JP2014171042A - Image signal processing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing apparatus capable of suppressing spread of width of a shoot component produced in contour emphasis processing.SOLUTION: An image processing apparatus includes: a BPF 5 with a mixing function that is formed by a plurality of taps for extracting a frequency component in a specific band corresponding to an edge portion from an input image signal; an edge detecting unit 2 for detecting pixels corresponding to an edge from the image signal; a strong edge extracting unit 3 for extracting a strong edge by comparing a signal level of the pixel detected in the edge detecting unit 2 with a predetermined edge determining threshold value to determine a strength level of the edge of the pixel extracted as the strong edge using the edge determining threshold value. In accordance with strength of the edge of the pixel input to a center tap of the BPF 5 with the mixing function, correction processing to an output value of a peripheral tap disposed at a position apart from the center tap by a predetermined distance or more is carried out using an output value of the center tap.

Description

  The present embodiment relates to an image signal processing apparatus.

  In recent years, the screen resolution of television displays has been increasing, and full high-definition and 4k models that are four times as high as full high-definition have been put into practical use. On the other hand, the image signal to be displayed remains compatible with high-definition and full high-definition formats. Therefore, when these image signals are reproduced on a 4k television, it is necessary to increase the image quality using a super-resolution technique or the like.

  Conventionally, contour enhancement processing for enhancing the contour of an image signal has been performed as one of the processing for improving the resolution of an image. The contour emphasis process is a process performed to improve the blur of the image accompanying the image enlargement process or to obtain further sharpness. In general, a high-frequency component of an image signal is extracted using a high-pass filter (hereinafter referred to as HPF) or a band-pass filter (hereinafter referred to as BPF), and is added to the image signal to perform edge enhancement.

  In such contour emphasis processing, in order to further improve the image quality, it is necessary to perform contour emphasis using not only the high frequency band but also the frequency components from the middle frequency range to the low frequency range. In this case, it is necessary to increase the number of taps of the filter and perform the calculation using the signal of the pixel at a position away from the pixel to be processed. However, if there is a strong edge in the tap away from the center tap, or if the level difference between the pixel signal value of the center tap and the pixel signal value of the tap away from the center tap is large, this There is a problem that the width of the chute component at the edge portion becomes wider due to the difference in level or the like.

JP 2007-221821 A JP 2005-260517 A JP 2003-338949 A

  Therefore, the present embodiment has been made in view of the above points, and an object thereof is to provide an image signal processing device capable of suppressing the spread of the width of a shoot component generated in the contour enhancement processing. .

  The image signal processing apparatus according to the present embodiment includes a first digital filter including a plurality of taps that extract a frequency component of a specific band corresponding to a contour portion of the image from the input image signal, and the image signal. An edge detection unit that detects a pixel corresponding to the contour portion, and extracts a strong edge by comparing a signal level of the pixel detected by the edge detection unit with a predetermined edge determination threshold, and extracts the strong edge A strong edge extraction unit that determines an edge strength level of the pixel using the edge determination threshold, and is located at the center of the plurality of taps of the first digital filter Depending on the strength of the edge of the pixel input to the center tap, which is a tap, the output of peripheral taps that are arranged at a predetermined distance or more away from the center tap. Value for, and performs the correction processing using the output value of the center tap.

1 is a block diagram illustrating a configuration of an image signal processing device 1 according to a first embodiment. The figure explaining the edge determination result and expansion result in the strong edge extraction part 3 and the edge expansion part 4. FIG. The block diagram explaining the structure of the mixer 502 in BPF5 with a mixing function. A comparison result of the edge determination result and a difference value and the signal word level threshold, the table showing the relationship between the tap outputs coefficient k _t and the center output coefficient k _c. The figure explaining the comparison of the pixel signal value before and behind correction in the mixing circuit 53. FIG. The block diagram explaining the structure of image signal processing apparatus 1 'concerning 2nd Embodiment. The block diagram explaining the structure of BPF11 with a replacement function. Table showing edge determination result, the relationship between BPF10 output coefficient _{k a} and BPF11 output coefficient _{k b.} The block diagram explaining the structure of the image signal processing apparatus 21 concerning 3rd Embodiment. Table for explaining an example of the value of the gain adjustment coefficient k _g for tap position signal is input pixel intensity edges. Diagram for explaining the change in the value of the gain adjustment coefficient k _g for tap position the intensity signal of the edge pixels are input. The block diagram explaining the structure of BPF5 with a mixing function concerning 1st Embodiment.

  Hereinafter, embodiments will be described with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram illustrating a configuration of an image signal processing apparatus 1 according to the first embodiment. As shown in FIG. 1, the image signal processing apparatus 1 according to the present embodiment includes an edge detection unit 2 that extracts an edge portion from an input image signal, and a strong edge that further extracts a strong edge portion from the detected edge portion. The extraction unit 3 and an edge extension unit 4 that extends a strong edge part to a peripheral region based on a predetermined condition.

  In addition, the image signal processing apparatus 1 of the present embodiment extracts a frequency component of a predetermined band from the input image signal, and moves away from the center tap based on information on the strong edge portion output from the edge extension unit 4. A BPF 5 with a mixing function as a first digital filter that performs a predetermined correction process on a tap value at a position, a coring unit 6 that removes a low-level signal component to reduce a noise component, and a predetermined Output from the limiter unit 8, a gain unit 7 that adjusts the value by multiplying the gain value, a limiter unit 8 that removes a signal component that exceeds a predetermined level and removes an excessively amplified and emphasized component, and a limiter unit 8. An adder 9 that generates a contour-enhanced image signal by adding the pixel signal and the original input image signal is also provided.

