FI94007B - Connection for peaking of a high resolution television picture - Google Patents

Description

5 94007

Switching to emphasize the sharpness of a high-definition television picture - Koppling för toppframhävning av en högupplösningste-levisionsbild

The invention relates to a high definition television receiver in which a luminance signal produced by a receiver decoder is processed to emphasize its sharpness.

10 HDTV in the European High Definition Television System

The (High Definition Television) studio standard has an aspect ratio of 16: 9 and comprises 1250 lines displayed at an interleaving ratio of 2: 1 (or later 1: 1) and a field frequency of 50 Hz. The picture is encoded when transmitted to a MAC-15-suitable HDMAC signal by compressing the luminance signal into its fourth part and the chrominance signal into its eighth part in a bandwidth reduction encoder BRE (Bandwidth Reduction Encoder). receiving. The line number of the transmission signal is thus half of the original, i.e. 625 lines, and the interleaving ratio is 2: 1. In view of the receiver, the so-called A DATV signal that transmits information about the motion content of an image.

25 i Compression is based on controlled subsampling controlled by motion information. This means that the less motion there is in the image, the higher the spatial resolution is used, and when the motion content of the image is large, the temporal resolution is added at the expense of the spatial.

*

In the above-mentioned bandwidth reduction encoder BRE, the image to be transmitted is divided into blocks of 16 * 16 pixels, which are derived from one of the three image processing branches based on their motion content. Each signal applied to a branch has its own sampling pattern. If there is no movement, the signal is controlled. 80 ms to the branch, where a subsampling pattern (decimation) is taken from each image field 2 94007 so that half of the pixels of the final image are transmitted. Samples are taken from each line and the decimated Teuva forms a quincunx sampling pattern. In this way, half of the compression is obtained and the other half is obtained by increasing the transmission time of one image from 40 ms to 80 ms. The name of the branch comes from this. If there are slow-moving objects in a part of the image, the video signal to be compressed is directed to a 40 ms branch, where the field transmission time is increased from 20 ms to 40 ms and only every other 10 quincunx-sampled fields are transmitted, i.e. every other sample is taken from every other field . If there are fast-moving parts in the image, the video signal to be compressed is directed to a 20 ms branch, where the bandwidth is reduced by a quarter so that every fourth sample of each original image-15 field is transmitted, i.e. there are samples from each line of the original image. Each branch of the BRE includes its own low-pass filter and a so-called "line shuffling" function, in which the format of a 1250-line image (line duration 32 / is) is converted to a MAC-compatible 625 line (line duration 64 μβ) format. In the 40 ms branch, this function is included in the subsampling itself. According to the information provided by the motion detector, the output of only one branch at a time is controlled as the output of the BRE.

25 The receiver's HDMAC decoder returns the compressed: picture 625/2: 1/50 Hz to the original HDTV format, ie 1250/2: 1/50 Hz. The decoder can be called a Bandwidth Restoration Decoder (BRD) and is analogous to the BRE described above. The BRD decoder 30 will now be described in more detail with reference to Figure 1, and the description relates to the luminance signal. The received sample frequency is 13.5 MHz and the output frequency of the BRD is 54 MHz. The encoder comprises a "line deshuffler" block 1, 3-6 field memories 2 connected e.g. to a cascade, a subsampling pattern reset 3 35 and three interpolation branches 4, 5 and 6. Each interpolation branch produces a complete HDTV field and the output of the selected branch is switched by switch C BRD output signal. The coupling

II

3 The 94007 output position is controlled by the motion information contained in the received DATV signal.

To generate the original line frequency, the deshuffler block 5 1 always generates two lines from one incoming line so that each new line is formed from every other sample of the incoming line. After the deshuffling operation, the luminance signal is delayed in the field memories 2. The field signals are then processed in the sub-sample pattern return 3 SSPC (Sub-Sample Pattern Converter), which reconstructs the original sub-sample patterns using the line memories. From the converter 3, the subsampling patterns are passed to branches 4, 5 and 6, where the original HDTV fields are formed by interpolation. In the 40 ms branch 6, the interpolation motion com-15 is compensated and utilizes the motion vectors reported in the DATV signal. Controlled by the DATV information, switch C connects one of the branches at a time to the output of the BRD, thus obtaining the original HDTV picture in the format 1250/2: 1/50 Hz.

