JP4868249B2 - Video signal processing device - Google Patents

Description

  The present invention relates to a video signal processing apparatus, and in particular, video signal processing for obtaining an output image that has been subjected to image enlargement / resolution conversion in a video display apparatus, video recording apparatus, video processing apparatus, computer apparatus, etc. that handles gradation images Relates to the device.

  With the recent digitization of video equipment, the spread of personal computers and digital cameras, and the introduction of ultra-high resolution video equipment, the importance of high-quality and high-resolution of existing video sources is increasing. For example, a method of enlarging an image by interpolating between pixels of a video signal of an existing video source is used for increasing the resolution.

  Widely known interpolation methods include nearest neighbor interpolation, linear interpolation, and three-dimensional convolution interpolation. The nearest neighbor interpolation method is a method in which the value of the nearest sample point of the converted image is used as an interpolation value, and has an advantage that the algorithm can be easily realized. The linear interpolation method is a method in which a ratio of distances from the surrounding four pixels at the interpolation position of the converted image is obtained, and interpolation is performed from the surrounding four pixel values according to the ratio, and the image becomes smooth. There are benefits. The three-dimensional convolution interpolation method uses a cubic function approximating the sinc function and performs calculation by convolution from 16 points around the interpolation position, and at the same time the image becomes smooth and has a sharpening effect. is there.

  However, in the conventional interpolation method as described above, since the surrounding information is evenly distributed to create interpolation data, a smooth interpolation image can be obtained at the non-edge portion, but the edge portion, particularly a slanting diagonal line It is inevitable that jaggy will stand out.

  For this reason, an interpolation method using orthogonal transformation / inverse transformation in which edge portions and non-edge portions are separately used and jaggies are not noticeable in the edge portion has been proposed (see, for example, Patent Document 1). .

  Also, the interpolation method corresponding to jaggies is conventionally known by using a monotonically increasing nonlinear function for interpolation between diagonally adjacent pixels and using a monotonically increasing nonlinear function or linear function for interpolating between vertically and horizontally adjacent pixels. (For example, refer to Patent Document 2).

  Furthermore, high-quality interpolation including diagonal lines is performed by repeatedly correcting the interpolated pixel value until the evaluation value obtained by the detailed function determined in advance based on the surrounding pixel values satisfies a preset condition. A method of performing the above has also been proposed (see, for example, Patent Document 3).

Japanese Patent No. 3578921 JP-A-5-284338 Japanese Patent Laid-Open No. 10-164358

  However, the interpolation methods described in Patent Documents 1 to 3 described above require processing of orthogonal transformation / inverse transformation, application of a nonlinear function using a memory table, and repeated calculation. In any case, there are obstacles in calculation cost, and there is a limit to the interpolation performance because it does not have a mechanism that only supports jaggies of diagonal lines at various angles.

  The present invention has been made in view of the above points, and detects the directionality of an edge of a video signal using a differential operator having a plurality of directions, obtains a weighted average along the detected edge, and obtains the maximum of neighboring pixels. It is an object of the present invention to provide a video signal processing apparatus that realizes high-performance oblique line interpolation with a relatively simple calculation by using an interpolation value that is limited between a value and a minimum value.

  In order to achieve the above object, a signal processing apparatus according to the present invention uses a differential operator having a plurality of directions when interpolating a video signal in a signal processing apparatus that interpolates pixel values of a video signal to increase resolution. The directionality detection means for detecting the directionality of the edge of the video signal by applying to the pixel values of the plurality of pixels set in advance, and the straight line indicating the directionality of the edge detected by the directionality detection means is a plurality of interpolation pixels When only passing through the interpolation position and does not pass through any of the plurality of preset pixel positions, a candidate value for the interpolation pixel is generated by linear interpolation of pixels around the interpolation pixel among the plurality of pixels, When the straight line passes only through the position of the first interpolation pixel at the center position among the plurality of interpolation pixels, the candidate value of the first interpolation pixel is set to two or more positions near the straight line among the plurality of pixels. Average the pixel values of the pixels And generating a candidate value of the second interpolation pixel at a position where the straight parallel line passes by averaging pixel values of two or more pixels at a position near the parallel line among the plurality of pixels. And candidate value generation means for generating a candidate value of the third interpolation pixel at a position where the straight line does not pass by linear interpolation of pixels around the third interpolation pixel among the plurality of pixels, and candidate value generation The candidate values generated by the means are limited by the maximum value and the minimum value among the pixel values of the central four pixels adjacent to the first interpolated pixel position in the top, bottom, left, and right of the plurality of pixels. And an interpolation value output means for outputting the candidate value as an interpolation value of the interpolation pixel.