Hereinafter, each component of the image signal processing apparatus 1 shown in FIG. 1 will be described individually. First, for the edge detection unit 2, a general edge detection filter configured by, for example, a Sobel filter or a Laplacian filter is used. The edge detection result is output to the strong edge extraction unit 3. In addition to the edge detection result, one or more edge determination threshold _Te is input to the strong edge extraction unit 3. Here, the edge determination threshold value _Te is a threshold value used for determining the level of edge strength of a pixel detected as an edge.

A method for determining a strong edge in the strong edge extraction unit 3 will be described with reference to an example shown in FIG. FIG. 2 is a diagram illustrating the edge determination result and the extension result in the strong edge extraction unit 3 and the edge extension unit 4. In the upper graph of FIG. 2, the horizontal axis indicates the arrangement of pixels in the horizontal direction, and the vertical axis indicates the signal level in each pixel. T _e1 and T _e2 shown on the vertical axis are edge determination threshold _values . Therefore, in the example shown in FIG. 2, two edge determination threshold values are set. Hereinafter, reference to an example shown in FIG. 2, a case of performing the edge determination for the nine pixels of S ₉ from S ₁ (and expansion).

The strong edge extraction unit 3 compares the input edge determination threshold _Te with the signal level of each pixel, extracts a strong edge, determines the strength level, and outputs the result as an edge determination result. Specifically, when the signal level of the pixel is lower than the edge determination threshold value _Te2 , the edge determination result is “0”, and it is determined that the edge is not a strong edge. Further, if less than the higher edge determination threshold T _e1 the edge determination threshold T _e2 is a signal level of the pixel, the intensity level of is determined to be a strong edge of "2". Further, when the signal level of the pixel is higher than the edge determination threshold value _Te1, it is determined that the pixel is a strong edge having a strength level of “3”. Therefore, the higher the numerical value of the edge determination result, the higher the edge strength of the pixel.

Among the _{S 1} shown in FIG. 2 of 9 pixels of _{S 9,} 4 single pixel of _{S 1, S 7, S 8} , S 9 , the signal level is lower than the edge determination threshold _{T e2.} Therefore, the edge determination result of these pixels is “0”, and it is determined that the edge is not a strong edge. _Furthermore, three pixels of S _2, S 3, _{S 5,} the signal level is lower than the edge determination threshold _{T e2} higher than the edge determination threshold _{T e1.} Therefore, it is determined that these pixels are strong edges having a strength level of “2”. Furthermore, the signal level of the two pixels S ₄ and S ₅ is higher than the edge determination threshold value T _e1 . Accordingly, these pixels are determined to be strong edges having a strength level of “3”. In FIG. 2, each edge determination result is shown below each pixel on the horizontal axis of the graph.

The edge extension unit 4 extends the strong edge portion using the edge determination result output from the strong edge extraction unit 3. Specifically, when the strong edge extraction unit 3 determines that the edge determination result is “0” and is not a strong edge, and the adjacent pixel is determined to be a strong edge, the pixel Is used to replace the strength level with “1”. In the example illustrated in FIG. 2, of the four pixels S ₁ , S ₇ , S ₈ , and S ₉ that are determined not to be strong edges, two pixels S ₁ and S ₇ are determined to be strong edges, respectively. It is adjacent to the pixels S ₂ and S ₆ . Therefore, the intensity level of the two pixels S ₁ and S ₇ is replaced from “0” to “1” to obtain a strong edge. In FIG. 2, the expansion result for each pixel is shown below the edge determination result. The extended edge determination result in the edge extension unit 4 is output to the mixed function BPF 5.

It should be noted that the level of intensity of a pixel that is extended to a strong edge in the edge extension unit 4 is not limited to “1”, and may be, for example, a level obtained by subtracting 1 from the intensity level of an adjacent pixel. Good. In this case, in the example illustrated in FIG. 2, for the pixel S ₁ adjacent to the pixel S ₂ having the strength level “2”, the strength level is replaced with “1”, and the strength level is “1”. 3 "for the pixel S ₇ and the adjacent pixel S ₆ is the level of strength" is replaced by 2 ".

  Next, the detailed configuration of the mixed function BPF 5 will be described with reference to FIGS. 12 and 3. FIG. 12 is a block diagram illustrating the configuration of the BPF 5 with a mixing function according to the first embodiment. The BPF 5 with a mixing function includes, for example, an FIR filter (shift register 501, multiplier 503, adder) composed of N taps and a plurality of mixers 502. A mixer 502 is connected to the tap output away from the center, and the output of the center tap and the tap output away from the center can be mixed.

  FIG. 3 is a block diagram illustrating the configuration of the mixer 502 in the BPF 5 with a mixing function. As shown in FIG. 3, the mixer 502 includes a difference calculation circuit 51 that calculates a difference between an output value from the center tap and an output value from the N taps, and a difference value calculated by the difference calculation circuit 51. Based on the comparison circuit 52 that compares a preset threshold value (signal level threshold value), the output from the comparison circuit 52, and the extended edge determination result output from the edge extension unit 4, the output value of each tap And a mixing circuit 53 as a correction circuit that mixes the output value of the center tap with a set ratio.

  In the image signal processing apparatus 1 according to the present embodiment, the contour enhancement processing is performed using also the frequency components from the middle range to the low range, and therefore it is necessary to perform the calculation using the signal of the pixel at a position away from the pixel to be processed. . Therefore, it is necessary to take in and calculate the signal of the pixels arranged in the horizontal direction from the pixel to be processed to a position about 5 to 7 pixels away from the front and back. For example, when the edge enhancement process is performed using pixel signals for seven pixels before and after the edge enhancement process target pixel in the horizontal direction, a 15-tap BPF 5 is used. At this time, the pixel signal of the contour emphasis target pixel is input to the center tap, and the pixel signals for the seven pixels before and after the contour target pixel are input to the front and rear taps. First, the pixel signal output from each tap and the pixel signal output from the center tap are input to the difference calculation circuit 51, and the difference is calculated. Accordingly, the difference calculation circuit 51 outputs a difference value corresponding to the number of taps.

