JP4422560B2 - Image processing circuit and image display device - Google Patents

Description

  The present invention relates to an image processing technique for reducing the pseudo contour and noise of a display image by performing signal processing on the image display signal and improving the display image quality.

  When a low-quality video source is played, contour lines called pseudo contours may appear. This pseudo contour is generated due to a gradation step that should not originally exist as a result of quantization being performed more than necessary on a smooth gradation original image, and this impairs image quality. It is the cause.

  As a method of reducing the pseudo contour and image noise, there is a method of passing an input signal through a low pass filter (LPF). When this LPF is configured by a 1-dimensional FIR (Finite Impulse Response) filter having 2k + 1 taps, the LPF output y (n) when the input signal sequence is x (n) is expressed by the following equation (1). Can be written as

ここでa(i)は、下記（２）式の関係を満足するような係数である。

Here, a (i) is a coefficient that satisfies the relationship of the following equation (2).

  In this method, since the high-frequency signal component of the original image is cut, the contour portion other than noise and pseudo contour is uniformly smoothed (linear filter processing), and the sharpness of the image is lost. There is.

  On the other hand, the input image signal is blurred through a spatial low-pass filter, and the difference between the blurred output signal and the original image signal is taken. If the difference is small, the blurred low-pass filter Based on the equations (3) to (5), both components are adaptively adapted so that an output can be obtained, and when it is large, this is regarded as a useful signal and the value of the original image signal is retained. There is a method of outputting (see, for example, Patent Document 1).

  This method makes it possible to avoid the drawbacks of a simple smoothing filter (linear filter) and to preserve a significant contour by using the nonlinear function F. Further, since this non-linear function works on the filter output, the image quality can be improved with a relatively small circuit scale. Hereinafter, processing using this nonlinear function is referred to as coring.

JP 2000-295497 A

  FIG. 2 is a diagram showing signal waveform examples at each stage when the technique described in Patent Document 1 is used.

  When the signal is smoothed using the method described in Patent Document 1, a shadow (ghost) of an edge that does not originally exist is displayed lightly at a position slightly away from the edge of the image. As shown in FIG. 2A, an edge waveform and a pseudo waveform are observed in the input original signal.

  This ghost phenomenon is a phenomenon peculiar to the coring process and occurs for the following reason. When an image having sharp edges (FIG. 2A) is processed by the LPF of the equation (1), the edges are smoothed with peripheral pixels. When coring processing according to the equation (2) is performed using the smoothed signal (FIG. 2B), the difference between the smoothed signal level and the input signal level in the vicinity of the edge (FIG. 2C). ) Is large, the edge is preserved. However, at a location slightly away from the edge, the difference between the smoothed result and the input signal becomes small, so that the information of the edge smoothed by the LPF remains (FIG. 2 ( D)), the influence of the edge appears as a ghost (FIG. 2E).

  Such a ghost phenomenon is applied not only to the case where a simple comparison process using a threshold value is applied as a nonlinear function as shown in equation (3), but also to the processing of complex input / output relationships using a LUT (Look Up Table). Even in this case, it will occur as a fundamental problem.

  The present invention has been made to solve the above problems, and suppresses the occurrence of ghosts in all display areas when performing a coring processing method that reduces pseudo contours and noise while preserving the contours of the original image. The purpose is to do.

  According to one aspect of the present invention, there is provided a low-pass filter used for an input image signal represented by a bit string, wherein a difference between a tap signal and a processing target input signal is calculated for each tap, and the calculated difference is calculated. A non-linear low-pass filter that multiplies a filter signal by a filter coefficient when a pixel signal to be processed is larger than a threshold value, and adds a result of the multiplication when the pixel signal is smaller, and outputs the result of the multiplication; A subtractor for subtracting the output of the non-linear filter from the signal; a coring processing circuit for performing a coring process with a non-linear function having a characteristic of blocking and attenuating the output of the subtractor at a certain frequency; and an output of the coring processing circuit And an adder for adding the output of the nonlinear low-pass filter.

  In the nonlinear filter, when an edge equal to or greater than the comparison threshold exists at a position away from the pixel of interest, the pixel signal used for the calculation in the vicinity of the edge is replaced with the element of interest by the action of the tap output selector. Here, the digital filter processing means that the pixel signal to be processed and its surroundings correspond to the product-sum operation of the prime signal and the filter coefficient, so that the influence of the edge can be eliminated from the result of the product-sum operation by this replacement processing. .

  Further, the non-linear filter sets a signal signal outside the display area to an appropriate constant level according to the input image, and outputs the signal together with the image signal within the display area from a signal from the display area signal generation circuit. Is entered.