20 However, when this image is displayed on the image surface, it appears soft. The reason for this is that due to the spatial subsampling at the transmitting end and the anti-crease filters of the HDMAC encoder, the transmitted and thus received HDMAC image did not contain as many frequency components as the original high-definition studio image. This-; the size of the hair components depends on the motion content of the image area. The mode representing the stationary image area, the 80 ms mode, contains the highest spatial plane frequency components, so the change bands at the edges are narrowest.

30 In the 40 ms slow motion mode, there are fewer spatial plane frequency components in the less than 80 ms mode, so the edge change bands are wider than in the 80 ms mode. The change bands are the widest in the fast-moving 20 ms mode because the spatial plane maximum-35 frequency components are the smallest in this mode.

The received image is further softened if 4 94007 FMH (FIR-Media Hybrid-filter) interpolators are used as interpolators for the signal processing branches of the receiver. They are median filters whose inputs have an FIR filter output. These interpolators are well suited for decoders in HDTV receivers, as the FMH interpolator does not bring individual pixels to the edges and narrow lines, the gray level of which differs from the environment and does not make the so-called diagonal edges. chess squares. Therefore, the output of the FMH interpolator is statistically close to the original image. Due to the softening effect of the interpolator, the out-10 input can be emphasized without interpolation errors becoming visible and disturbing.

As is known, the human eye, when watching a television image, creates an impression of sharpness mainly on the basis of luminance information pe-15. As a result, sharpening the edges of the luminance information results in a sharper image for the viewer and creates the impression that the resolution of the image increases. Sharpening the edges means that at the edge points in the image where the luminance changes greatly, 20 changes are sharpened further. Such points are e.g.

the outline of the subject when the subject and the background are different in brightness, the edges of the texts, etc. Such an emphasis on the luminance differences of the edges is commonly used in the English term "peaking", which is referred to in the text below as "peaking". Peak: The difficulty with using an HDMAC receiver is that the output of the HDMAC encoder is one of the outputs of three signal processing branches. As stated above, the maximum frequency components of the spatial plane of the image produced by each branch are different and thus also the change bands of the edges are different. In this case, applying the same peaking method to the output of all branches results in blocks with different resolutions being distinguished in the displayed image, and the image quality deteriorates.

The softening of the high-definition picture described above can be avoided by the spike circuit according to the invention, by which the luminance signal received from the HDMAC decoder is processed by:

(I

35 5 94007 that the sharpness effect of the displayed image is emphasized. The spike connection is characterized by what is stated in the characterizing part of claim 1.

According to the invention, the sharpness effect of the luminance image is emphasized by emphasizing the mid-frequency components of the image, the image of which appears sharper due to the highlighted edges. Highlighting the mid-frequency components is an effective way to sharpen the image, as most of the edge energy in the dra-10 cheese plane is concentrated on the mid-frequency components. The input of the silicon circuit is the output signal of the HDMAC encoder, and the circuit operates differently depending on which of the three interpolation branches of the encoder is currently connected to the output signal of the encoder. The decision information of the operating mode of the circuit is preferably obtained from the DATV information, which, as is known, also determines the interpolation branch to be connected to the output of the encoder. The purpose of the decision-controlled peaking is to produce a uniform output regardless of the current interpolation branch connected to the encoder output, so that the blocks 20 of the different branches do not differ from each other in the displayed image.