  A difference value between the pixel signal of each tap and the pixel signal of the center tap is input to the comparison circuit 52. A signal level threshold value set in advance to a predetermined value is also input to the comparison circuit 52, and is compared with the difference value calculated by the difference calculation circuit 51. That is, for each of the N taps, the difference value is compared with the signal level threshold value, and the comparison result is output to the mixing circuit 53.

  Based on the extended edge determination result for the center tap pixel among the extended edge determination results input from the edge extension unit 4 and the comparison result input from the comparison circuit 52, the mixing circuit 53 Correction of the pixel signal value of the tap is performed. The correction of the pixel signal value may be performed for all taps, but may be performed only for a tap located at a certain distance or more from the center tap (for example, a tap located at a distance of 3 taps or more from the center tap). For example, when the BPF 5 is 15 taps, if correction is performed only for taps that are located 3 taps or more away from the center tap, the number of taps to be corrected is 10 taps. The correction of the pixel signal value is performed using the following equation (1).

Output value of the tap × tap output coefficient k _t
+ Output value of center tap × center output coefficient k _c (1)
The tap output coefficient k _t and the center output coefficient k _c are set according to the extended edge determination result of the pixel signal of the center tap and the comparison result between the difference value and the signal level threshold value. An example of the tap output coefficient k _t and the center output coefficient k _c is shown in FIG. Figure 4 is a table showing a comparison result of the edge determination result and the difference value and the signal level threshold, the relationship between the tap outputs coefficient k _t and the center output coefficient k _c.

As shown in FIG. 4, when the pixel of the center tap is a strong edge and the level is determined to be “3”, 1 is set to the tap output coefficient k _t and 0 is set to the center output coefficient k _c. Is done. That is, in this case, the pixel signal values output from all the taps are used as they are, and no correction is performed. Next, when it is determined that the pixel of the center tap is a strong edge and the level is “2”, 0.25 is set for the tap output coefficient k _t and 0.75 is set for the center output coefficient k _c. Is done. These set values are substituted into equation (1), and the pixel signal value is corrected for the correction target tap. That is, in this case, the pixel signal value output from the tap to be corrected and the pixel signal value output from the center tap are mixed at a ratio of 0.25: 0.75 and replaced as the pixel signal value of the tap. Is done.

If it is determined that the center tap pixel is a strong edge and the level is “1”, the tap output coefficient k _t is set to 0.125, and the center output coefficient k _c is set to 0.875. The That is, in this case, the pixel signal value output from the correction target tap and the pixel signal value output from the center tap are mixed at a ratio of 0.125: 0.875 and replaced as the pixel signal value of the tap. Is done.

Furthermore, when it is determined that the pixel of the center tap is not a strong edge (level “0”) and the difference value is larger than the signal level threshold, 0 is set for the tap output coefficient k _t and 1 is set for the center output coefficient k _c. Is set. That is, in this case, the pixel signal value output from all the taps is replaced with the pixel signal value of the center tap. Further, when it is determined that the center tap pixel is not a strong edge (level “0”) and the difference value is smaller than the signal level threshold, the tap output coefficient k _t is 1 and the center output coefficient k _c is 0. Is set. That is, in this case, the pixel signal values output from all the taps are used as they are, and no correction is performed.

The pixel signal value correction processing in the mixing circuit 53 will be described using a specific example shown in FIG. FIG. 5 is a diagram for explaining comparison of pixel signal values before and after correction in the mixing circuit 53. As shown in FIG. 5, when the edge enhancement process is performed using pixel signals in a range of 7 pixels before and after the pixel to be processed, taps P ₋₇ , P ₋₆ , P ₋₅ , P ₋₄ , P ₋₃ , _{P -2, P -1, P 0} , P 1, P 2, P 3, P 4, P 5, composed of 15 taps _P 6, P ₇ BPF15 is used. The tap P ₀ is a center position, is inputted pixel signal of the pixel to be processed, the tap P _-7 from the tap P _-1 is placed after 1-7 pixels in the horizontal direction from the pixel to be processed A pixel signal of the pixel is input. Further, the tap P ₇ from the tap P ₁ pixel signals of the pixels arranged in the 1-7 pixels before the horizontal direction from the pixel to be processed is input.

  FIG. 5 shows the tap output value, the difference between the center tap output value and the tap output value, and the tap output value (correction value) after the replacement processing by the mixing circuit 53 for each tap. Yes.

In the example shown in FIG. 5, correction is performed only for taps that are located at a position 3 or more taps away from the center tap. Accordingly, as shown in FIG. 5, the tap _{P -2,} P _-1, 5 taps P _0, P 1, _{P 2} is set as the normal region is not corrected, the rest of the tap _{P -7, P -6 , P -5, P -4, P} -3, P 3, P 4, P 5, 10 taps P 6, _{P 7} is set as replacement area to be corrected.

First, the output values of the taps P _{−7 to} P _{7 and} the output value of the center tap P ₀ are input to the difference calculation circuit 51. The difference calculation result of each tap in the difference calculation circuit 51 is as described in the row of “difference from center tap” in FIG. That is, the tap _{P -7} is the output value is 64, the output value of the center tap _{P 0} is 190, the difference becomes 190-64 = 126. Since the output value of the tap P- ₆ is 64, the difference is 190−64 = 126. Since the output value of the tap P- ₅ is 80, the difference is 190-80 = 110. Since the output value of the tap P- ₄ is 120, the difference is 190-120 = 70. Since the output value of the tap P- ₃ is 180, the difference is 190-180 = 10. Since the remaining 10 taps _P -2 to P ₇ output value is 190, the difference becomes 190-190 = 0. These difference calculation results are output to the comparison circuit 52.

