preferred embodiment of the invention referring to the de-interlacer
• Figure 3 illustrates the nominal pixel positions corresponding to nine considered edge directions
• Figure 4 illustrates the pixel positions used for various line interpolations
• Figures 5 and 6 represent edge direction calculation for a series of directions corresponding to 90o, 45o, 30o, 7o, 4o and -45o, -30o, -7o, -4o;
• Figure 7 is the high frequency detector for direction 90o according to the preferred embodiment of the invention.
• Figure 8 is the high frequency detector for direction 45o and -45o according to the preferred embodiment of the invention
• Figure 9 is the high frequency detector for direction 30o and -30o according to the preferred embodiment of the invention
• Figure 10 is the high frequency detector for direction 7o and -7o according to the preferred embodiment of the invention.
• Figure 11 is the high frequency detector for direction 4o and -4o according to the preferred embodiment of the invention.
• Figure 12 is the edge direction selector according to the preferred embodiment of the invention.
• Figure 13 is the algorithm used for minimization in the edge direction selector according to the preferred embodiment of the invention.
• Figure 14 represents the edge binary filters in consolidations 1, 3, 4, 5;
• Figure 15 illustrates the edge binary filters for directions 45o, -45o, 30o, -30o in consolidation 2;
• Figure 16 illustrates the edge binary filters for 7o, -7o in consolidation 2
• Figure 17 illustrates the edge binary filters for 4o, -4o in consolidation 2;
• Figure 18 represents direction decision block diagram in a pseudo code format;
• FIG 19 is a block diagram of the proposed motion detector according to the preferred embodiment of the invention, also shown in Figures 2 and 21;
• Figure 20 is a block diagram of a proposed vertical motion detector according to the preferred embodiment of the invention
• Figure 21 is a general functional block diagram illustrating a line doubling apparatus or method for chrominance component
• Figure 22 is a general functional block diagram illustrating the preferred embodiment of the invention referring to the field frequency converter;
• Figure 23 is a functional block diagram illustrating the proposed detector for the frame inserting condition shown in Figure 22;
• Figure 24 is the movement indice calculator block diagram for the embodiment shown Figure 22;
• FIG 25 is a high level flowchart of the scene change detector according to the preferred embodiment of the invention also shown in Figure 22;
• Figure 26 illustrates fixed and adaptive threshold detectors respectively according to the preferred embodiment of the invention shown in Figure 21;
• Figure 27 represents the proposed technique for frame interpolation and frame insertion according to the preferred embodiment of the invention, also shown in Figure 21;
• Figure 28 represents the separable vertical temporal filters for frame interpolation according to the preferred embodiment of the invention.
• Fig. la illustrates the three main parts of an up converter 10.
• the first part of the system is a video frame frequency or field rate converter 12, which accepts, according to the preferred embodiment of the invention, a lower frequency video signal, such as a 59.94 Hz interlaced video signal input 14 and outputs a higher video frame frequency signal, such as a 60 Hz interlaced video signal 16.
• the output video signal output 16 is an intermediate interlaced video signal 16 which has exactly 60 video fields by second and thus complies with the HDTV field frequency standard.
• the intermediate video signal 16 then enters a de-interlacer 20 whose function is to provide a progressive video signal 22 having the same input picture aspect ratio of 4:3 as the signal 14.
• the progressive video signal 22 allows a resizer 24 to convert easily its input signal 22 into a progressive HDTV video signal 26 having an aspect ratio of 16:9.
• the resizer 24 is mainly composed of separable vertical and horizontal digital interpolation filters. Since the filtering technique is relatively well known in the art, the resizer 24 will not be discussed in detail in the present text.
• the progressive HDTV video output 26 can be further transformed, if necessary, into an interlaced HDTV signal 28 by deleting the appropriate lines in each image or picture. This line decimation or deletion is not shown in Fig 1.
• the present invention also apply to the European version of the HDTV standard, where the HDTV signal has a frequency of 50 Hz. Since the standard interlaced video signal 30 has the same frequency, in such a system the field frequency converter 12 is no longer needed. As illustrated in Fig. lb the interlaced video signal 30 is directly fed into the de-interlacer 20. In other words, according to the preferred embodiment of the invention, the de-interlacer 20 is functional for any field rate video input.
