KR100213109B1 - Circuit for improving picture quality by using noise reduction and histogram equalization and method thereof - Google Patents

Abstract

The circuit of the present invention detects a noise of an impulse component existing in an input video signal, and a noise reduction unit that outputs a noise-removed signal by tripping the detected impulse noise and a noise-removed signal, And a histogram equalizer for mapping the noise-free signal to a new gray level according to a cumulative distribution function value to output an equalized signal. The input signal is firstly removed, By histogram equalization of the signal, the contrast can be improved, and at the same time, the noise can be suppressed from being amplified and the image quality can be improved.

Description

Image Quality Improvement Conference using Noise Reduction and Histogram Equalization

FIG. 1 is a block diagram of an image quality improvement circuit according to the present invention.

FIG. 2 is a detailed block diagram of the picture quality improvement circuit shown in FIG.

The present invention relates to a picture quality improvement circuit using noise reduction and histogram equalization, and more particularly to a circuit and method for improving image quality by histogram equalizing an image obtained by removing noise from an input image after noise is removed .

In order to meet the demand for high quality image display devices such as TV sets and the like, image quality improvement techniques have been used in video processing fields such as TV signals and VCR signals, A histogram equalization technique based on mapping to a gray level on an image plane has emerged.

The histogram of the gray level provides an overall depiction of the appearance of the image. A properly adjusted gray level for a given image improves the appearance or contrast of the image.

Among many methods for improving the contrast, histogram equalization, which is a method of improving the contrast of a given image according to the distribution of images in the image, is most widely known, which is disclosed in the following paragraphs [1] and [2]: [1] JS Lim [2] RCGonzalez and P. Wints, Digital Imaging Processing, Addison-Wesley, Reading, Nassachusetts, 1977. [2] Two-Dimensional Signal and Image Processing, Prentice Hall, Englewood Cliffs,

Also useful applications of the histogram equalization method including medical image processing and radar image processing are disclosed in the following documents [3], [4]: [3] J. Zimmerman, S. Pizer, E. Staab, E. Perry, W. McCartney. and B.Brenton, 'Adaptive histogram equalization for contrast enhancement,' 'IEEE Trans. Medical Imaging, pp.3O4-312, Dec.1988, [4] Y. Li, W. Wang, and DYYu , Application of adaptive histogram equalization to x-ray chest image, '' Proc. of the SPIE, pp. 513-514, vol.1.2321, 1994.

In general, histogram equalization has the effect of stretching the dynamic range, so that histogram equalization flattenes the gray distribution of the resulting image and consequently improves the contrast of the image.

This characteristic of the well-known histogram equalization may be a drawback in practical cases. That is, due to the characteristic of improving the contrast at the time of increasing the dynamic range, when the noise (impulsive noise) included in the video signal is equalized to the histogram, the noise component is also improved and the noise is amplified. There was a problem.

Accordingly, it is an object of the present invention to provide an image quality improvement circuit that removes noise included in an input video signal and then histograms the signal from which noise has been removed to output an equalized signal.

It is another object of the present invention to provide a picture quality improvement method for eliminating noise included in an input video signal and then histogram equalizing the noise-free signal.

In order to achieve the above object, an image quality improvement circuit according to the present invention includes a histogram equalizing circuit for adjusting a gray level of a video signal expressed by a predetermined number of levels to improve contrast, A noise reduction means for detecting a noise of an impulse component existing in a video signal and trimming the detected impulse noise to output a noise-free signal; and a noise reduction means for inputting the noise-removed signal and accumulating And histogram equalizing means for calculating a distribution function value and outputting the equalized signal by mapping the noise canceled signal to a new gray level according to the cumulative distribution function value.

According to another aspect of the present invention, there is provided a method for improving image quality using histogram equalization for improving contrast by adjusting a gray level of a video signal expressed by a predetermined number of levels : (a) detecting noise of an impulse component existing in an input video signal, trimming the detected impulse noise, and outputting a noise-canceled signal; And (b) receiving the noise-free signal, calculating an accumulated distribution function value based on a gray-level distribution in units of an image, mapping the noise-removed signal to a new gray level according to the cumulative distribution function value, And outputting a signal.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of a picture quality improvement circuit using noise reduction and histogram equalization and a method thereof will be described with reference to the accompanying drawings.

