JP4062793B2 - Pixel number converter - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a pixel number conversion device, and more particularly to a pixel number conversion device that converts the number of pixels of an image signal.
[0002]
[Prior art]
The TV screen has 483 effective scanning lines in the case of a standard television signal, and 1035 effective scanning lines in the case of a high-definition television signal.
[0003]
Therefore, when the standard television signal is displayed on the television monitor for the high-definition television signal, the number of scanning lines is converted to 7:15.
When a standard television signal is displayed on a fixed single display pixel number such as a liquid crystal display, the number of pixels is 800 × 600 for SVGA (Super Video Graphics Array), so the scanning line is approximately 4: 5. Is converted and displayed.
[0004]
Furthermore, when the television signal is enlarged by a little less than 10%, which is necessary for slightly overscan display on a liquid crystal display, or when a personal computer signal is displayed on a CRT with overscan display, the image is out of the screen. In the case of reducing about 10% necessary for preventing the protrusion, the conversion is performed at a ratio of 10:11 or 10: 9.
[0005]
When importance is attached to image quality in such scanning line number conversion, interpolation is generally performed with an interpolation filter that approximates sin (x) / x of the interpolation function.
[0006]
[Problems to be solved by the invention]
However, in contrast to the conventional conversion of the number of scanning lines (hereinafter referred to as pixel number conversion) as described above, the number of scanning lines of the input original signal is smaller at the time of enlargement conversion than the number of scanning lines for display. Therefore, the number of pixels must be increased. For this reason, there has been a problem that it looks blurred as compared with the original signal of the display device.
[0007]
In addition, even if an ideal interpolation filter is used, there is a problem in that image quality deteriorates because information finer than the information of the original image cannot be created by interpolation.
[0008]
The present invention has been made in view of the above points, and an object of the present invention is to provide a pixel number conversion device that improves image quality at the time of pixel number conversion.
[0009]
[Means for Solving the Problems]
In the present invention, in order to solve the above-described problem, in a pixel number conversion device that converts the number of pixels of an image signal, a pixel number conversion unit that performs reduction conversion or enlargement conversion of the number of pixels, and a line memory at the time of the reduction conversion A low-pass filter that limits the band of the image signal, an enhancer that shares the line memory during the expansion conversion, and emphasizes a high-frequency component of the image signal, and the low-pass filter and the enhancer A switch that switches according to the reduction conversion or the enlargement conversion, and the enhancer performs a high-pass filtering process to extract a high-frequency component from the enlarged image signal; And high frequency emphasis means for adding high frequency components whose high frequency characteristics are changed in accordance with the enlargement ratio. Pixel number converter.
[0010]
Here, the enlargement interpolation means enlarges the image signal to generate an enlargement interpolation image signal. The high frequency component extraction means extracts a high frequency component from the enlarged interpolation image signal. The high-frequency emphasis unit adds a high-frequency component obtained by changing the high-frequency characteristics according to the enlargement ratio to the enlarged interpolated image signal, and emphasizes and outputs the high-frequency.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a principle diagram of a pixel number conversion apparatus according to the present invention. The pixel number conversion apparatus 1 performs enlargement conversion of the number of pixels of the image signal.
[0012]
The pixel number conversion apparatus 1 includes an enlargement interpolation unit 10 and an enhancer 20. The enhancer 20 includes a high frequency component extracting unit 21 and a high frequency emphasizing unit 22.
[0013]
The enlargement interpolation means 10 is a low-pass filter (hereinafter referred to as LPF), and enlarges and interpolates an input image signal I based on the enlargement ratio to generate an enlargement interpolation image signal Im.
[0014]
The high-frequency component extraction unit 21 performs high-pass filtering to extract a high-frequency component from the enlarged interpolation image signal Im.
The high frequency region emphasizing unit 22 adds a high frequency component obtained by changing the high frequency characteristics according to the enlargement ratio to the enlarged interpolated image signal Im. Then, the high frequency is emphasized and the converted image signal Ia is output.