  According to the present invention, high-performance oblique line interpolation with few errors can be realized by a relatively simple calculation process.

  Next, an embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a block diagram of an embodiment of a video signal processing apparatus according to the present invention. As shown in the figure, the video signal processing apparatus 100 according to the present embodiment generates a luminance signal Y and two types of color signals Cb and Cr based on the input three primary color signals (input RGB signals). Signals output from the matrix circuit 101, the enlargement processing unit 102 for the luminance signal Y, the enlargement processing unit 103 for the color signal Cb, the enlargement processing unit 104 for the color signal Cr, and the enlargement processing units 102 to 104 And a second matrix circuit 105 that generates an enlarged RGB signal.

  Next, the operation of this embodiment will be described.

  The first matrix circuit 101 converts an input RGB signal, which is a video signal to be increased in resolution, into a luminance signal Y and two types of color signals Cb and Cr using a known matrix operation. The Y enlargement processing unit 102 receives the luminance signal Y output from the first matrix circuit 101 as an input, detects the edge of the image of the luminance signal Y and its directionality, and detects the detected edge, as will be described later. The weighted average value obtained by weighted averaging the pixel values along the pixel by the weight determined by the directionality of the edge is limited by the value determined by the maximum value and the minimum value of a plurality of peripheral pixels. Then, enlargement processing using the limit value as an interpolation value is performed.

  On the other hand, the Cb enlargement processing unit 103 and the Cr enlargement processing unit 104 receive the color signals Cb and Cr output from the first matrix circuit 101 as inputs, respectively, and receive the input color signals Cb and Cr as well-known. Enlargement processing is performed by interpolation using linear interpolation. The second matrix circuit 105 receives, as inputs, the enlarged luminance signal Y, the enlarged color signal Cb, and Cr output from the enlargement processing units 102, 103, and 104, respectively, based on them. And the three primary color signals of the enlarged R (red) signal, G (green) signal, and B (blue) signal are output.

  Next, details of the enlargement processing method by the Y enlargement processing unit 102, which is the main part of the present embodiment, will be described.

  FIG. 2 is a flowchart for explaining the operation of the main part of the embodiment of the apparatus of the present invention. This flowchart shows the procedure of the enlargement process by the Y enlargement processing unit 102.

  First, using the pixel values of a total of 24 pixels (hereinafter referred to as y [n]) at the positions indicated by 0 to 23 in FIG. 3 of the input luminance signal Y, o0 and o1 in FIG. , O2, o3, and o4, an edge and its directionality are detected in order to calculate interpolation values at five positions (step S1).

  Note that “edge” refers to a boundary between regions where the above characteristics change rapidly when a region having similar features such as density value, color, and pattern is defined as one region. . The directionality detection is a process of selecting one from 12 types of directionality patterns described later by a predetermined method. Details of the selection method will be described later.

  In this step S1, a differential operator having a plurality of directions shown in FIGS. 4 to 15 is applied to the pixel value (luminance value) of the luminance signal Y to detect an edge and its directionality. Here, the “differential operator” means, for example, a pixel density in a feature extraction filter described in a well-known document (Koichi Sakai: “Introduction to Digital Image Processing”, Corona, Inc., P.30-32). It is a general term for operators for taking the difference. This known document discloses an operator of a Prewitt filter that is a differential filter extended to the vicinity region, and an operator of a Sobel filter that is a filter having a larger central weight in the Prewitt filter. These operators have a coefficient arrangement so that vertical and horizontal edges can be detected for a total of 9 pixels, 3 pixels in the vertical direction and 3 pixels in the horizontal direction.

  On the other hand, in the present embodiment, the directionality (angle) of 12 edges can be detected as will be described below by using a 24-pixel operator having the arrangement shown in FIG. Yes.

  Now, the horizontal directionality indicated by the dotted line D0 in FIG. 4 is referred to as directionality 0, and the vertical directionality indicated by the dotted line D1 in FIG. Similarly, the directivity indicated by dotted straight lines D2 to D11 in FIGS. 6 to 15 is referred to as directivity 2 to directivity 11, respectively. Further, the pixel arrangement of 24 pixels in FIGS. 4 to 15 showing the directionality k (k = 0, 1, 2,..., 11) is the same as that in FIG. The coefficient of the pixel indicated by ○ arranged on one side of ˜D11 is “10”, and the coefficient of the pixel indicated by ● arranged on the other side is “−10”. However, in FIG. 6 and FIG. 7, the four pixels located on the straight lines D2 and D3 are not counted, the coefficient of the pixel indicated by ◯ is “12”, and the coefficient of the pixel indicated by ● is “−12”. The straight lines D0 to D11 all pass through the center position (position of the interpolation pixel o0) of the arrangement of 24 pixels.