The peaking circuit preferably performs both vertical enhancement and horizontal enhancement on the incoming signal. Of course, only horizontal or vertical highlighting can be performed. Both highlights are accomplished by adding a low-level peak-limited, band-pass filtered, and low-edge limited signal to the original signal. In each highlight, the bandpass filter to be used is selected so that the filtering 30 depends on which encoder interpolation branch the input signal of the spike circuit is from. The peak constraint used in both enhancements comprises, first, preventing the accentuation of dark areas of the image and, second, preventing low-amplitude channel noise from being accentuated. Because the ih-35 miss eye is sensitive to large gray level changes in dark areas of the image, the spike in these areas would appear disturbingly as channel noise. This is prevented by subtracting a certain fixed value from the input signal. After the bandpass filter 6,9007, undue noise gain is limited by eliminating the gain with small values at the filter output. This is based on the fact that the channel noise 1a has a smaller amplitude compared to the image information, so that the 5 low-amplitude signals, which are probably noise, are not added at all as an enhancement signal. Finally, the highlight signals are added by the desired amount weighted to the original signal. The weight is adjustable by the TV viewer.

10

The invention will now be described more clearly with reference to the accompanying schematic figures, in which Figure 1 shows a known HDMAC decoder to which a peaking block according to the invention has been added, Figure 2 is a possible block diagram of a peaking circuit, Figure 3 shows a curve in small gray planes, Figure 4 illustrates Fig. 5 shows the basic switching field of a horizontal FIR filter, Fig. 6 illustrates the basic switching field of a vertical FIR filter, Fig. 7 is a table of some possible filter coefficients for FIR filters.

25: As shown in Figure 1, the switch C is known to switch the output of the encoder to one of the interpolation branches 4, 5 or 6. The position of the switch C is determined by the branch decision information obtained from the DATV signal processing block 8. The luminance signal produced by the encoder is applied to a peaking circuit 7 according to the invention, to which the branch decision information from block 8 and the peaking intensity setting information W are also fed. This information can be set by the television viewer. In addition, a DZ value calculated in block 9 can be applied to the peaking circuit 7. The calculation is performed on the basis of the signal-to-noise ratio measured from the test line of the received HDMAC signal. The DZ value can also be user-adjustable, in which case block 9 is not required at all. The DZ-ar prevents the undue emphasis of noise from occurring.

7 94007

In the peaking circuit 7, a vertical and horizontal emphasis is performed on the luminance signal, taking into account the different frequency responses of the interpolation branches 4, 5, 6. The output signal of the silicon circuit is a luminance signal in which the center frequency components are highlighted, so that the image looks sharp to the viewer.

Figure 2 is a simplified block diagram of a spike circuit 7 according to an embodiment. The circuit can be divided into 10 consecutive similar parts, parts A and B.

Part A performs vertical frequency enhancement on the luminance signal and Part B performs horizontal frequency enhancement. In principle, both parts of the emphasis work in the same way, so first the operation of Part A will be described in more detail, and then the operation of Part B only in so far as it differs from the operation of Part A.

The luminance signal obtained from the decoder, which is thus the lu-20 luminance signal produced by one of the three signal processing branches 4, 5 or 6, is fed to an adder 25, where vertical frequency emphasis is added to it. This enhancement of the center frequency components is formed in a parallel branch comprising an adder 21, a bandpass filter 22, a noise enhancement blocking circuit 23 and an adjustable attenuator 24.

The organ 21 serves as the first peak constraint as follows: Since the human eye is sensitive to large changes in the gray level in dark image areas, dark areas should not be emphasized. This is prevented by subtracting a certain value K from the luminance signal to be highlighted in the member 21. The purpose of this operation is shown in Fig. 3, in which Fig. 3A shows the luminance signal entering the peaking circuit as a function of time. The higher the instantaneous value of the signal, the lighter the image area it represents. With an eight-bit signal, the maximum grayscale values are between 1 ... 254. 35 Thus, at interval ΔΤ, the signal has a small grayscale value, so it represents a dark-toned image. In order not to emphasize the signal in this area, a certain value K is subtracted from it in the member 21, whereby after the member the signal is as shown in Fig. 3B. Negative values resulting from the reduction have been replaced by zeros. This signal is the input signal of the filter 22. The value K can be chosen freely, a preferred value would be 100.