[0015]
Here, when the high-frequency emphasis unit 22 emphasizes the high-frequency in two dimensions, the high-frequency component extraction unit 21 includes a two-dimensional transversal high-pass filter. The coefficients at this time are shown in FIG.
[0016]
As described above, the pixel number conversion apparatus 1 according to the present invention is configured to include the enhancer 20 including the high-frequency component extraction unit 21 and the high-frequency emphasis unit 22 at the output of the enlargement interpolation unit 10.
[0017]
Then, the high frequency component is added to the enlarged interpolated image signal Im converted in number of pixels, and the high frequency is emphasized according to the enlargement ratio, thereby suppressing image blur. As a result, the image quality can be improved.
[0018]
Next, conversion of the number of pixels when an LPF having a cosine roll-off characteristic is used as a finite response filter of the interpolation function sin (x) / x, for example, when the ratio of the number of pixels after conversion is 10:11 Will be described.
[0019]
Consider a cosine roll-off interpolation LPF that is processed by the least common multiple clock fm of the input signal clock fi and the output signal clock fo and cuts off fi / 2.
FIG. 3 is a diagram showing frequency characteristics of the cosine roll-off interpolation LPF in the case of 10:11. The vertical axis represents gain (| A |) and the horizontal axis represents frequency (fi × 10).
[0020]
FIG. 4 is a diagram showing a table of coefficients of the cosine roll-off interpolation LPF in the case of 10:11. The coefficient Cn (n = 0 to 20) of the table t1 shows only half from the center, and is scaled by 64 times assuming an actual circuit for integer processing, and then rounded at the integer part.
[0021]
FIG. 5 is an overall block diagram of the cosine roll-off interpolation LPF in the case of 10:11. In the figure, IC1a to IC4a and IC11a are D flip-flops, IC5a to IC8a are 2-input 1-output multipliers, IC9a is a 4-input 1-output adder, SW1a to SW4a are switches for selecting a coefficient Cn, and IC10a is switch-switching This is an hex counter for control.
[0022]
The coefficient Cn is for an interpolation filter (operated by fm) (not shown). The IC 1a is a D flip-flop for timing adjustment.
Next, the connection relationship of each circuit element of the cosine roll-off interpolation LPF 100a will be described. IC1a to IC4a are serially connected. The image signal before conversion is input to the D terminal of the IC 1a. The Q terminal of IC1a and the D terminal of IC2a are connected, the Q terminal of IC2a and the D terminal of IC3a are connected, and the Q terminal of IC3a and the D terminal of IC4a are connected. An input clock fi synchronized with the image signal before conversion is input to the clock terminals of IC1a to IC4a.
[0023]
On the other hand, SW1a is a switch for selecting one coefficient from the coefficients of C12 to C20, SW2a is a switch for selecting one coefficient from the coefficients of C1 to C11, and SW3a is one of the coefficients of C0 to C10. A switch for selecting a coefficient, SW4a is a switch for selecting one coefficient from the coefficients C11 to C20.
[0024]
Each Q terminal of IC1a-IC4a is connected with one input terminal of IC5a-IC8a. The other input terminals of IC5a to IC8a are connected to switch terminals of SW1a to SW4a.
[0025]
An output clock fo synchronized with the converted image signal, that is, (11/10) × fi, is input to the clock terminals of the IC 10a and the IC 11a.
The four output terminals of the IC 10a are connected to the switch switching control terminals SW1a to SW4a, respectively. The four input terminals of IC9a are connected to the output terminals of IC5a to IC8a, respectively, and the output terminal of IC9a is connected to the D terminal of IC11a. Then, the IC 11a outputs the converted image signal in which the number of pixels is converted.
[0026]
In this block diagram, the frequency relationship is fo = (11/10) × fi, fm = 11 × fi. The coefficient Cn is for an interpolation filter operating at fm, and in reality, only one sample is included in 11 samples. Therefore, only C11 + C0 + C11, C10 + C1 + C12, C20 + C9 + C2 + C13,... The output may be output as fo.
[0027]
In the pixel number expansion conversion such as 10:11 as described above, since the number of pixels must be increased by interpolation from the input original signal, it is more blurred than the original signal of the display device. It looks and the image quality deteriorates.