In the comparison circuit 52, a difference calculation result of each tap P _-7 to P ₇ input from the difference calculation circuit 31, and a signal level threshold are compared. For example, as illustrated in FIG. 5, when the signal level threshold is set to 50, it is determined that the difference calculation result of the four taps P _{−7 to} P ₋₄ is larger than the signal level threshold. Eleven taps of taps P _{−3 to} P ₇ are determined to have a smaller difference calculation result than the signal level threshold. These determination results in the comparison circuit 52 are output to the mixing circuit 53.

In the mixing circuit 53, based on the determination result of each tap P _-7 to P ₇ which is input from the comparing circuit 52, the extended edge determination result is input from the edge extension unit 4, the tap output coefficient k _t and the center The value of the output coefficient k _c is determined. For example, as illustrated in FIG. 5, when the edge determination result of the center tap extended is level “0”, taps P _{− 7 to} P ₋₄ determined that the difference calculation result is larger than the signal level threshold. 4 taps are determined as tap output coefficient k _t = 0 and center output coefficient k _c = 1 from the table of FIG. Further, 11 taps of the determined taps P _{-3 to} P ₇ whose difference calculation result is smaller than the signal level threshold value are similarly output from the table of FIG. 4 from the tap output coefficient k _t = 1 and the center output coefficient k _c = 0 is determined.

Of the 15 taps, set in the area replacement to be corrected, tap _{P -7, P -6, P -5} , P -4, P -3, P 3, P 4, P 5, P ₆ and the values of the tap output coefficient k _t and the center output coefficient k _c determined for 10 taps of P ₇ are substituted into the above equation (1) to correct the pixel value. Specifically, the output values after correction of the four taps of the taps P _{−7 to} P ₋₄ are (output value of each tap) × 0 + 190 × 1 = 190. That is, the output values of these four taps are replaced with the output values of the center tap. Further, the corrected output values of the six taps of taps P ₋₃ , P ₃ , P ₄ , P ₅ , P ₆ , and P ₇ are (output value of each tap) × 1 + 190 × 0 = (each tap Output value). Therefore, the output values of these six taps are used as they are.

  With the above processing, the value shown in the “tap output after replacement” line in FIG. 5 is output to the multiplier 503 of the BPF 5 with a mixing function as the mixing circuit 53 (output of the mixer 502) as an output value after correction of each tap. Output and calculated. The calculation result is output to the coring unit 6.

  In the coring unit 6, the low-level signal component is removed from the image signal obtained by extracting the frequency component of the predetermined band output from the BPF 5 with a mixing function, and the noise component is reduced. The gain unit 7 adjusts the value by multiplying the image signal after noise removal output from the coring unit 6 by a predetermined gain value. In the limiter unit 8, a signal component having a level equal to or higher than a predetermined input level is limited to a predetermined level from the adjusted image signal output from the gain unit 7 and is prevented from being excessively amplified or emphasized. Finally, in the adding unit 9, the image signal output from the limiter unit 8 and the original input image signal are added to obtain a desired contour-enhanced image signal.

  As described above, according to the present embodiment, when performing edge enhancement processing of an image signal using not only a high frequency but also a frequency band from a middle frequency to a low frequency, that is, from a video signal using a BPF having a large number of taps. When contour enhancement processing is performed by extracting a predetermined frequency band component and adding it to the original input image signal, the output of taps located away from the center tap according to the level of edge strength of the center tap Correct the value. That is, when the center tap is not an edge (or the edge strength level is low), the output value is set so that the level difference between the output value of the tap away from the center tap and the output value of the center tap is small. If the center tap is a strong edge, the output values of all the taps are used without correction. As a result, the expansion of the width of the chute component due to the output value of the tap located at a position away from the center tap can be prevented.

  In the above-described example, the edge extension unit 4 is provided that extends the strong edge extracted by the strong edge extraction unit 3 to the peripheral region (performs smoothing so that the edge does not become sharp). The edge determination result of the strong edge extraction unit 3 may be input to the BPF 5 with a mixing function without providing the.

Further, in the above-described example, leveling of strong edges is performed using two edge determination threshold values, but one edge determination threshold value or three or more edge determination threshold values may be set. Furthermore, the values of the tap output coefficient k _t and the center output coefficient k _c are not limited to the values shown in FIG. 4, and optimum values can be set according to the characteristics of the input image signal. .

  In addition, taps for which output values are corrected in the mixed function BPF 5 are not limited to taps that are 3 taps or more away from the center tap, and all taps may be corrected, or conversely, 4 taps or more apart. The number of output value correction target taps such as taps may be reduced.

  Furthermore, in the above example, BPF is used to extract a desired frequency band component, but other digital filters such as HPF and LPF (low pass filter) may be used.

(Second Embodiment)
In the image signal processing apparatus 1 according to the first embodiment described above, when the tap output value is corrected in the BPF 5 with the mixing function, if the center tap is not a strong edge, the tap output value and the center tap output value are In accordance with the comparison result between the difference value and the signal level threshold value, it has been determined whether or not to replace with the output value of the center tap. In this embodiment, a general frequency band component is extracted. Two BPFs of BPF having a function of replacing the output value of the tap to be corrected with the output value of the center tap according to the comparison result between the BPF, the difference value between the tap output value and the center tap output value, and the signal level threshold are provided. The difference is that the output values from both BPFs are mixed and output at a set ratio according to the edge strength of the center tap. In the image signal processing apparatus 1 ′ of the present embodiment, the constituent elements other than the BPFs 10 and 11 and the selective mixing unit 12 are the same as those in the first embodiment, and thus the same reference numerals are given and description thereof is omitted.