5

The filter 22 can operate in three different modes depending on the state of the control signal input to the filter. Because each branch of the encoder has a different subsampling pattern, the resolution of the image areas they produce and the other lanes of the edges are different. Therefore, the filter is such that it has its own unique mode for each signal processing branch of the encoder in order to obtain the best possible highlighting results. In principle, when in a certain mode, the filter is a bandpass filter at the center frequency of the encoder branch 15 corresponding to this mode. Analogous to the signal processing branches, the filter modes can be named 80 ms mode, 40 ms odd field mode, 40 ms even field mode, and 20 ms mode. When the DATV signal processing block 8 directs, for example, an 80 ms signal-20 link processing branch 5 (Fig. 1) to the output of the encoder, the decision signal from said block directs the filter to operate in its 80 ms mode. The same happens for the other branches, so the filter always operates at the center frequency of the current output signal of the encoder.

25

The filter mode simply means that the filter coefficients of the FIR filter are selected differently in each mode. This is illustrated in Figure 5, which shows the basic circuit of a horizontal FIR filter. The filter comprises a set of 30, in this case four one-pixel delays pd1, pd2, pd3 and pd4, five weighting elements al ... a5, each of which weights the derived sample value so that the incoming luminance sample is weighted by kl, the previous luminance sample by k2, etc. The weighted luminance-35 samples are summed in an adder 51, the output of which provides a filter output signal. If desired, it can be scaled. As can be seen, the pixel delays form a filter window with five consecutive snow lines of the same line.

II

9 94007 nance samples. The filter coefficients are called a set of numbers [kl, k2, k3, k4, k5]. The coefficients can be e.g. [-1, 0, 2, 0, -1]. Changing these coefficients changes the property of the filter, so each mode corresponds to certain multipliers of 5 met. The odds are changed programmatically.

Figure 6 shows a known basic circuit of a vertical FIR filter. It is basically analogous to the connection of Figure 5, but the delays of one pixel have been replaced by the delay of the yh-10 den line. The arrangement allows the filter window to be one sample wide in the horizontal direction but to comprise five samples in the vertical direction. By increasing the number of line memories, the height of the window can be increased. In the vertical direction, each sample is weighted with its own weighting factor kl ... k5, the weighted samples are added together and the summed vertically-filtered signal is obtained from the output. As with the horizontal filter, the filter coefficients are called the number sequence [kl, k2, k3, k4, k5]. Changing these coefficients changes the filter property of the vertical-20 auxiliary filter, so that each mode corresponds to certain coefficients. The odds are changed programmatically.

The output of the filter can be limited without visibly affecting the peaking result. Without limitation, the output of the filter is between [-438, 438] when the input is between [16, 235]. This would require a 10-bit output. If the filter output is limited to [-128, 127], an 8-bit output will suffice, simplifying the system.

30 This limitation cannot be seen in the peaking result, as most of the filter output values are concentrated within the limit range.

The noise enhancement blocking circuit 23 prevents undue noise enhancement 35 from occurring. The channel noise has a small amplitude compared to the image information, so the output signal of the blocking circuit is zero if the amplitude (absolute value) of the signal coming from the filter is less than the limit value, i.e. assuming 94007 10 the incoming signal as noise. The threshold is set to fixed or can be set to adaptive by automatically calculating the signal-to-noise ratio from the test line of the MAC frame and determining the threshold DZ from this ratio. If the 5 limit value DZ is fixed, it can be made as a factory setting or the user can set it as desired with the remote control, for example. Figure 4 shows in curve form the output of the highlight blocking circuit as a function of the input. When the input is between -DZ / 2, DZ / 2, the blocking circuit sets its output to zero.

10 The shaded square around the origin in the figure is thus the so-called death zone, within which the input values do not appear in the output. In the figure, the input value and thus the output value is limited to L if the filter output limit to 8-bit is used above. The restriction is not necessary. If the DZ interval is determined from the signal-to-noise ratio of the received signal, as the S / N ratio increases, the DZ interval can be reduced and thus the effect of emphasis on lower signal levels can be improved.