[0028]
Next, the deterioration in image quality will be described from the waveform response. In order to simplify the description, the interpolation filter will be described using an example of linear interpolation instead of cosine roll-off.
[0029]
FIG. 6 is a diagram showing the waveform response of the interpolation filter. (A) is a waveform response before pixel number conversion, (B) is a diagram showing sample points interpolated in the waveform response before pixel number conversion, and (C) is a waveform response after pixel number conversion. . The vertical axis is amplitude, and the horizontal axis is time.
[0030]
In (A), white circles in the figure indicate sample points of the original signal S. In addition, the rising slope of the original signal S is Sa.
In (B), the black circles in the figure are sample points when 11 times the sample points from the sample points (white circles) of the original signal S are obtained by interpolation calculation, and the square marks are 1 of the sample points (black circles). Sample points of / 10 are shown.
[0031]
(C) shows an enlarged signal S-1 obtained by extending the sampling points (square marks) of (B) to the sampling period of the original signal S. The rising slope of the enlarged signal S-1 is Sb.
[0032]
As shown in the figure, the rising slope Sa of the original signal S becomes the slope Sb after enlargement, so that the rising slope becomes small and the image appears visually blurred. This is more conspicuous as the enlargement ratio is larger.
[0033]
Therefore, the pixel number conversion apparatus 1 of the present invention enhances the output signal after enlargement interpolation to make the image blurring at the time of enlargement visually inconspicuous.
Next, the enhancer 20 will be described. FIG. 7 is a diagram showing a configuration of the enhancer 20. The enhancer 20 includes a high frequency component extracting unit 21 and a high frequency emphasizing unit 22.
[0034]
IC 21a and IC 21b are D flip-flops, IC 21c and IC 21d are multipliers, and IC 21e is a 3-input 1-output adder. IC 22a is a multiplier, and IC 22b is a 2-input 1-output adder.
[0035]
The connection relationship of circuit elements will be described. The enlarged interpolation image signal Im output from the enlargement interpolation means 10 is input to the D terminal of the IC 21a. The Q terminal of IC 21a is connected to the D terminal of IC 21b, the input terminal of IC 21c, and the input terminal of IC 22b. Also, an output clock fo synchronized with the converted image signal is input to the clock terminals of the ICs 21a and 21b.
[0036]
The three input terminals of the IC 21e are connected to the D terminal of the IC 21a, the output terminal of the IC 21c, and the Q terminal of the IC 21b, respectively. And it adds with ratio of -1: 2: -1 with respect to each output.
[0037]
The output terminal of the IC 21e is connected to the input terminal of the IC 21d, and the output terminal of the IC 21d is connected to the input terminal of the IC 22a. The output terminal of the IC 22a is connected to the input terminal of the IC 22b.
[0038]
In this way, the enhancer 20 that operates with the fo clock is installed in the output (enlarged interpolation image signal Im) of the pixel number conversion of the linear interpolation. Then, the enhancer 20 takes out the high frequency component with the HPF of −1: 2: −1, and adds an appropriate amount thereof (according to the enlargement ratio) to the enlarged interpolated image signal Im.
[0039]
FIG. 8 is a diagram showing frequency characteristics of the present invention in the case of 10:11. The vertical axis represents gain (| A |), and the horizontal axis represents frequency (fi × 10). As can be seen from the characteristics, the midrange of the pass band is raised compared to the frequency characteristics of the cosine roll-off in FIG.
[0040]
Next, this will be explained with a waveform response. FIG. 9 is a diagram showing the waveform response of the present invention in the case of 10:11. The vertical axis is amplitude, and the horizontal axis is time.
The sample points obtained by the interpolation calculation are indicated by square marks, and the output through the enhancer 20 is indicated by black square marks.
[0041]
Here, the values of the sample points (square marks) obtained by the interpolation calculation are 20, 20 (s), 20 (t), 65 (u), 75 (v), 75, 75.
The output of the sample position is ((−1 * s + 2 * t−1 * u) / 4) * 0.5 by the processing in the high frequency component extraction means 21 of the enhancer 20 shown in FIG.