FIG. 1 is a block diagram of an image quality improvement circuit according to the present invention.

The circuit of the present invention includes a noise reduction unit 100 for detecting impulse noise included in an input video signal and trimming the detected impulse noise and a histogram equalizer 200 for histogram equalizing the noise- ).

FIG. 2 is a detailed block diagram of the picture quality improvement circuit shown in FIG. 1, in which the operation of the circuit shown in FIG.

2, the window generator 102 of the noise reduction unit 100 outputs a plurality of samples (here, two samples) of windows W _L and W _M and an input sample Y [i] [j] do. Here, the size of the first window W _L is 5 × 5, and the size of the second window W _M is 3 × 3.

Further, the window generator 102 outputs the noise canceled signal Y _N [i] [j] fed back from the selector 116 in the repeated noise reduction mode according to the repetitive / non-repetitive noise reduction mode control signal S Replace with input samples.

The first mean and variance calculator 104 receives each sample of the re-one window shame generated by the window generator 102 and calculates an average sample value A (W _L ) of the first window W _L by equation (1) _L ), and calculates an average (D _L ) of absolute deviations of the samples of the first window (W _L ) by the equation (2).

The first ideal value detector 106 compares the absolute difference between the input sample Y [i] [j] and the average sample value A _L of the first window W _L by the absolute value of the absolute deviation multiplied by the predetermined parameter k If it is greater than the average (kD _L ) , It is determined that an impulse component is included in the input sample (Y [i] [j]) and a first abnormal value detection signal is output.

The second average and deviation calculator 108 receives each sample of the second window W _M generated by the window generator 102 and calculates an average sample value of the second window W _M A _M ), and calculates an average (D _M ) of absolute deviations of the samples of the second window (W _M ) by the equation (4).

In the second outlier detector 110, the absolute difference between the input sample (Y [i] [j]) and the average sample value (A _M ) of the second window (W _M) If it is greater than the average (kD _M ) , It is determined that the impulse component is included in the input sample (Y [i] [j]) and the second outlier detection signal is output.

The selection control signal generator 112 selects the output of the tri-averaging filter 114 when all of the first and second outlier detection signals generated by the first and second outliers 106 and 110 are generated, (Y [i] [j]) output from the adder 102 as it is.

On the other hand, in the trimming mean filter 114, the sample of the second window W _M generated in the window generator 102 is trimmed by the equation (5). In the embodiment of the present invention, the second window W _M is used as the trimming window T _M , but the size of the trimming window can be varied.

here,

ego, If It now indicates an outlier in two windows (W _m ).

Also,

, Which represents the number of untreated samples.

The degree of trimming is controlled by the parameter k. For example, if k = ∞, there is no trimming sample, and if k = 0, all samples in the window are trimmed. Therefore, various noise cancellation characteristics can be obtained by appropriately selecting this parameter depending on the channel condition.

The selector 116 selects the output y of the trimmed mean filter 114 only when an abnormal value is detected in both the first and second ideal value detectors 106 and 110 according to the selection control signal generated by the selection control signal generator 112 select _tr [i] [j]) and if not, to as by-pass the input samples (Y [i] [j]) that is output from the window generator 102, the signal signal (Y _N [i] removing noise [ j].

In addition, when the video signal interlaced in the intra-field mode is used, the horizontal filter 118, the absolute value circuit 120, and the comparator 122 may be further configured.

The horizontal filter 118 calculates the horizontal volume H by calculating a center vector of the first window W _L generated by the window generator 112 and a predetermined weight vector.

The absolute value circuit 120 indicated by ABS calculates the absolute value of the horizontal amount H calculated by the horizontal filter 118.