20 Reference is further made to Figure 2. The signal from the anti-jamming circuit is applied to a peaking weight control member 24 which weights the highlighting signal by a desired value W before summing it to the luminance signal from the encoder in adder 25. The control thus determines how much mid-frequency peaking signal is attenuated image, i.e. how much the luminance signal is emphasized. The weighting value is user-adjustable, allowing the TV viewer to adjust the sharpest picture according to their own subjective preferences. The default value is a weighting value of 1/8.

The implementation of vertical enhancement in part A has been described above. The coupling used in horizontal enhancement in part B is basically very similar, comprising a means 26 for eliminating dark area enhancement, a four-mode filter 27, a noise suppression circuit 28, a weight control element 29 and an adder 210. A. Filters 22 and 27

II

11 94007 differ in that the input samples of the filter 22 are vertically consecutive samples (each sample from a different line) while the samples of the filter 27 are from the same line. With respect to the operation of the filter, reference is made to the description of Figure 6.

The table in Figure 7 shows some possible filter coefficients for mid-bandpass FIR filters. The principle of FIR filters per se is within the scope of the basic knowledge of a person skilled in the art 10 and will not be discussed further in this context. The numerical values in the third column describe the weighting factors by which successive samples in the horizontal direction (horizontal peak) are weighted before summation, and the numerical values in the second column describe the weighting factors by which successive samples in the vertical direction (vertical peak) are weighted before summation. In the first column, "Mode" is the branch of the decoder whose output is peaked. As will be appreciated, there are four modes, and in each mode the DATV data processing block provides the filter-20 with a setting that adapts the filter to the characteristics of the decoder branch connected to the output of the decoder and thus the input of the spike circuit.

The invention does not place any restrictions on the implementation of FIR filters. The size of the window in both horizontal and vertical directions can be selected as desired. Vertically FIR filters theoretically require 2 HDTV line memories in 80 ms mode and 4 HDTV line memories in 40 ms and 20 ms modes. In practice, vertical FIR filters can be shortened in 40 ms and 20 ms modes to the same as in 80 ms mode, allowing the same filter to operate in all branches. Horizontal FIR filters do not need to be limited for cost reasons.

35 The sharpening of the luminance signal can be effectively emphasized by the highlighting circuit of the center frequency components of the image shown. Since the filters are individually designed for each processing branch of the MAC decoder, the different frequency responses of these 94007 12 will be taken into account. It should be noted that, while remaining within the scope of the appended claims, the spike coupling can be implemented in other ways than by performing a vertical emphasis first and then a Horison-5 vertical emphasis. Equally, Horizontal highlighting can be performed first, followed by vertical highlighting. It is also possible to proceed by passing the signal from the MAC decoder to two parallel branches, one with vertical enhancement and the other with horizontal enhancement, and then the enhancements are added to the original signal. It is also possible to combine the filters used in vertical enhancement and horizontal enhancement together into a two-dimensional filter. It is further possible to combine the parts so that after the adder in which the fixed value K (dark area enhancement inhibition) is subtracted from the luminance signal, the signal is passed to two parallel filters, one performing Horizontal enhancement and the other vertical enhancement. The highlighted signals are summed and routed after the noise suppression circuit DZ to be summed weighted to the original signal.

»

II

Claims (14)