  According to the signal processing circuit of the present invention, it is possible to suppress the occurrence of ghosts related to the edge without losing the sharpness of the edge while removing noise and pseudo contour, and high quality in all display areas. Display is possible.

  In the image processing apparatus according to the present invention, the circuit characteristics can be changed as necessary to enable / disable the function, so that an appropriate image can be displayed according to the characteristics of the input signal source and the user's preference. There is an advantage.

  In this specification, an output for each delay unit is referred to as a “tap”. Hereinafter, an image processing technique according to an embodiment of the present invention will be described with reference to the drawings. In the following description, for simplification of description, the signal is handled in one dimension, but the same processing can be applied in the case of two dimensions. First, the principle of the present invention will be briefly described with reference to FIG. In FIG. 2, the pixel of interest is x (a), an edge exceeding the comparison threshold exists at a position away from the pixel of interest, and a pixel near the edge is x (b). At this time, the tap length of the filter is 2k + 1, and the distance between a and b is shorter than k. When a general linear filter is used, since x (b) is used in the filter calculation for the pixel of interest x (a), x (b) is reflected in the calculation result. As a result, as shown in FIG. Ganamaru. In contrast, the image processing apparatus according to the present invention uses a nonlinear filter. If a non-linear filter is used, all computations in this case use x (a), so the processing result at the position of a is not affected by x (b), and the result is stored as it is. In the case of a level change smaller than the comparator threshold value such as a pseudo contour, the output is smoothed as in the conventional linear filter. Therefore, since the filter output is not rounded at the edge exceeding the comparator threshold value, no ghost is generated even if the coring processing from FIG. 2C to FIG. 2E is performed.

  FIG. 1 is a block diagram showing a configuration example of an image processing circuit according to an embodiment of the present invention. As shown in FIG. 1, the image processing circuit includes a nonlinear LPF 2, a subtracter 3, a coring processing circuit 4, and an adder 5. The LPF 2 receives an input image signal and a display area signal and generates an out-of-display-area signal generating main signal 1 and gives an output signal to the subtracter 3 and the adder 5. More specifically, the LPF 2 is an FIR-type nonlinear LPF having a tap length of 2k + 1. The non-linear LPF 2 receives the output from the out-of-display area signal generation circuit 1 and delays the signal for each stage, and each stage including the output signal from the out-of-display area signal generation circuit 1. A plurality of tap output selectors 22 that respectively receive the outputs from the delay units 21, a plurality of coefficient multipliers 23 that respectively receive the outputs from the respective tap output selectors 22, and an addition that adds the output signals of the plurality of coefficient multipliers 23 And a container 24. Each of the tap output selectors 22 includes a subtractor 221, an absolute value circuit 222, a comparator 223, and a selector 224. First, the operation of the nonlinear LPF 2 will be described.

  The signal input to the non-linear LPF 2 from the signal generating circuit 1 outside the display area is sent to the tap output selector 22 for each sampling clock by the delay unit 21. In the tap output selector 22, the difference between the tap input signal and the signal of interest (tap center signal: i = n) is calculated by the subtractor 221, and the calculation result is input to the absolute value circuit 222. Since the tap input pixel is a signal at a position separated from the target (signal processing) pixel by the sampling clock, the difference between the tap input signal and the target signal (tap center signal: i = n) is large. This shows that a large change such as an edge exists at a position separated from the target pixel by the sampling clock.

  The signal level difference information output from the absolute value circuit 222 is input to the input terminal A of the comparator 223. The comparator 223 has a comparator threshold value designated by an external register connected to the other input terminal B. When the A input is larger than the B input, the output Y is set to “0”, and the A input is more than the B input. If it is small, “1” is output as the output Y.

  The signal of interest and the tap input signal are input to the selector 224 and controlled according to the output of the comparator 223. When the output of the comparator 223 is “0”, the signal of interest is output and the output is “1”. In the case of, a tap input signal is selected and output. Therefore, the tap output selector 22 outputs the signal of interest when there is a change equal to or greater than the comparison threshold, such as an edge, at a position separated from the pixel of sampling clock by the sampling clock, and otherwise outputs the tap output signal as before. By outputting, the influence of an edge existing at a position away from the pixel of interest can be removed.

  The output signal of the tap output selector 22 is multiplied by filter coefficients a (n−k) to a (n + k) by the coefficient multiplier 23. Since the filter output is the sum of 2k + 1 multiplication data by the adder 24, the nonlinear LPF 2 is affected by the edge that has been a problem when using the conventional filter at a position separated from the target pixel by the sampling clock. Can be reduced. When the difference from the edge portion is equal to or less than the comparison threshold value, it functions as a normal LPF, and the image is smoothed as usual. By changing this comparison threshold, the degree of removal of the influence (ghost) of the edge can be changed. It is desirable that the comparator threshold value is held in a register or the like and appropriately rewritten from the outside according to the display contents.