The comparator 122 compares the water level H output from the absolute value circuit 120 with a predetermined constant T and generates a comparison control signal when the water level H is larger than a constant T. [

The selector 116 selects the _trimmed output _ytr [i] from the trimming mean filter 114 according to the selection control signal generated by the selection control signal generator 112 and the comparison control signal generated by the comparator 122, j] or an input sample Y [i] [j]. That is, in the selector 116, the selection control signal indicating the case where an abnormal value is detected in both the first and second abnormal value detectors 106 and 110 and the horizontal control signal H calculated in the horizontal filter 118 are smaller than the predetermined constant T the only when the comparison control signal representing a larger when the input select tree mid-output _{(y tr [i] [j} ]) and if not, select the input samples (Y [i] [j] ) to the noise cancellation signal ( Y _N [i] [j]).

On the other hand, the configuration of the histogram equalizer 200 is implemented by, for example, a typical histogram equalization technique. However, the present invention can be applied to any configuration as long as the histogram equalizer 200 can improve the contrast by using the histogram.

The frame histogram calculator 202 of the histogram equalizer 200 inputs the noise canceled signal Y _N [i] [j] = X _k ) output from the noise reduction unit 100 on a screen basis to obtain a gray level distribution And calculates a probability density function p (X _k ) expressed by equation (8) based on the gray level distribution obtained. The screen unit could be a field, but here it is a frame.

Here, n _k represents the number of times the gray level (X _k ) appears in the image {X}, and n represents the total number of samples in the image {X}.

The CDF calculator 204 receives the probability density function p (X _k ) output from the frame histogram calculator 202 and calculates the cumulative distribution function c (X _k ) using equation (9).

The CDF memory 206 updates the cumulative distribution function value c (X _k ) calculated by the CDF calculator 204 on a frame-by-frame basis in accordance with the synchronization signal SYNC, (c (X _k )) to the mapper 208. Here, the sync signal is a field sync signal if the picture unit is a field, the frame sync signal is a frame, and the CDF memory 206 is used as a buffer.

Mapper 208, the cumulative distribution function value (c (X _k)) to input the read and noise removal from the CDF memory 206 corresponding to the noise reduction of the input signal (X _k) output from the noise reduction unit 100 And maps to the gray level from 0 to X _L-1 for the image {X _k }.

The noise canceled signal X _k output from the noise reduction unit 100 is delayed by one frame so as to input the input image signal of the same frame as the cumulative distribution function value output from the CDF calculator 204 to the retarder 208 A frame memory can be further configured.

The present invention can be applied to a wide range of fields related to image quality improvement of a video signal. That is, it can be applied to broadcasting equipment, radar signal processing system, medical engineering, home appliances, and the like.

As described above, the present invention improves the contrast by suppressing the noise of the input signal first and histogram equalizing the noise-removed signal, and at the same time, suppressing amplification of noise and improving the image quality.

Claims (16)