が方向性ｋにおいて最大になる場合、方向性ｋのエッジが存在すると考える。図２のステップＳ１では、方向性０〜１１について（１）式の演算を行い、微分作用素計算結果の絶対値ｘの最大値が得られる方向性を検出する。 If the coefficient of the pixel is c _k [n] and each of the 24 pixel values is y [n] (n is the number of each pixel shown in FIGS. 4 to 15),

Is the maximum in direction k, it is considered that there is an edge of direction k. In step S1 of FIG. 2, the calculation of the expression (1) is performed for the directivity 0 to 11, and the directivity at which the maximum value of the absolute value x of the differential operator calculation result is obtained is detected.

Subsequently, it is determined whether or not the maximum value x of the differential operator calculation result indicating the detected directionality is less than a predetermined value th1 (step S2). If the maximum value is less than th1, A non-edge is determined and linear interpolation is performed (step S3). In particular,
o0 = (y [8] + y [9] + y [14] + y [15]) / 4
o1 = (y [8] + y [9]) / 2
o2 = (y [8] + y [14]) / 2
o3 = (y [9] + y [15]) / 2
o4 = (y [14] + y [15]) / 2
As described above, the pixel values (interpolation values) of the interpolation pixels o0 to o4 are calculated based on the pixel values of four or two pixels located in the vicinity.

  On the other hand, when it is determined in step S2 that the maximum value of the absolute value x of the differential operator calculation result is equal to or greater than the predetermined value th1, the detected directionality is determined to be an edge and determined as the edge. The candidate values of the interpolation value by the weighted average are calculated for each of the directional characteristics (step S4). In the calculation of candidate values for the interpolation value, linear interpolation is also used everywhere, so the following relationship is determined in advance.

bo0 = (y [8] + y [9] + y [14] + y [15]) / 4
bo1 = (y [8] + y [9]) / 2
bo2 = (y [8] + y [14]) / 2
bo3 = (y [9] + y [15]) / 2
bo4 = (y [14] + y [15]) / 2
th2 = max (y [8], y [9], y [14], y [15]) − min (y [8], y [9], y [14], y [15])
Moreover, although it is estimation of the candidate value of the interpolation value before finally limiting, for convenience, the candidate value is also described by the expression o0, o1, o2, o3, o4, which is the same as the interpolation value.

  Here, th2 means the dynamic range of the current pixel around the interpolation position, and is used for case classification to prevent an edge estimation error or the like in the following calculation. Also, the pixels indicated by the numbers 8, 9, 14, and 15 in FIGS. 4 to 15 are “center 4 pixels” that occupy an important position for determining the interpolation value, and y [8], y [9], y [14] and y [15] are hereinafter referred to as “center four pixel values”.

  Next, a method for calculating candidate values when each directionality 0 to 11 is detected will be described.

(1) When Directionality 0 is Detected as Edge Directionality In this case, the straight line D0 indicating the directionality 0 shown in FIG. 4 passes through only the interpolation positions of the interpolation pixels o0, o2, and o3, and the video signal No candidate position of the interpolation value is calculated by the same method as the linear interpolation. Therefore, candidate values for the interpolation values of o0 to o4 in the directionality 0 shown in FIG. 4 are as follows.

o0 = bo0, o1 = bo1, o2 = bo2, o3 = bo3, o4 = bo4
(2) When Directionality 1 is Detected as Edge Directionality In this case as well, the straight line D1 indicating the directionality 1 shown in FIG. 5 is the interpolation position of o0, o1, and o4 interpolation pixels as in (1). Only when the positions of the 24 pixels 0 to 23 of the video signal do not pass through, and the candidate value of the interpolation value is calculated by the same method as the linear interpolation. Therefore, candidate values for the interpolation values of o0 to o4 in the directionality 1 shown in FIG. 5 are as follows.

o0 = bo0, o1 = bo1, o2 = bo2, o3 = bo3, o4 = bo4
(3) When Directionality 2 is Detected as Edge Directionality Of the candidate values for interpolation values o0 to o4 in directionality 2 shown in FIG. 6, o1 to o4 are set to the same values as linear interpolation, and o0 For, the pixel values y [9] and y [14] of the pixels 9 and 14 located on the straight line D2 indicating the directionality 2 in FIG. 6 are averaged. Therefore, in this case, it is as follows.

o0 = (y [9] + y [14]) / 2
o1 = bo1, o2 = bo2, o3 = bo3, o4 = bo4
(4) When Directionality 3 is Detected as Edge Directionality Among the candidate values for interpolation values o0 to o4 in the directionality 3 shown in FIG. 7, o1 to o4 are set to the same values as the linear interpolation, and o0. For, the pixel values y [8] and y [15] of the pixels 8 and 15 located on the straight line D3 indicating the directionality 3 in FIG. 7 are averaged. Therefore, in this case, it is as follows.