  By the way, the image signal is composed of a low frequency component and a high frequency component. Therefore, when the output result of the LPF is subtracted from the original signal by the subtracter 3, a high frequency component can be extracted and works as an HPF (High Pass Filter). Coring is signal processing for blocking or attenuating a high frequency component above a certain level. The coring processing circuit 4 can be configured by a comparator 42 and a selector 43 as shown in FIG. The high frequency component signal input to the coring processing circuit 4 has a higher frequency as its absolute value increases (the 0 level as a signal corresponds to a DC component). Therefore, the output of the absolute value circuit 41 and the frequency level to be cut off are compared by the comparator 42. If the output of the absolute value circuit 41 does not exceed the cut off level, the selector 43 outputs 0 and the output of the absolute value circuit 41 is cut off. If the level is exceeded, the input signal is output as it is, so that only minute change components such as noise and pseudo contour can be removed from the signal.

  When the signal is attenuated nonlinearly, for example, an attenuation curve as shown in FIG. 3B may be written in an LUT composed of ROM (Read Only Memory) / RAM (Random Access Memory) or the like. ). Further, since the frequency of the signal to be attenuated is only a portion relatively close to the DC component, in FIG. 3A, an attenuation upper limit level is given instead of the cutoff level, and the selector selects the LUT output. If configured, the capacity of the LUT can be reduced (FIG. 3C). Furthermore, since the difference between the input signal to the coring circuit and the output signal level indicated by the corresponding attenuation curve is generally small, only this difference information is stored in the LUT and the difference is subtracted from the LUT input. The capacity can be further reduced (FIG. 3D).

  If the LUT capacity is reduced in this way, there is an advantage that the occupied capacity related to the ROM and RAM capacity can be reduced and the cost can be reduced. In addition, when a high-speed RAM must be used for this LUT due to the operating speed of the memory element, etc., the capacity of data downloaded from the external ROM is reduced. Therefore, it contributes to the capacity reduction of the external ROM, and since the amount of data is reduced, there is an advantage that the time required for the initial load can be reduced and the startup time can be shortened. Further, by suppressing the capacity of one LUT, it is possible to hold a plurality of LUTs in advance and select them as necessary.

  The frequency and amplitude characteristics of the high-frequency components that are cut off or attenuated by the coring processing circuit 4 can be switched according to the input image and the comparison threshold value of the nonlinear LPF 2, or can be changed by rewriting to smooth the noise. And the degree of preservation of edges and textures can be optimally given.

  The adder 5 adds the output of the coring circuit 4 to the output of the nonlinear LPF 2. The output of the nonlinear LPF 2 has edge ghosts removed, and the coring circuit 4 attenuates unnecessary high frequency components (noise, pseudo contour, etc.), and necessary edge information above the cutoff / attenuation frequency remains. Therefore, when these signals are added together, an output image signal that eliminates noise and pseudo contour from the input image signal, does not lose edge sharpness, and does not generate edge ghost can be obtained.

  In the above description of the arithmetic processing, it is assumed that the input image is infinitely continuous, but the actual input image signal is discontinuous because there is no data signal outside the display area. There are many cases. In general, when performing a filter operation on an image signal, if the processing pixel is located outside the half tap length from the edge of the image, a part of the tap input signal is out of the display area, When correct image data does not exist, there is a problem that the corresponding tap input becomes invalid and a correct calculation result cannot be obtained. As a method for improving this point, there is a method in which a temporary signal is generated outside a display area that does not originally exist, added to the original signal, and the whole is processed. As a specific tentative signal, mirror processing that makes the display area line-symmetric mirror image data exist outside the display area at the edge of the display image area, fixing black (level 0) or white (level 1) There is a method of making a signal masked with data.

  Among these, the mirror processing is a method in which mirror image data that is line-symmetric with respect to the display area exists outside the area with the edge of the display image area as a boundary, and is used as temporary data and added to the original signal. When this method is used, normal processing can be performed even if the processing pixel is at the end of the display region, and the obtained result has no adverse effect such as ghost, and there is an advantage that high image quality can be obtained. However, since it is necessary to replace the mirror image data for each sampling clock with respect to tap input outside the area, there is a problem that the scale of the signal generation circuit outside the display area increases.