• Fig. 2 illustrates the proposed de-interlacer block diagram for interlaced digital luminance video input.
• the present invention provides an improved edge direction detector 44 which is part of the overall de-interlacer 20.
• the proposed system is an adaptive interpolator that combines the results from a pure temporal interpolation for picture still part, with a pure vertical interpolation for vertically moving horizontal edges, and with an edge-based steered spatio-temporal interpolation for the general case.
• the edge direction number is extended up to 9, namely 4o, 7o, 30o, 45o and 90o for positive directions and - o, -7o, -30o and -45o for negative directions, as illustrated in Fig 3.
• the proposed directions are chosen approximately in a logarithmic order according to the human visual system. In fact, the picture result will be more pleasant if nearly horizontal edges are carefully interpolated.
• the interlaced video input 005 is applied to two field delays 32 and 33 connected in series. These two field delays provide respectively two delayed video signals 34 and 35.
• the video signals 35, 34 and 30 represent respectively the past, the present and the future video fields.
• These signals are sent in a suitable manner to the three interpolators 38, 40 and 42 and to the three detectors 44, 46 and 48.
• the cited detectors control in turn the system adaptation in order to provide a final interpolated signal output.
• the first interpolator 38 namely the temporal interpolator 38, provide an average signal from the past frame signal 35 and the future frame signal 30.
• the temporal output 50 is given by the following expression:
• the vertical interpolator 40 accepts only as input the present field video signal 34.
• the vertical interpolator's outputs 40 also designated by VF, is given by the following equation:
• a and A' are values of adjacent pixels respectively corresponding to the preceding and to the following existing lines in the vertical direction of the pixel to be interpolated.
• F is value of existing pixel adjacent to pixel A in the vertical direction. F is also vertically adjacent to A as illustrated by Fig 3.
• SST+4o (4A8 + 4A'-8 + 2B0 + 2C0 - D16 - D'-16 - E16 - E'-16) / 8 (4)
• SST-7o (4A-4 + 4A'4 + 2B0 + 2C0 - D-8 - D'8 - E-8 - E'8) / 8(5)
• SST+7o (4A4 + 4A'-4 + 2B0 + 2C0 - D8 - D'-8 - E8 - E'-8) / 8 (6)
• SST-30o (4A-2 + 4A'2 + 2B0 + 2C0 - D-4 - D'4 - E-4 - E'4) / 8 (7)
• SST+30o (4A2 + 4A'-2 + 2B0 + 2C0 - D4 - D'-4 - E4 - E'-4) / 8 (8)
• Equation 1 is chosen in order to reduce the possible additive noise by a factor of 3dB.
• Equation 2 illustrates the simplest four taps half-band filter and equations 3 to 10 are edge directed versions of a vertical temporal half-band filter, defined by the following equation:
• VT (4A0 + 4A'0 + 2B + 2C - DO - DO - E0 - E'O) / 8 (16)
• the SST90 which is the expression of a vertical interpolation, is also a vertical temporal filter similar to the one defined in Eq. 16.
• the vertical bandwidth of the filter in Eq. 11 is larger than the one defined in Eq. 16, when temporal frequency is nearly zero. This feature has been selected because the human visual system is more sensitive to still parts in a picture.
• the filter described in Eq. 11 is different from the VT filter used in US-Pat- Application # 08/916960, mainly because it has a better vertical bandwidth and a shaper transition roll-off.
• the two interpolated signals VT and SST are fed into a selector 60 which is controlled by a binary signal 62 provided from the vertical motion detector 46.
• the selector 60 When the control binary signal 62 is "ON”, the selector 60 outputs an SF signal 64 which is chosen to be the vertical inte ⁇ olator VF output 56. Otherwise, when the control binary signal 62 is "OFF", the multiplexer 60 selects the steered spatio-temporal inte ⁇ olator output 58.
• the selector SF output 65 and the temporal inte ⁇ olator TF output 50 are combined in a temporal adapter 66 in order to provide a final inte ⁇ olated video signal 68, for non existing lines of interlaced video input signal.
• the temporal adapter 66 is controlled by the motion indicative value 70, delivered by the temporal motion detection 48.
• the inte ⁇ olated video lines signal 68 and the existing video lines signal 72 are combined by the multiplexer 74. in order to generate the progressive luminance signal 22.
• edge direction detector 44 Associated with the three above-mentioned inte ⁇ olation techniques, are the edge direction detector 44, the vertical motion detector 46 and the temporal motion detector 48.