The configuration of the image signal processing apparatus 1 ′ of this embodiment will be described with reference to FIG. FIG.
It is a block diagram explaining the structure of the image signal processing apparatus 1 'concerning 2nd Embodiment. As shown in FIG. 6, the image signal processing apparatus 1 ′ of the present embodiment includes a general BPF 10 that extracts a desired frequency band component from an input image signal, and the same frequency as the BPF 10 from the input image signal. After extracting the band component, for the tap to be corrected, the output value of the tap is replaced with the output value of the center tap according to the comparison result between the difference value between the tap output value and the center tap output value and the signal level threshold value. The BPF 11 with a replacement function as a second digital filter having a function and the output value of the BPF 10 and the output value of the BPF 11 with a replacement function at a predetermined ratio set according to the edge determination result of the pixel to be subjected to contour enhancement And a selective mixing unit 12 for mixing and outputting. Other components are the same as those of the first embodiment described with reference to FIG.

  The configuration of the BPF 11 with a replacement function will be described with reference to FIG. FIG. 7 is a block diagram for explaining the configuration of the BPF 11 with a replacement function.

The BPF 11 with a replacement function includes an FIR filter having the same number of taps as the taps of the BPF 10, and replacement circuits 112a to 112e and 112k to 112o as many as the number of taps to be corrected.

FIG. 7 shows, as an example, a BPF 11 with a 15-tap replacement function in which 10 taps that are 3 or more taps away from the center tap are correction target taps. In other words, it is composed of 14 delay circuits 111a to 111n, 10 replacement circuits 112a to 112e, 112k to 112o, and 15 multiplication circuits 113a to 113o. Ten replacement circuits 112a to 112e and 112k to 112o are provided one for each of 10 taps that are separated from the center tap by 3 taps or more, and between the output terminal of the delay circuit and the input terminal of the multiplication circuit. Is arranged. (However, since the leftmost tap has no delay circuit, a replacement circuit 112a is disposed between the input terminal of the pixel signal and the input terminal of the multiplication circuit 113a.)
The ten replacement circuits 112a to 112e and 112k to 112o are input with the corresponding tap output value, the signal level threshold value, and the center tap output value. For example, the leftmost replacement circuit 112a has a pixel signal input to the BPF 11 with a replacement function, which is an output value of the leftmost tap, an output value of the delay circuit 111g, which is an output value of the center tap, and a signal level threshold value. Entered.

  In replacement circuits 112a to 112e and 112k to 112o, the difference between the output value of each corresponding tap and the output value of the center tap is calculated. The calculated difference value is compared with a signal level threshold value. When the difference value is smaller than the signal level threshold value, the output values of the corresponding taps are output as they are to the multiplication circuits 113a to 113e and 113k to 113o. On the other hand, when the difference value is larger than the signal level threshold value, the output value of the center tap is output to the multiplication circuits 113a to 113e and 113k to 113o. That is, the replacement circuits 112a to 112e and 112k to 112o have a function of replacing the output value of the tap with the output value of the center tap when the level difference between the tap output value and the center tap output value is large. ing.

  As described above, the BPF 11 with a replacement function extracts and corrects a desired frequency band component from the input image signal, and outputs it to the selection mixing unit 12. The selection mixing unit 12 also receives an output from the BPF 10 and an extended edge determination result output from the edge extension unit 4.

  Based on the extended edge determination result for the center tap pixel among the extended edge determination results input from the edge expansion unit 4, the selection mixing unit 12 outputs the output value from the BPF 10 and the BPF 11 with a replacement function. Selective mixing with the output value is performed. The pixel signal value is corrected using the following equation (2).

The output value of the BPF 10 × BPF 10 outputs the coefficient _{k a}
+ Output value × substituted function BPF 11 BPF 11 outputs the coefficient _{k b} ... formula (2)
BPF10 output coefficient _{k a} and BPF11 output coefficient _{k b} is set according to the extended edge determination result of a pixel signal of the center tap. BPF10 an example of an output coefficient _{k a} and BPF11 output coefficient _{k b} shown in FIG. 8, an edge determination result is a table showing the relationship between BPF10 output coefficient _{k a} and BPF11 output coefficient _{k b.}

As shown in FIG. 8, a pixel is strong edge of the center tap, when it is determined that the level is "3", 0 is set to 1, BPF 11 outputs the coefficient _{k b} is the BPF10 output coefficient _{k a} Is done. That is, in this case, the output value from the BPF 11 with a replacement function is not used, and only the output value from the BPF 10 is selected and output to the coring unit 6. Next, a pixel is strong edge of the center tap, when it is determined that the level is "2", 0.25 to BPF10 output coefficient _{k a,} setting 0.75 to BPF11 output coefficient _{k b} Is done. That is, in this case, the output value from the BPF 10 and the output value of the BPF 11 with a replacement function are mixed at a ratio of 0.25: 0.75 and output to the coring unit 6.

Further, a pixel is strong edge of the center tap, when it is determined that the level is "1", 0.125 to BPF10 output coefficient _{k a,} 0.875 is set in the BPF11 output coefficient _{k b} The That is, in this case, the output value from the BPF 10 and the output value of the BPF 11 with a replacement function are mixed at a ratio of 0.125: 0.875 and output to the coring unit 6.

Furthermore, if it is determined not to be pixel intensity edge center tap (level "0"), 1 is set in the BPF10 output coefficient _{k a} 0, BPF 11 outputs the coefficient _{k b.} That is, in this case, the output value from the BPF 10 is not used, and only the output value from the BPF 11 with a replacement function is selected and output to the coring unit 6.

  As described above, according to this embodiment, two BPFs having the same number of taps and the same frequency characteristics are used, and only one BPF (BPF 11 with a replacement function) is separated from the center tap by a predetermined distance or more (the correction target). If the difference between the output value of the center tap and the output value of the tap is greater than the signal level threshold value, the output value of the tap is replaced with the output value of the center tap and output to the selection mixing unit 12. . The selection mixing unit 12 selects either the output from the BPF 10 or the output from the BPF 11 with a replacement function based on only the (extended) edge determination result of the center tap, or mixes and outputs at a predetermined ratio do it.