  On the other hand, the method of uniformly masking the outside of the display area with fixed data of 0 or 1 is very simple in terms of circuit, and the display area outside signal generation circuit 1 does not require any special device. However, since the image becomes discontinuous inside and outside the display area, the obtained result has a problem that interference occurs near the edge of the screen.

  By the way, in the nonlinear LPF 2 of the present invention, even if the image becomes discontinuous inside and outside the boundary of the display area, if the difference is larger than the comparison threshold value, no ghost is generated in the display area (FIG. 4A). (B), FIG. 4 (A) shows the case where the difference between the inside and outside of the display area is smaller than the comparator threshold value, and FIG. 4 (B) shows the case where the difference inside and outside the display area is larger than the comparator threshold value. Therefore, there is a method of adaptively setting to 0 or 1 according to the setting method when the data signal outside the display area is generated in the signal generating circuit 1 outside the display area.

  FIG. 5 is a diagram showing a configuration example of such an out-of-display-area signal generation circuit 1. As shown in FIG. 5, the end of the display area is detected by a DE (Data Enable) signal indicating the display area of the input image, and the level of the input image signal at the end in the display area end-threshold comparator 11 is 0 ( The selector 12 is controlled to set 1 (maximum) as a signal outside the display area if close to (minimum), and to set 0 (minimum) as a signal outside display area if the level of the input image signal is close to 1 (maximum). To do. For this determination, for example, if the image signal is given in 8 bits, the most significant bit can be used. Since the comparison threshold given to the nonlinear LPF 2 does not exceed 1/2 of the level range that the input signal can take, such simple determination is sufficient, and there is no deterioration in image quality, and the circuit configuration is simple. Works well with.

  Since the input image signal that has passed through the display area outside signal generation circuit 1 has a large edge inside and outside the display area, ghosts from outside the display area can be removed by the nonlinear LPF 2. Therefore, in the image processing apparatus according to the present embodiment, it is possible to obtain a high-quality processing result in the entire display image area.

  FIG. 6 is a block diagram illustrating a configuration example of a liquid crystal display device including the image processing circuit according to the present embodiment. In this example, a liquid crystal television with a built-in tuner will be described as an example of the liquid crystal display device. The liquid crystal television includes a TV signal processing circuit 6, and this TV signal processing circuit 6 includes an antenna 61, a tuner 62, a video signal processing circuit 63, and a scaler 64, and the data signal and the synchronization signal from the scaler 64 are as described above. To the image processing circuit. The liquid crystal panel module 7 includes a timing controller 71, a source driver 72, a gate driver 73, and a TFT liquid crystal panel 74. The timing controller 71 receives a data signal and a synchronization signal from the image processing circuit, and the timing controller 71 outputs the data signal. By sending a scanning signal to the source driver 72 to the gate driver, display is performed on the screen of the TFT liquid crystal panel 74.

  In this configuration example, a comparison threshold is given to the nonlinear LPF 2 via the register setting circuit 8, and an attenuation setting is given to the coring processing circuit 4. The register setting circuit 8 is fetched from the user interface 81 via the microcomputer 82, a user interface (I / F) 81 realized by combining a switch such as a remote controller and a menu display, a microcomputer 82 for controlling the user interface 81 It can be configured by a register 83 for storing setting information. The setting state of the register 83 is displayed on the television screen 74 by, for example, an OSD (On Screen Display: information display on display screen) function.

  As a function of the image processing circuit according to the present embodiment, on the image quality adjustment menu displayed on the OSD, it is displayed with names such as noise reduction (noise removal), S / N (Signal / Noise) improvement, smoother, It is convenient to be able to select the ON / OFF of the function and the level of smoothing from ON / OFF, weak (sharp) to strong (soft), etc. In the menu setting, the microcomputer 82 passes the image processing circuit when “OFF” is selected, and outputs data of the coring to the register 83 so that the result of coring is output when “ON” is selected. Write the signal switching setting value. Similarly, the comparator threshold value and the attenuation setting are written in the register 83 in accordance with the menu setting weak (sharp) to strong (software).

  The image processing circuit operates according to the value stored in the register 83. Specifically, when “strong” is set, the cutoff frequency is set high in the coring processing circuit 4 or the LUT is selected so as to increase the attenuation, and the comparator threshold value is set in the nonlinear LPF 2 accordingly. Make it high. On the other hand, when “weak” is set, the cutoff frequency is set low in the coring processing circuit 4 or the LUT is selected so that the attenuation amount is small, and the comparison threshold is lowered in the nonlinear LPF 2 accordingly. . With these functions, in the liquid crystal television, the user can perform an appropriate image quality setting suitable for the video source.