• the pu ⁇ ose of the temporal motion detector 48 is to locate rapidly moving or approximately still parts in an image. For doing so, it uses as input the low-pass filtered video signals 76 and 77 instead of their original ones 30 and 35 for possibly noisy signals.
• the pu ⁇ ose of the vertical motion detector 46 is to locate moving horizontal lines in a video image sequence.
• the edge direction detector 44 has two functions: the first one is to choose the best direction among the nine (9) possible directions for performing a steered inte ⁇ olation. Its second function is to compute a compromise inte ⁇ olation for insufficiently reliable nearly horizontal edges that are detected.
• the decision process is performed in two steps: first, the image is inte ⁇ olated in all possible directions and the direction having the minimum variation is selected.
• the nine directional inte ⁇ olators 80 receive as inputs the three video signals 30, 34 and 35 from the past, the present and the future fields. The inte ⁇ olations are described by the nine Eqs.
• Figs. 5 and 6 illustrates a detailed view of the edge direction calculators for the directions 90o, 45o, 30o, 7o, 4o, -45o, -30o, -7o, -4o.
• each inte ⁇ olated input signal for example 82a, is sent into a horizontal low-pass filter 90 to remove eventually noise and strengthen horizontal edge.
• the impulse response of this linear-phase filter 90 is (1, 3, 4, 3, 1).
• the filter outputs 92, 94, 96, 98, 100 and 102, 104, 106 and 108 are then applied individually to their respective directional variation calculator, numbered 110 through 126.
• each calculator is a directional high-pass filter whose impulse response is given in Figs 5 and 6.
• the calculator outputs 110 though 126 are sent into absolute value devices, numbered 130 through 146, in order to convert the initial values into magnitudes of variation in the possible nine directions.
• the absolute value devices outputs are sent into their respective low-pass filters 151 through 166 to smooth out any eventual noise.
• its low-pass filter may only be a vertical filter having an overall gain two times lower than the one of the other directions.
• the low-pass filters for the other directions may be identical and may comprise a vertical filter and a horizontal filter. The impulse responses of these filters are shown in greater detail in Figs 5 and 6.
• the directional inte ⁇ olator 80 output 82[a-i], is sent as previously mentioned into the horizontal and vertical high frequency detectors 86.
• the pu ⁇ ose of the high frequency detectors is to locate high horizontal or vertical texture regions of the image that could introduce errors in the edge direction estimation process.
• the detectors 86 may need to be different for each considered direction.
• Fig. 7 illustrates the high frequency detector 86a for vertical direction according to the preferred embodiment of the invention.
• the corresponding inte ⁇ olated signal 82a is applied to the input of the detector 86, which detects the high frequency intensity by applying a Laplacian 170, followed by and an absolute value device 172.
• Its output 174 represents the magnitude of the high frequency signal and is applied to a detector 176, which is simply a level comparator.
• the comparator output is a binary signal 178 which is equal to 1 if the input signal 174 is greater than a threshold value 180. If not, the binary signal 178 is equal to 0.
• the threshold value 180 is set to be 40, according to the preferred embodiment of the invention.
• the binary output signal 178 is sent into a consolidating device, which can link together some isolated detections in a moving window 3x3. Details of the mentioned consolidating device are provided in same Fig. 7: it may be comprise an appropriate delay 182, a filter 184, a comparator 186, and an OR gate 188.
• the gate output signal 190a represents a binary map of the high frequency region for the direction 90o.
• Fig. 8 illustrates the high frequency detectors 86 for the directions 45o and - 45o.
• a pixel is said to be in a high frequency zones if the magnitudes of the horizontal or vertical high frequency components in the input signal 82b or 82f exceed some threshold values, with the exception of a specific high frequency pattern corresponding to the considered direction.
• the horizontal and vertical high frequency components are detected respectively by the high-pass filters 192 and 194.
• the mask 196 defines the specific pattern for the 45o direction.
• the high-pass filters 198 and 200 do the same function while the mask 202 defines the -45o pattern.
• the impulse responses of these filters as well as details of the detection process are given in Fig. 8.