  That is, correction processing of an image signal based on two types of determination results (determination result by comparison with a signal level threshold and edge determination result) performed by the BPF 5 with a mixing function of the image signal processing device 1 of the first embodiment. In the image signal processing apparatus 1 ′ of the present embodiment, the correction process based on the determination result based on the comparison with the signal level threshold value is assigned to the BPF 11 with a replacement function, and the correction process based on the edge determination result is assigned to the selection mixing unit 12. Is going. Therefore, similarly to the image signal processing device 1 of the first embodiment, it is possible to prevent the expansion of the width of the shoot component due to the output value of the tap located at a position away from the center tap, and the image signal processing device. Since the 1 ′ component is functionally divided, it is excellent in maintainability and expandability.

  Note that, similarly to the image signal processing device 1 of the first embodiment, the image signal processing device 1 ′ of the present embodiment extends the strong edge extracted by the strong edge extraction unit 3 to the peripheral region (the edge is steep). Although the edge extension unit 4 is provided), the edge determination result of the strong edge extraction unit 3 may be input to the selection mixing unit 12 without providing the edge extension unit 4. .

Further, in the above-described example, leveling of strong edges is performed using two edge determination threshold values, but one edge determination threshold value or three or more edge determination threshold values may be set. Furthermore, the value of BPF10 output coefficient k _a and BPF11 output coefficient k _b is not limited to the values shown in FIG. 8, it is possible to set an optimum value depending on the characteristics of the image signal input .

  In addition, the tap for which the output value is corrected in the BPF 11 with a replacement function is not limited to a tap that is 3 taps or more away from the center tap, and all taps may be corrected, or conversely, 4 taps or more apart. The number of output value correction target taps such as taps may be reduced.

  Furthermore, in the above example, BPF is used to extract a desired frequency band component, but other digital filters such as HPF and LPF (low pass filter) may be used.

(Third embodiment)
The image signal processing apparatuses 1 and 1 'according to the first and second embodiments described above use a multi-tap digital filter in order to increase the output image quality, and not only a high frequency but also a low frequency from a middle frequency. When edge enhancement processing is also performed using frequency components over the band, there is a strong edge in the tap away from the center tap, or the pixel signal value of the tap away from the center tap pixel signal value. In order to prevent the width of the chute component at the edge from becoming wide due to the level difference, correction processing is performed on the output value of each tap of the digital filter. It was. However, when the correction process is performed on the output value of each tap of the digital filter, the circuit scale added for the correction process becomes large, or two kinds of determinations are necessary to determine the conditions for the correction process. There has been a problem that the processing becomes complicated.

  Therefore, an object of the present embodiment is to provide an image signal processing device capable of suppressing the spread of the shoot component generated in the contour enhancement process with a simple process while suppressing an increase in circuit scale. .

  FIG. 9 is a block diagram illustrating the configuration of the image signal processing device 21 according to the third embodiment. As shown in FIG. 9, the image signal processing device 21 of the present embodiment includes an edge detection unit 22 that detects an edge from an input image signal, and a strong edge extraction unit that extracts a strong edge from the detected edges. 23, a strong edge position determination unit 24 that determines which tap position the extracted strong edge corresponds to among the N taps constituting the BPF 26 described later, and a determination result of the strong edge position determination unit 24 Accordingly, a gain adjustment unit 25 for adjusting a gain value used in a gain unit 28 described later is included.

  Further, the image signal processing device 21 of the present embodiment removes noise by removing the BPF 26 composed of N taps, which extracts a frequency component of a predetermined band from an input image signal, and a low-level signal component. A coring unit 27 for reducing components, a gain unit 28 for adjusting a value by multiplying by a predetermined gain value, and a limiter for removing a signal component exceeding a predetermined level to remove an excessively amplified and emphasized component And an adder 30 that adds the pixel signal output from the limiter unit 8 and the original input image signal to generate an edge enhanced image signal.

  Hereinafter, each component of the image signal processing device 21 shown in FIG. 9 will be described individually. First, the edge detection unit 22 is a general edge detection filter configured by, for example, a Sobel filter or a Laplacian filter. The edge detection result is output to the strong edge extraction unit 23. In addition to the edge detection result, an edge determination threshold value is input to the strong edge extraction unit 23. Here, the edge determination threshold value is a threshold value used for extracting a strong edge from the edge of a pixel detected as an edge. If the signal level of a pixel detected as an edge is higher than the edge determination threshold, the pixel is extracted as a strong edge.

  The strong edge position determination unit 24 determines to which tap the pixel signal extracted as a strong edge in the strong edge extraction unit 23 is input from among the N taps constituting the BPF 26. The tap position where the signal of the pixel of the strong edge is input is output to the gain adjustment unit 25.

In the gain adjustment unit 25, the gain value used in the gain unit 28 is adjusted using the in-tap position information of the pixel signal of the strong edge input from the strong edge position determination unit 24. Specifically, it adjusts the gain value by multiplying the gain adjustment coefficient k _g relative set gain value provided from outside. Gain adjustment coefficient k _g is the signal of the pixel intensity edges when located tap center tap or center tap vicinity is set to a large value is set to a smaller value as the distance from the center tap in the opposite.

The relationship between the tap position the intensity signal of the edge of the pixel is input and the gain adjustment coefficient k _g, specifically described with reference to an example shown in FIGS. 10 and 11. Figure 10 is a table illustrating an example of the value of the gain adjustment coefficient k _g for tap position signal of the pixel intensity edge is input. Figure 11 is a diagram illustrating a change in the value of the gain adjustment coefficient k _g for tap position signal of the pixel intensity edge is input. 10 and 11 show the case where the BPF is configured with 15 taps.