  As described above, according to the signal processing circuit according to the present embodiment, it is possible to suppress the occurrence of ghosts related to the edge without losing the sharpness of the edge while removing noise and pseudo contour, and a high-quality display can be achieved. It becomes possible. In addition, the image processing apparatus according to the present embodiment is configured so that the circuit characteristics can be changed as necessary to enable / disable the above functions. There is an advantage that an appropriate image can be displayed according to the user's preference.

  The present invention is applicable to an image processing device and a display device using the image processing device, such as a liquid crystal television.

It is a block diagram which shows the structural example of the image processing circuit by one embodiment of this invention. FIGS. 2A to 2E are diagrams showing how a ghost occurs when an input image signal is processed by a conventional method. 3A to 3D are diagrams showing configuration examples of the coring processing circuit according to the embodiment of the present invention. FIG. 4A is a diagram illustrating the operation of the nonlinear LPF according to an embodiment of the present invention at the end of the display region, and FIG. 4A is a diagram when the difference between the inside and the outside of the display region is smaller than the comparator threshold. B) is a diagram when the difference between the inside and outside of the display area is larger than the comparator threshold value. It is a figure which shows the structural example of the signal generation circuit outside a display area by one embodiment of this invention. It is a block diagram which shows the structural example of a liquid crystal display device provided with the image processing circuit by this Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Out-of-display-area signal generation circuit, 2 ... Nonlinear LPF, 3 ... Subtractor, 4 ... Coring processing circuit, 5 ... Adder, 6 ... TV signal processing circuit, 7 ... Liquid crystal panel module, 8 ... Register setting circuit, DESCRIPTION OF SYMBOLS 11 ... Display area edge-threshold comparator, 12 ... Selector, 21 ... Delay device, 22 ... Tap output selector, 23 ... Coefficient multiplier, 24 ... Adder, 41 ... Absolute value circuit, 42 ... Comparator, 43 ... Selector 44 ... LUT, 61 ... antenna, 62 ... tuner, 63 ... video signal processing circuit, 64 ... scaler, 71 ... timing controller, 72 ... source driver, 73 ... gate driver, 74 ... TFT liquid crystal panel, 81 ... user interface, 82 ... Microcomputer, 83 ... Register, 221 ... Subtractor, 222 ... Absolute value circuit, 223 ... Comparator, 224 ... Selector, 441 ... Absolute value circuit, 442 ... Comparator 443 ... LUT, 444 ... selector, 445 ... absolute value circuit, 446 ... comparator, 447 ... LUT, 448 ... subtractor, 449 ... selector.

Claims (8)

A low-pass filter that is used for an input image signal represented by a bit string, and calculates the difference between the tap signal and the input signal to be processed for each tap, and the calculated difference is greater than a threshold value A pixel signal to be processed, or a tap signal if it is small, is multiplied by a filter coefficient, the result of the multiplication is added and output, and the output of the nonlinear filter is subtracted from the input image signal A coring processing circuit for coring processing using a nonlinear function having a characteristic of blocking and attenuating the output of the subtractor at a certain frequency, an output of the coring processing circuit, and an output of the nonlinear low-pass filter an image processing apparatus having an adder for adding the bets,
The signal level outside the display area of the input image signal input to the nonlinear low-pass filter is set to the maximum value or the minimum value, and the set signal outside the display area is combined with the image signal within the display area. An image processing apparatus comprising an out-of-display-area signal generation circuit that outputs to the nonlinear low-pass filter. 2. The image processing apparatus according to claim 1 , wherein the out-of-area signal generation circuit selects a signal level outside the display area to be set based on a signal level in the display area. The out-of-display-area signal generation circuit is set so that a level difference from the in-display area exceeds the threshold value, and the non-display-area signal set is combined with the image signal in the in-display area and the non-linearity the image processing apparatus according to claim 1 or 2, characterized in providing a low-pass filter. The image processing apparatus according to any one of claims 1 to 3, characterized in that it has a first setting unit for setting from outside the threshold used in the nonlinear low-pass filter. The image processing according to any one of claims 1 to 4, characterized in that it has a second setting section for changing from outside the cutoff frequency or the damping characteristic used at said coring circuit apparatus. The coring processing circuit stores a difference between an input signal to the coring processing circuit and a signal indicated by a corresponding nonlinear attenuation curve, adds the input signal and the difference, and outputs the coring output The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus . Wherein at least one of the functions of the low-pass filter or the coring processing circuit, according to any one of claims 1, characterized in that a switch which is switched on and disabling effective until 6 Image processing device. The image display apparatus having an image processing apparatus according to any one of claims 1 to 7.

Images