[0042]
Thus, when the value of each sample position is calculated, it becomes 0 (s), -5.6 (t), 4.4 (u), 1.3 (v), 0.
Then, the value of the output through the enhancer 20 (black square mark) is added to add 20, 20 (s), 14.4 (t), 69.4 (u), 76.3 (v ), 75, 75.
[0043]
Compared with FIG. 6, it can be seen that the rising slope Sb before enhancement becomes the slope Sc after enhancement, and the rising slope rises and approaches the slope Sa of the original signal. Therefore, the visual blur is improved.
[0044]
Next, a case where the enhancement amount is adaptively controlled according to the enlargement ratio will be described. The degree to which the slope of the rising edge of the original signal becomes smaller after enlargement and appears visually blurred becomes larger as the enlargement ratio increases.
[0045]
Therefore, the effect can be further enhanced by increasing the enhancement amount (high frequency enhancement amount) in conjunction with the enlargement ratio. FIG. 10 is a diagram showing the configuration of the present invention when the enhancement amount is controlled according to the enlargement ratio.
[0046]
For the circuit described with reference to FIG. 7, a coefficient setting unit 22 a is newly provided in the high frequency emphasizing unit 22. In the figure, for example, the enhancement amount can be controlled adaptively based on the enlargement rate, such as an enhancement amount of 0.5 if the enlargement rate is 1.1, and an enhancement amount of 0.7 if the enlargement rate is 1.2.
[0047]
Further, the HPF characteristics of the high frequency component extraction means 21 are not determined considering the conversion ratio of the pixel conversion or the characteristics that make the pixel conversion close to the ideal characteristics, but the blurring of the image as a result of the display is obtained. Since the purpose is to make it difficult to feel, it is advantageous that the frequency be close to the display upper limit of the display device. If it is lower than this, interference such as ringing is worrisome. Conversely, if it is higher, the effect is reduced.
[0048]
Therefore, when the resolution of the display device is lower than the upper limit Nyquist frequency determined from the output clock frequency, a transversal HPF (BPF) having a coefficient of -1: 2: -1 is used for the high frequency component extraction means 21. Better results.
[0049]
Next, the configuration and operation when nonlinear characteristic processing is provided inside the enhancer 20 will be described. FIG. 11 is a diagram showing an enhancer in the case where processing for nonlinear characteristics is provided. The same circuit elements as those in FIG.
[0050]
In the enhancer 20a, a non-linear characteristic processing unit 23 is provided between the output of the high frequency component extraction unit 21 and the input of the high frequency emphasizing unit 22.
The nonlinear characteristic processing means 23 performs nonlinear characteristic processing on the output of the high frequency component extraction means 21. That is, when the output from the high frequency component extraction means 21 is added by the IC 22b, the small amplitude component is removed so that the noise component included in the input signal is not enhanced together with the signal.
[0051]
Furthermore, even if a large amplitude signal is emphasized, if the display device exceeds the dynamic range of the display device, the number of pixels may increase, or the color may become colored, which may have an adverse effect. Make restrictions.
[0052]
Next, a description will be given of a pixel number conversion apparatus that performs both horizontal pixel number conversion and vertical pixel number conversion. FIG. 12 is a diagram showing a pixel number conversion apparatus when both horizontal pixel number conversion and vertical pixel number conversion are performed.
[0053]
The expansion interpolation unit 10b includes a horizontal pixel number conversion unit 11 (LPF in the horizontal direction) that performs linear interpolation of the number of horizontal pixels, and a vertical pixel number conversion unit 12 (LPF in the vertical direction) that performs linear interpolation of the number of vertical pixels. Composed.
[0054]
The two-dimensional enhancer 20b includes a 3 × 3 two-dimensional transversal HPF 21-1 (having the coefficients shown in FIG. 2), a multiplier IC 22b, and an adder IC 22a. Moreover, you may provide the nonlinear characteristic processing means 23. FIG.