For example, as illustrated in FIG. 10, when a signal of a pixel with a strong edge is input to the front and rear two taps including the center tap P ₀ , that is, the positions of the taps P _{−2 to} P ₂ , the gain adjustment coefficient k _g is set to 1. Is set. When a signal of a pixel with a strong edge is input at a position (P ₋₃ , P ₃ ) 3 taps away from the center tap P ₀ , the gain adjustment coefficient k _g is set to a value smaller than 1, for example, 0.8. The When a signal of a pixel with a strong edge is input at a position (P ₋₄ , P ₄ ) 4 taps away from the center tap P ₀ , the gain adjustment coefficient k _g is a value smaller than 0.8, for example 0.4 Set to

Similarly, when a signal of a pixel with a strong edge is input at a position (P ₋₅ , P ₅ ) 5 taps away from the center tap P ₀ , the gain adjustment coefficient k _g is set to 0.3, When a signal of a pixel with a strong edge is input at a position (P ₋₆ , P ₆ ) that is 6 taps away from the center tap P ₀ , the gain adjustment coefficient k _g is set to 0.2. Then, a position farthest from the center tap _{P 0 (P -7, P 7} ) If the signal of the pixel intensity edge is inputted, the gain adjustment coefficient _{k g} is set to 0.1.

In the case where the signal intensity edge pixels to a plurality of taps of the BPF is input, focusing on tap near to the center tap sets the gain adjustment coefficient k _g. For example, when the signal of the pixel of the strong edge is input to the position (P ₋₇ ) that is 7 taps away from the center tap and the position (P ₃ ) that is 3 taps away, the position (P ₃ ) that is 3 taps away since attention is paid to the gain adjustment coefficient _{k g} is set to 0.8.

Relationship tap position the intensity signal of the edge of the pixel is input and the gain adjustment coefficient k _g is, when it is set as shown in the table of FIG. 10, the gain adjustment coefficient k _g how over time It will be described with reference to FIG. As shown in the uppermost stage of FIG. 11, in the pixels S ₁₁ , S ₁₂ ,..., S ₄₁ , S ₄₂ aligned in the horizontal direction, it is assumed that the pixel S ₂₅ is a strong edge. Since the pixels are read into the BPF 26 in the order in which they are arranged in the horizontal direction, the pixel S ₂₅ that is a strong edge is first read into the tap of the BPF 26 when an operation is performed on the pixel S ₁₈ that is located seven before the pixel S _25. It is.

In other words, when the signal of the pixel S ₁₈ to the center tap are input, the signal of the pixel S ₂₅ which is a strong edge in a position away 7 taps from the center tap are input. (Refer to the top row of “pixels input to BPF 26” in the middle stage of FIG. 11.) Therefore, the gain adjustment coefficient k _g used for the processing of the pixel S ₁₈ is represented in the table of FIG. A gain adjustment coefficient _kg of 0.1 is set when a pixel signal of a strong edge is input at a position (P ₇ ) that is 7 taps away from the center tap P ₀ .

When operations on pixel S ₁₈ ends in BPF 26, pixels that are loaded into each tap moves to the previous tap, the pixel S ₂₆ is read at the end of the tap. That is, 15 pixels of pixels S _{12 to} S ₂₆ are read into the BPF ₂₆ , and the signal of the pixel S ₁₉ is input to the center tap. At this time, the signal of the pixel S ₂₅ a strong edge is input to a position away 6 taps from the center tap. (Stage in the middle of FIG. 11, see the second row from the top of "pixels input to BPF26".) Thus, the gain adjustment coefficient k _g used for processing of the pixel S _19, in the table of FIG. 10 Then, 0.2 is set as the gain adjustment coefficient _kg when the signal of the pixel of the strong edge is input at a position (P ₆ ) 6 taps away from the center tap P ₀ .

In this way, the signal of the pixel S ₂₅ that is a strong edge moves to the previous tap one by one as the processing progresses, so that each time the calculation target pixel read into the center tap changes, The distance gets closer by one tap. Therefore, if the signal of the pixel S ₂₀ is the next operand is input to the center tap position of the tap signal of the pixel S ₂₅ which is a strong edge is input becomes a position away 5 taps from the center tap. As the process proceeds and the pixel signal input to the center tap moves to the pixels S ₂₁ , S ₂₂ , S ₂₃ , and S ₂₄ , the tap of the tap to which the signal of the pixel S ₂₅ that is a strong edge is input. The position also moves from the center tap 4 taps before, 3 taps before, 2 taps before, 1 tap before. Therefore, the gain adjustment coefficient _{k g} also will change to 0.2,0.3,0.4,0.8,1,1, the larger direction.

When the processing further proceeds and the signal of the pixel S ₂₅ which is a strong edge is input to the center tap and the calculation is performed, the gain adjustment coefficient k _g is set to 1. After this, the signal of the pixel S ₂₅ that is a strong edge moves to the previous tap one by one as the process progresses, so every time the calculation target pixel read into the center tap changes, The distance is one tap away. Therefore, as the processing proceeds and the pixel signal input to the center tap moves to the pixels S ₂₆ , S ₂₇ , S ₂₈ , S ₂₉ , S ₃₀ , S ₃₁ , S ₃₂ , the pixel that is a strong edge after one tap from position also the center tap of the tap signal S ₂₅ is input, after two-tap, 3-tap, after four taps, five taps, 6-tap, moves the post 7 taps. Therefore, the gain adjustment coefficient _{k g} also 1,1,0.8,0.4,0.3,0.2,0.1, and will change to decrease direction.

Incidentally, if the signal of the pixel _{S 25} which is a strong edge is outside the range to be read into BPF 26, i.e., when the pixel signal input to the center tap of the pixels _{S 11} ~ pixels _{S 17,} and the pixel _{S 33} ~ pixels for S _42, the gain adjustment coefficient k _g is set to 1, the normal gain is used.