• the detector's binary signal outputs are the signals 190b and 190f, respectively for directions 45o and -45 o.
• Fig. 9 illustrates the high frequency detectors 86 for the directions 30o and -
• the two detectors 86 are similar to those presented in Fig. 8. The only difference resides in the specific high frequency pattern. For the direction 30o, this is provided by the filter 208, while for the direction -30o, it is provided by the filter 210.
• the detector's binary signal outputs are 190c and 190g respectively for the directions 30o and -30o.
• Figs. 10 and 11 illustrate four identical detectors 86 for the four nearly horizontal directions 7o, -7o, 4o, and -4o.
• a horizontal filter 212 detects the horizontal high frequency component.
• a vertical filter 214 does the same job for the vertical high frequency component.
• a vertical mask 216 may be used for the specific pattern. Other detection details are also given in these Figures.
• the detector 86 outputs the signals 190d, 190h, 190e and 190i respectively for the directions 7o, -7o, 4o and -4o.
• the edge direction calculator 84 outputs 220[a-i] and the high frequency detector outputs 190[a-i] are sent together into the edge direction selector 222 better shown in Fig. 12.
• the nine edge direction variations 220 are sent to a minimum selector 224 which outputs a first signal 226 comprising data related to a minimum value and its corresponding direction, and a second signal 228 having corresponding data. If two or more equal minimum values are detected, the device 224 selects only one direction according to the following priority: 90o, 45o, - 45o, 30o, -30o, 7o, -7o, 4o and -4o.
• the first signal 226, the second signal 228, and the 90o variation signal 220a are applied to a logic device 230, whose pseudo-code is better shown in Fig. 13.
• the logic device 230 will select only the minimum (the first minimum) direction or the vertical direction. The second minimum is used for a consistent comparison pu ⁇ ose.
• the logic device 230 provides nine binary outputs 232a and 234 through 248 respectively for the nine directions 90o, 45o, 30o, 7o, 4o, -45o, -30o, -7o and -4o. The output of the selected direction is set to "1" while the others are set to "0".
• the selected direction can be reset to "0" if the considered pixel is located in an unreliable high frequency region detected by the presence of the high frequency binary signals 190.
• each of the eight selector outputs is thus validated by a different AND gate 250a through 250h, with the negation of its respective detected high frequency binary signal.
• the nine resulting outputs are 232a through 232i, each representing one of the nine possible directions 90o, 45o, 30o, 7o, 4o, -45o, -30o, -7o and -4o.
• the edge direction selector output signals 232[a-i] are spiky and comprise many inconsistently isolated directions or discontinuities along a main edge. Consequently, it may be necessary to reinforce the detection results before taking a final decision.
• the outputs 232[a-i] are sent into the edge binary filters 127, as shown in Fig. 2, in order to consolidate the decision to be taken.
• the binary signals 232[b-i] may further be submitted to four (4) or five (5) consecutive phases of consolidation as better shown in Figs. 14 and 15. These five phases can be described successively as a horizontal, a directional, a vertical, another horizontal and finally a logical vertical consolidation.
• each one may comprise eight (8) binary filters followed by level decision devices working in parallel. Each filter and its associated decision device may be used for one direction only.
• the filter masks and the level detectors are given in Figs 14 and 15.
• the impulse response of the first consolidation filter 260 is (1,1,1,1,1) wherein the central coefficient corresponds to the current pixel position.
• the threshold 262 in the detector 264 is set equal to 2.
• the consolidation 2, also illustrated in Fig. 15 is directional and specific to one considered direction.
• the consolidations 3 and 4 shown in Fig. 14 have similar structure to that of the first consolidation.
• the consolidation 5 which is employed for nearly horizontal directions, is composed of four logical filters for four considered directions. Each filter is a linear vertical filter followed by a logical device running the code described by the pseudo code given in Fig. 14, and provides two binary outputs 266 and 268.
• the outputs 268[d, e, h, i] corresponding to the above-mentioned directions and called "Mix”, may represent some composite condition in the output image line inte ⁇ olation.