[0055]
As described above, when both the horizontal pixel number conversion and the vertical pixel number conversion are performed, after interpolation with the horizontal and vertical linear LPFs, only one set such as a non-linear circuit is required if the two-dimensional enhancer 20b is used. An image with good visual characteristics can be obtained.
[0056]
Next, a description will be given of a pixel number conversion apparatus that shares a band-limited LPF and an enhancer. In the pixel number conversion, both reduction and enlargement processes are often performed. For example, when a standard television signal and a high-definition television signal are switched and displayed on a VGA display panel, the number of effective scanning lines per field of the original signal is about 240 and about 517, respectively. When the number of pixels is converted to the number of vertical dots 480, the former is enlarged and the latter is reduced.
[0057]
In the case of reduction, since the frequency of the signal is increased by the conversion of the number of pixels and the interference of sampling aliasing occurs, the band is generally limited by the LPF before the conversion of the number of pixels. Since this LPF often uses a transversal filter, the configuration is similar to that of an enhancer.
[0058]
Do not zoom in and out at the same time. Therefore, by sharing the band-limited LPF before the pixel number conversion and the enhancer after the pixel number conversion, the circuit can be reduced. In particular, when the vertical filter is made into an IC, the line memory used for the delay is shared. Thus, the circuit scale can be reduced.
[0059]
FIG. 13 is a diagram illustrating a configuration of a pixel number conversion apparatus that shares a band-limited LPF and an enhancer. IC33a and IC33b are line memories, IC33c and IC33dIC33e, IC32a is a multiplier, IC33f and IC33g, and IC32b are adders.
[0060]
A connection relationship of circuit elements of the pixel conversion device 3 will be described. An image signal is input to the a terminal of the switch SW1 and the b terminal of the switch SW2. The c terminal of the switch SW1 is connected to the input terminal of the line memory IC 33a and one input terminal of the adder IC 33f.
[0061]
The output terminal of the line memory IC 33a is connected to the input terminal of the multiplier IC 33c and the input terminal of the line memory IC 33b. The output terminal of the line memory IC 33b is connected to the other input terminal of the adder IC 33f.
[0062]
The output terminal of the adder IC33f is connected to the terminal a of the switch SW3 and the input terminal of the multiplier IC33d. The output terminal of the multiplier IC 33d is connected to the b terminal of the switch SW3.
[0063]
One input terminal of the adder IC33g is connected to the output terminal of the multiplier IC33c, and the other input terminal is connected to the c terminal of the switch SW3.
The output terminal of the adder IC33g is connected to the input terminal of the multiplier IC33e, and the output terminal of the multiplier IC33e is connected to the c terminal of the switch SW4.
[0064]
The a terminal of the switch SW4 is connected to the a terminal of the switch SW2, and the c terminal of the switch SW2 is connected to the input of the pixel number conversion means 30. The output of the pixel number conversion means 30 is connected to the b terminal of the switch SW1 and one input terminal of the adder IC 32b.
[0065]
The b terminal of the switch SW4 is connected to the input terminal of the multiplier IC 32a, and the output terminal of the multiplier IC 32a is connected to the other input terminal of the adder IC 32b.
Next, the operation will be described. In the case of reduction in which the switches SW1 to SW4 are switched upward at the time of reduction and downward at the time of enlargement, the input signal first enters a 1: 2: 1 vertical LPF using the line memory ICs 33a and 33b, and the output thereof is the pixel number conversion means 30. The conversion result is output as it is. In the case of reduction, there is no blur, so enhancement control is not necessary.
[0066]
In the case of enlargement, the input signal is first input to the pixel number conversion means 30, and the conversion result enters the -1: 2: -1 vertical HPF using the line memory ICs 33a and 33b, and the output is enhanced.
[0067]
In the case of enlargement, an input band-limited LPF is not necessary. With such an operation, the line memory ICs 33a and 33b and the filter addition portion are shared, so that processing for two circuits can be performed with one set of circuits.
[0068]
In particular, the line memory ICs 33a and 33b have 90k bits (1920 dots * 2 lines * 8 bits * 3ch) per circuit when RGB3ch processing is performed on, for example, an 8-bit HDTV signal. Therefore, it is possible to increase the effect of reducing the circuit scale, power consumption, and the like as compared with the case of not sharing.