The image quality enhancement circuit using histogram equalization for adjusting the gray level of a video signal expressed by a predetermined number of levels to improve contrast; A noise reduction unit that detects noise of an impulse component existing in an input video signal and trims the detected impulse noise to output a noise-removed signal; And a noise reduction unit for calculating a cumulative distribution function value based on a gray level distribution of a picture unit by inputting the noise canceled signal and mapping the noise removed signal to a new gray level according to the cumulative distribution function value to output an equalized signal And a histogram equalizing means for performing histogram equalization. The apparatus of claim 1, wherein the noise reduction means comprises: window generating means for generating a plurality of windows of different predetermined sizes including input samples; An abnormal value detection means for detecting an average of absolute values of samples and an average value of samples of each window to detect that the input sample is abnormal; Selection control signal generating means for generating a selection control signal when the input sample is detected as an abnormal value by the abnormal value detecting means; Trimming means for outputting a trimming signal using a trimming window of a predetermined size; And selection means for outputting the trimmed signal as an output signal if an impulse component exists in the input sample according to the selection control signal and otherwise outputting an input sample as an output signal. . The picture quality improvement method according to claim 2, wherein the window generating means replaces the output signal with the input sample in the repeated noise reduction mode according to the repetitive / non-repetitive noise reduction mode. 2. The apparatus of claim 2, further comprising: means for computing a horizontal magnitude by computing a predetermined weight vector and a predetermined horizontal vector including the input sample; And comparison means for comparing the detected watermark with a predetermined parameter to generate a comparison control signal. 5. The apparatus of claim 4, wherein the selecting means selects the tripled signal as an output signal if an impulse component exists in the input sample according to the selection control signal and the comparison control signal and the detected horizontal amount is larger than a predetermined parameter. And if not, outputs the input sample as an output signal as it is. 3. The noise reduction circuit of claim 2, wherein the window generating means generates a first window and a second window of different sizes including an input sample. The method of claim 6, wherein the outlier detection means comprises an impulse component in the input sample if the absolute difference between the input sample and the average sample value of the first window is greater than the average of the absolute deviation of the first window multiplied by the predetermined parameter And outputs a first ideal value detection signal; And determines that the impulse component is included in the input sample if the absolute difference between the input sample and the average sample value of the second window is larger than the average of absolute deviations of the second window multiplied by the predetermined parameter, And outputting the second ideal value detector. The apparatus of claim 1, wherein the histogram equalization means comprises: first calculation means for inputting the noise-free signal, calculating a gray level distribution on a screen basis, and outputting a probability density function value; Second calculation means for calculating a cumulative distribution function based on the probability density function value; And mapping means for mapping the noise-accumulated signal to a gray level according to a calculated cumulative distribution function value. 9. The apparatus according to claim 8, further comprising: a delay unit for delaying the noise canceled signal for one screen to input a signal of the same screen as the cumulative distribution function value calculated by the second calculation unit to the mapping unit, And further comprising a memory. 9. The image processing apparatus according to claim 8, further comprising a buffer for updating the cumulative distribution function value calculated by the second calculation means on a screen basis and outputting the accumulated cumulative distribution function value for one screen before the update to the mapping means, The image is displayed on the screen. A method for enhancing picture quality using histogram equalization that adjusts a gray level of a video signal expressed by a predetermined number of levels to improve contrast, comprising: (a) detecting a noise of an impulse component existing in an input video signal; And outputting a noise canceled signal by tripping the detected impulse noise; And (b) receiving the noise-free signal, calculating an accumulated distribution function value based on a gray-level distribution in units of an image, mapping the noise-removed signal to a new gray level according to the cumulative distribution function value, And outputting a signal. 12. The method of claim 11, wherein step (a) comprises: generating a plurality of windows of different sizes including an input signal; (a2) obtaining an average of absolute values of the samples and a mean value of the samples for each window; (a3) detecting whether or not an impulse component is present in the input signal by using an average of absolute values of the samples and the average values of the sampled windows; ; And (a4) if there is an impulse component in the input signal, outputting the samples in the trimming window having a predetermined size, and otherwise, bypassing the input signal as it is. 13. The method of claim 12, wherein, in step (a3), when an impulse component is present in at least two windows, it is determined that an impulse component exists in the input signal. 14. The method of claim 12, wherein the step (a4) comprises the steps of: (a4-1) calculating a horizontal vector having a predetermined size including the input signal and a predetermined vector to calculate a horizontal amount; And (a4-2) if an impulse component exists in the input signal and the detected horizontal amount is larger than the predetermined parameter, the trimming unit samples trimming in the trimming window of a predetermined size and outputs the trimming result as an output signal. Otherwise, the input signal is bypassed The method of claim 1, The method of claim 12, wherein the step (a4) further comprises: (a4-11) trimming the samples in one window and outputting the trimmed signal as an output signal when an impulse component exists in the input sample, Otherwise outputting the input sample as it is as an output signal; And (a4-12) replacing the input sample with the output signal, moving to the next input sample, and feeding back to the step (a). 12. The method of claim 11, wherein the step (b) comprises: (bl) calculating a gray level distribution on a screen basis by inputting the noise canceled signal and outputting a probability density function value; (b2) calculating a cumulative distribution function value based on the probability density function value; And (b3) mapping the noise-free signal to a gray level according to the calculated cumulative distribution function value.