• the final direction decision output 54 represents the chosen inte ⁇ olation direction with or without mixed condition and is sent both to the steered spatio-temporal inte ⁇ olator 42 and to the vertical motion decision device 272, as illustrated in Fig. 2.
• the inte ⁇ olator 42 For a mixed condition, the inte ⁇ olator 42 combines the directional and the vertical inte ⁇ olations into a mean value as described by Eqs.. 12-15. Otherwise, the inte ⁇ olation is strictly directional.
• the vertical motion device 272 only the direction information carried by the signal 54 may be considered.
• Fig. 21 which illustrates the adaptive line doubling technique for video image chrominance component.
• Applicants have found that an adaptation based on moving or still parts in a picture is good enough even for nearly horizontal edges.
• the structure remains unchanged comparatively to the one proposed in the above- mentioned US Patent Application.
• the only difference resides in the fixed VT inte ⁇ olation filter 280, which is now described by Eq. 11.
• Fig. 22 illustrates the preferred embodiment of the invention related to the frame frequency converter or to the field rate converter 12. Even if the following paragraphs describe mainly a frame frequency converter for an standard interlaced video signal, it is to be noted that the same technique may be used for converting the frame frequency of a progressive video signal, still from a lower frame frequency to a higher frame frequency. This particular feature is rarely needed, but is still useful for conversion of standard progressive video signals into HDTV progressive video signals which have higher frame frequency.
• the frame frequency converter 12 receives the standard interlaced video signal 14 having a field frequency of 59.9400599402 fields/s.
• This standard video signal 14 is fed simultaneously into the clock and field synchronization generator, into the buffer memory 302, into the frame counter 304 and into the frame insertion detector 2107.
• the buffer memory means 302 are means that are used for generating an accelerated video signal 308 and may be a FIFO device, which reads the digital video input signal 14 using a 59.94Hz field synchronization control signal 306, and delivers at its output an accelerated video signal 308 having a field frequency of 60Hz, using the 60Hz field synchronization control signal 307.
• the accelerating means also called the buffer memory means 302 may also accept an insert frame control signal 310 for freezing a frame of video input in order to equilibrate the output video rate.
• the clock and field synchronization generator 300 provides the control signals 306 and 307 while the frame insertion detector 312 delivers the control signal 310.
• the frame counter 304 receives the video signal input 14 and counts from 0 to
• the counter provides a time frame or window in which one new video picture should be inserted in order to get a total of 1001 pictures for each incoming sequence of 1000 pictures.
• This ratio 1001/1000 is necessary for providing a 60Hz video output starting from an initial frequency of 59.94Hz.
• the functional block diagram illustrated in Fig. 22 shows the proposed frame insertion detector 15 for the case of an interlaced video signal.
• the pu ⁇ ose of the detector is to determine the right moment in a given sequence of 1000 consecutive existing frames to insert a frame.
• This frame may be duplicate from an existing frame or, preferably, may be a new inte ⁇ olated frame from adjacent existing frames
• the proposed detector 15 may examine the sequence of frames in order to detect the following situations: a)- static or nearly static picture sequence, b)- sudden scene change, c)- reduced motion activity and d)- end of the time frame.
• the static situation a) is obvious, since the newly created picture is easy to be inte ⁇ olated.
• the scene change situation b) is also understandable, since in such case, picture inte ⁇ olation artifacts become not evident for the human visual system.
• the situation c) is a compromise technique in a dynamic picture sequence: an insertion will be made if motion activity is reduced below an adaptive varying threshold.
• the situation d) is self-explanatory and happens when no other situation has occurred until the counter reaches 1000 frames.
• the standard interlaced video signal 14 is fed into a motion indice calculator 320, illustrated in details in Fig. 24, which evaluates the average of absolute difference between the frames.
• This value herein called the motion activity indice 322
• the fixed thresholding 324 is simply a level detector, as better shown in Fig. 26, which is used for static or nearly static picture detection. It provides an FT output 330 which may be one of the signals meaning that the condition for possible picture insertion has arrived.