[0069]
As described above, the pixel number conversion apparatus 1 according to the present invention is configured to enhance the output at the time of enlargement and reduce blurring with respect to the pixel number conversion of a television or the like. Further, the pixel number conversion device 3 of the present invention is configured to share the LPF for limiting the bandwidth of the input for reduction and the enhancer for enlargement. Thereby, the circuit can be reduced. In this case, the effect of reducing the vertical line memory is particularly great.
[0070]
In the above description, the signal delay in the digital processing is omitted, but the circuit scale and the like are almost the same as the circuit scale in consideration of the signal delay.
[0071]
【The invention's effect】
As described above, the pixel number conversion device of the present invention emphasizes the high-frequency by adding the high-frequency components whose high-frequency characteristics are changed according to the enlargement ratio after the image signal is enlarged and interpolated. Output. As a result, it is possible to improve the image quality when converting the number of pixels.
[Brief description of the drawings]
FIG. 1 is a principle diagram of a pixel number conversion apparatus according to the present invention.
FIG. 2 is a diagram illustrating coefficients of a two-dimensional transversal high-pass filter.
FIG. 3 is a diagram illustrating frequency characteristics of a cosine roll-off interpolation LPF in the case of 10:11.
FIG. 4 is a diagram showing a table of coefficients of a cosine roll-off interpolation LPF in the case of 10:11.
FIG. 5 is an overall block diagram of a cosine roll-off interpolation LPF in the case of 10:11.
FIG. 6 is a diagram illustrating a waveform response of an interpolation filter. (A) is a waveform response before pixel number conversion, (B) is a diagram showing sample points interpolated in the waveform response before pixel number conversion, and (C) is a waveform response after pixel number conversion. .
FIG. 7 is a diagram illustrating a configuration of an enhancer.
FIG. 8 is a diagram showing frequency characteristics of the present invention in the case of 10:11.
FIG. 9 is a diagram showing a waveform response of the present invention in the case of 10:11.
FIG. 10 is a diagram showing a configuration of the present invention when the enhancement amount is controlled in accordance with the enlargement ratio.
FIG. 11 is a diagram illustrating an enhancer in the case where a process for nonlinear characteristics is provided.
FIG. 12 is a diagram illustrating a pixel number conversion apparatus when both horizontal pixel number conversion and vertical pixel number conversion are performed.
FIG. 13 is a diagram illustrating a configuration of a pixel number conversion apparatus that shares a band-limited LPF and an enhancer.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Pixel number conversion apparatus, 10 ... Expansion interpolation means, 20 ... Enhancer, 21 ... High frequency component extraction means, 22 ... High frequency enhancement means, I ... Image signal, Im ... Expansion interpolation Image signal, Ia: Output signal.

Claims (5)

In a pixel number conversion device that converts the number of pixels of an image signal,
A pixel number converting means for performing reduction conversion or enlargement conversion of the number of pixels;
A low-pass filter that limits the band of the image signal using a line memory during the reduction conversion;
An enhancer that emphasizes high frequency components of the image signal by sharing the line memory during the enlargement conversion;
A switch for switching the low-pass filter and the enhancer according to the reduction conversion or the enlargement conversion;
Have
The enhancer performs high-pass filtering processing, and extracts a high frequency component from the enlarged image signal;
Has high frequency emphasis means that adds high frequency components whose high frequency characteristics are changed according to the magnification ratio
A pixel number conversion device characterized by the above.   2. The pixel number conversion device according to claim 1, wherein the high frequency component extraction means is a transversal high pass filter having a coefficient of -1: 2: -1.   2. The pixel number conversion apparatus according to claim 1, further comprising nonlinear processing means for nonlinearly processing the high frequency components.   2. The pixel number conversion apparatus according to claim 1, wherein the high-frequency component extraction means is a two-dimensional transversal high-pass filter.   5. The pixel number conversion apparatus according to claim 4, wherein the high-frequency emphasis means enhances the high-frequency in two dimensions.

Images