In this way, the gain adjustment unit 25, is multiplied against the strong based on the tap in the position information of the edge pixels of the signal to set the gain adjustment coefficient k _g, set gain values given them externally To adjust the gain value. The adjusted gain value is output to the gain unit 28.

  The BPF 26 extracts and outputs a frequency component of a predetermined band (including not only a high frequency but also a low frequency to a middle frequency) from an input image signal. In the coring unit 27, a low-level signal component is removed from the image signal extracted from the frequency component of a predetermined band output from the BPF 26, and the noise component is reduced. In the gain unit 28, the image signal after noise removal output from the coring unit 27 is multiplied by the adjusted gain value output from the gain adjustment unit 25 to adjust the value. In the limiter unit 29, signal components having a predetermined input level or higher are removed from the adjusted image signal output from the gain unit 28, and excessively amplified and emphasized components are removed. Finally, the adder 30 adds the image signal output from the limiter 29 and the original input image signal to obtain a desired contour-enhanced image signal.

  As described above, according to the present embodiment, when performing edge enhancement processing of an image signal using not only a high frequency but also a frequency band from a middle frequency to a low frequency, that is, from a video signal using a BPF having a large number of taps. When contour enhancement processing is performed by extracting a predetermined frequency band component and adding it to the original input image signal, the gain unit 28 according to the distance between the tap to which the signal of the pixel of the strong edge is input and the center tap. The gain value used in is adjusted. That is, when a pixel with a strong edge is located at a center tap or a tap near the center tap, a preset gain value (or a value close thereto) is used, but a pixel with a strong edge is located at a tap away from the center tap. If this is the case, decrease the gain value. Thereby, even when a strong edge exists at a position away from the center tap, it is possible to prevent the width of the shoot component from being expanded.

In the above-described example, BPF is used to extract a desired frequency band component, but other digital filters such as HPF and LPF (low pass filter) may be used. The value of the gain adjustment coefficient k _g is not limited to the values shown in the table of FIG. 10, it is possible to use an appropriate value depending on the characteristics of the image signal and BPF inputted.

  Each “unit” in this specification is a conceptual one corresponding to each function of the embodiment, and does not necessarily correspond to a specific hardware or software routine on a one-to-one basis. Therefore, in the present specification, the embodiment has been described assuming a virtual circuit block (unit) having each function of the embodiment.

  Although several embodiments of the present invention have been described, these embodiments are illustrated by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1 ... Image signal processing apparatus, 2 ... Edge detection part, 3 ... Strong edge extraction part, 4 ... Edge expansion part, 5 ... BPF with a mixing function, 6 ... Coring part, 7 ... Gain part, 8 ... Limiter part, 9 ... adder,

Claims (8)

A first digital filter composed of a plurality of taps for extracting a frequency component of a specific band corresponding to a contour portion of the image from an input image signal;
An edge detection unit that detects pixels corresponding to the contour portion from the image signal;
The signal level of the pixel detected by the edge detection unit is compared with a plurality of threshold values for edge determination to extract a strong edge, and the intensity level of the edge of the pixel extracted as the strong edge is A strong edge extraction unit for determination using a plurality of edge determination thresholds;
With
In the first digital filter, when the pixel input to a center tap that is a tap located at the center among the plurality of taps of the first digital filter is not extracted as a strong edge, the peripheral When the difference value between the tap output value and the center tap output value is greater than a predetermined signal level threshold value, the output value of the peripheral tap that has been determined that the difference is greater than the threshold value is the output value of the center tap. When the pixel input to the center tap is extracted as a strong edge, the output value of the peripheral tap and the output value of the center tap are set as the edge strength of the center tap. The lower the tap, the higher the ratio of the output value of the center tap is. And JP further comprising a correction circuit for performing a mixing process of the output value, the image signal processing apparatus. A first digital filter composed of a plurality of taps for extracting a frequency component of a specific band corresponding to a contour portion of the image from an input image signal;
An edge detection unit that detects pixels corresponding to the contour portion from the image signal;
A strong edge is extracted by comparing the signal level of the pixel detected by the edge detection unit with a predetermined threshold for edge determination, and the intensity level of the edge of the pixel extracted as the strong edge is used for the edge determination A strong edge extraction unit for determination using a threshold;
A plurality of taps of the first digital filter, and a predetermined distance or more from the center tap according to the strength of the edge of the pixel input to the center tap that is a tap located at the center. An image signal processing apparatus, wherein correction processing is performed on output values of peripheral taps arranged at distant positions using output values of the center tap.   The image processing apparatus according to claim 2, wherein a plurality of the threshold values for edge determination are set in the strong edge extraction unit.   In the correction process when the pixel input to the center tap is not extracted as a strong edge, the difference value between the output value of the peripheral tap and the output value of the center tap is lower than a predetermined signal level threshold value. 4. The image according to claim 2, wherein when the difference is larger, the replacement process replaces an output value of the peripheral tap, which is determined that the difference is larger than a threshold, with an output value of the center tap. 5. Processing equipment.   In the case where the pixel input to the center tap is extracted as a strong edge, the correction process mixes the output value of the peripheral tap and the output value of the center tap at a predetermined mixing ratio, and The image processing apparatus according to claim 2, wherein the image processing apparatus is a mixing process for generating an output value.   The image processing apparatus according to claim 5, wherein the mixing ratio is set such that the ratio of the output value of the center tap increases as the edge strength of the center tap decreases.   The image processing apparatus according to claim 2, wherein the first digital filter includes a correction circuit that performs the correction process.   A second digital filter comprising the same number of taps as the first digital filter and extracting a frequency component in the same specific band as the first digital filter from the input image signal; and the first digital filter A selection mixing unit that mixes the output of the filter and the output of the second filter at a predetermined ratio in accordance with the intensity of the edge of the pixel input to the center tap; The image processing apparatus according to claim 4, wherein the filter includes a replacement circuit that performs the replacement process.