• the scene change detector 326 provides also a binary output 332 when the motion indice difference between two successive frames is larger than a given threshold.
• the corresponding scene change detector block diagram is illustrated in greater details in Fig. 25.
• the adaptive thresholding 328 may comprise a first order low-pass filter 334 providing an output 336.
• the weighted value 338 changed by a possible factor of 0.9 coming from the low-pass filter output 336 is used as an adaptive varying threshold for the detector 340. If the instantaneous frame motion activity indice is smaller than the threshold signal 338, the detector AT output will be "ON" giving the signal for a possible picture insertion moment.
• the multiplexer 334' s function is to quickly change the varying threshold when a sudden scene change is detected.
• the frame counter detector 350 also provides a Count - 0 signal, 354, for the beginning of a time interval.
• This signal is sent to a logic device 356.
• the four binary signals, 332, 330, 342 and 352 are fed into an OR gate 358 in order to provide an output signal 360 designating a possible picture insertion signal.
• This signal 360 is also sent into the logic device 356 in order to further provide the insert frame control signal 310, which will have a value "ON" only once during the 1000 frames time window.
• the frame control signal 310 is transmitted into buffer memory 302 for freezing condition and also to the multiplexer 364 for selecting the new inte ⁇ olated picture 366 from frame inte ⁇ olator 368.
• Fig. 27 illustrates the proposed technique for the frame inte ⁇ olator module 368 and for the insertion means or frame insertion module 364.
• the proposed frame insertion technique is performed according to the preferred embodiment of the invention by substituting one incoming frame by two new and inte ⁇ olated frames (substituting two incoming fields with four new and inte ⁇ olated fields, for the interlaced case), when the inserting condition is detected.
• Fig. 27 also illustrates successively in time the various field positions.
• the present frame 400 composed of the two fields C and D is to be deleted.
• the two new frames composed are to be inserted.
• the time distance between the previous fields B and the inte ⁇ olated field P may be, according to the best mode of the invention, 3/5 of the incoming field interval.
• the normalized distance between P and C is thus 3/10.
• the proposed inte ⁇ olation technique for the field P is based only on its two nearest existing fields, B and C. Similarly, it should be the fields C and D that are used for the calculation of the inte ⁇ olated field Q, and so on.
• New field inte ⁇ olation may be performed in two separate steps: a) the vertical inte ⁇ olation for the missing lines in the existent fields and b) the temporal inte ⁇ olation for the new fields to be inserted.
• Fig. 28 illustrates these separate inte ⁇ olation filters. The reader is invited to take note that the pixel and the line notations are described in Fig. 27.
• the vertical inte ⁇ olation for the missing lines is provided from a half-band filter whose impulse response is described by the following: (-8,0,40,64,40,0,-8)/64.
• the temporal filter for new field inte ⁇ olation is a two taps filter in which the coefficients 3 ⁇ and V* are used and represent practical fixed-point values, approximately equal to the 3/5 and the 3/10 above mentioned normalized field distances.
• the coefficient % is associated with the nearest existent field for the considered inte ⁇ olated field.
• the coefficient is associated with the other existent field employed for the inte ⁇ olation.
• P2 may be described by the following expression:
• the inte ⁇ olated video pictures 366 may be selected by the insert frame control signal 310 and by the multiplexer 364 in order to provide the output, a 60Hz interlaced video signal 16.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Computer Graphics (AREA)
• Television Systems (AREA)

Priority Applications (1)

Applications Claiming Priority (5)

Related Child Applications (1)

Publications (1)

Family

ID=25683154

Family Applications (1)

Country Status (5)

Families Citing this family (15)

Family Cites Families (8)

• 1999
  • 1999-04-01 EP EP99911555A patent/EP1084577A1/de not_active Withdrawn
  • 1999-04-01 JP JP2000542915A patent/JP2002510946A/ja active Pending
  • 1999-04-01 WO PCT/CA1999/000286 patent/WO1999052281A2/en not_active Application Discontinuation
  • 1999-04-01 DE DE69928134T patent/DE69928134D1/de not_active Expired - Lifetime
  • 1999-04-01 AU AU30219/99A patent/AU3021999A/en not_active Abandoned

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events