94007 13 A horizontal and vertical enhancement circuit for enhancing the sharpness of a picture in a high-definition television receiver having an HDMAC decoder subsampling pattern restoring circuit 5 (3) for generating original subsampling patterns of luminance samples, three interpolating branches connected to said restoring circuit sub-sampling patterns representing stationary image areas of the high-definition image, the interpolators (6) of the second branch form a sub-sampling pattern representing the low-motion image areas of the high-definition image, and the interpolators (4) of the third branch form a high-resolution image of in the decoder, in response to the branch decision information of the motion data signal (DATVj decoder (8)), a luminance sample formed by the signal processing branch (4, 5 or 6) indicated by it is connected t the output of the HDMAC decoder as its luminance output signal, characterized in that the connection comprises: - first limiting means (21, 26) which replace sample values smaller than the fixed value (K) of the luminescence output signal with zero samples to form a limited signal, 25. at least one adjustable An FIR filter (22, 27) to which a limited signal is applied to the associated memories to form a sample window and whose center frequency is set in response to branch decision information produced by the DATV signal decoder (8) to match the center of the signal processing branch (4, 5 or 6) connected to the decoder output. Coupling according to Claim 1, characterized in that there are two filters, one of which is an adjustable vertical -FIR filter and the other of which is an adjustable horizontal -FIR filter. The FIR filter and that the coupling further comprises summing means (25, 210) operatively coupled to the filters for summing the vertically filtered and horizontally filtered signals to the luminance output signal of the decoder. 14 94007 A circuit according to claim 1 or 2, characterized in that the filter is an adjustable vertical FIR filter (22) in which a limited signal is applied to successive line memories to form a vertical sample window and whose center frequency is set in response to a DATV signal decoder (8). ) to correspond to the center frequency of the signal processing branch (4, 5 or 6) connected to the decoder output. A circuit according to claim 1 or 2, characterized in that the filter is an adjustable horizontal FIR filter (27) in which a limited signal is applied to successive single-sample memories to form a horizontal sample window and whose center frequency is set in response to a DATV signal decoder (8) to the branch decision information to correspond to the center frequency of the signal processing branch (4, 5 or 6) connected to the output of the decoder. Coupling according to Claims 3 and 4, characterized in that - the vertical FIR filter (22) and the horizontal FIR filter (27) are multimode bandpass filters and one filter mode corresponds to each signal processing branch (4, 5, 6) of the decoder. , 25. in each mode, the filter coefficients of the filters (k1, k2, k3 ...) are determined in response to the branch decision information decoded from the DATV signal such that the center frequency of the filters corresponds to the center frequency of the signal processing branch (4, 5, 6) currently connected to the decoder output. 30 Coupling according to one of the preceding claims, characterized in that second limiting means (23, 28) are connected to each filter (22, 27) for limiting the output signal of the filter so that the samples are replaced by zero samples when the value of the filtered signal is within the estimated noise range (DZ). / 2) inside, preventing the addition of mere noise to the luminance signal. 15 94007 Coupling according to Claim 6, characterized in that in the limiting means (23, 28) the output signal of the filter is limited so that when the value of the filtered signal is close to the boundary of the noise interval 5 within the estimated noise interval (DZ / 2), its sample value is slidably reduced. blank samples do not occur in a jump-like manner. A circuit according to claim 6, characterized in that it has calculation means (9) for determining the estimated noise interval (DZ / 2) from the signal-to-noise ratio calculated from the test line of the automatically received MAC frame. A circuit according to claim 2, characterized in that the output signal of each filter (22, 27) is connected to a weighting means (24, 29) which, in response to a weighting factor (W), weights said output signal before it is applied to the summing means (25, 210). Coupling according to Claim 9, characterized in that the weighting factor (W) is set by the television user. A circuit according to claim 2, characterized in that both the horizontal boost circuit (B) and the vertical boost circuit (A) are separate circuits, wherein the horizontal boost circuit (B) comprises at least a first restricting means (26), an adjustable horizontal -FIR -filter (27) and summing means (210) and the vertical enhancement circuit (A), respectively, comprise at least a first 30 limiting means (21), an adjustable vertical FIR filter (22) and summing means (25). Coupling according to Claims 6 and 11, characterized in that both the horizontal enhancement coupling (B) and the vertical enhancement coupling (A) have their own second restraining means and their own weighting means. 16 94007 A circuit according to claim 11, characterized in that the horizontal circuit (B) and the vertical circuit (A) are connected in series with respect to the luminance output signal of the decoder. 5 A circuit according to claim 11, characterized in that the horizontal circuit (B) and the vertical circuit (A) are connected in parallel with respect to the luminance output signal of the decoder. 10