o0 = (y [8] + y [15]) / 2
o1 = bo1, o2 = bo2, o3 = bo3, o4 = bo4
(5) When Directionality 4 is Detected as Edge Directionality Among o0 to o4 in directionality 4 shown in FIG. 8, the candidate value for the interpolation value o0 at the center is represented by a straight line D4 indicating directionality 4. The pixel values of the nearest four pixels 9, 10, 13, 14 are averaged. The candidate values for the interpolation values o2 and o3 are obtained by averaging the pixel values of three pixels near the interpolation position close to the parallel line of the straight line D4 passing through the interpolation position. However, if the absolute value of the difference between the two pairs across the o0 position of the four pixels 9, 10, 13, 14 exceeds th2, the average calculation has a high probability of causing an error. Performs the same operation as. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [13] + y [14] + y [9] + y [10]) / 4
o1 = bo1
o2 = (y [9] + y [13] + y [8]) / 3
o3 = (y [10] + y [14] + y [15]) / 3
o4 = bo4
if (abs (y [10] −y [9])> th2 || abs (y [13] −y [14])> th2)
{o0 = (y [9] + y [14]) / 2; o2 = bo2; o3 = bo3}
(6) When Directionality 5 is Detected as Edge Directionality Among o0 to o4 in directionality 5 shown in FIG. 9, the candidate value for the interpolation value of o0 at the center is represented by a straight line D5 indicating directionality 5. The pixel values of the nearest four pixels 7, 8, 15, 16 are averaged. The candidate values for the interpolation values of o2 and o3 are obtained by averaging pixel values of three pixels near the interpolation position close to the parallel line of the straight line D5 passing through the interpolation position. However, if the absolute value of one of the two pairs of the four pixels 7, 8, 15, 16 across the o0 position exceeds th 2, the average calculation has a high probability of causing an error. Performs the same operation as. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [7] + y [8] + y [15] + y [16]) / 4
o1 = bo1
o2 = (y [7] + y [14] + y [15]) / 3
o3 = (y [8] + y [9] + y [16]) / 3
o4 = bo4
if (abs (y [7] −y [8])> th2 || abs (y [15] −y [16])> th2)
{o0 = (y [8] + y [15]) / 2; o2 = bo2; o3 = bo3}
(7) When Directionality 6 is Detected as Edge Directionality Among o0 to o4 in directionality 6 shown in FIG. 10, the candidate value for the interpolated value o0 at the center is a straight line D6 indicating directionality 6. The pixel values of the nearest four pixels 4, 9, 14, 19 are averaged. The candidate values for the interpolation values o1 and o4 are obtained by averaging pixel values of three pixels near the interpolation position close to the parallel line of the straight line D6 passing through the interpolation position. However, if the absolute value of one of the differences between two pairs across the o0 position of the four pixels 4, 9, 14, 19 exceeds th2, the average calculation has a high probability of causing an error. Performs the same operation as. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [4] + y [9] + y [14] + y [19]) / 4
o1 = (y [4] + y [8] + y [14]) / 3
o2 = bo2
o3 = bo3
o4 = (y [9] + y [15] + y [19]) / 3
if (abs (y [4] −y [9])> th2 || abs (y [19] −y [14])> th2)
{o0 = (y [9] + y [14]) / 2; o1 = bo1; o4 = bo4}
(8) When Directionality 7 is Detected as Edge Directionality Among o0 to o4 in directionality 7 shown in FIG. 11, the candidate value for the interpolation value o0 at the center is represented by a straight line D7 indicating directionality 7. The pixel values of the nearest four pixels 3, 8, 15, 20 are averaged. The candidate values for the interpolation values o1 and o4 are obtained by averaging the pixel values of three pixels near the interpolation position close to the parallel line of the straight line D7 passing through the interpolation position. However, if the absolute value of one of the two pairs of the four pixels 3, 8, 15, and 20 across the o0 position exceeds th2, the average calculation has a high probability of causing an error. Performs the same operation as. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [3] + y [8] + y [15] + y [20]) / 4
o1 = (y [3] + y [9] + y [15]) / 3
o2 = bo2
o3 = bo3
o4 = (y [8] + y [14] + y [20]) / 3
if (abs (y [3] −y [8])> th2 || abs (y [15] −y [20])> th2)
{o0 = (y [8] + y [15]) / 2; o1 = bo1; o4 = bo4}
(9) When Directionality 8 is Detected as Edge Directionality In this case, the straight line D8 indicating the directionality 8 shown in FIG. 12 passes through the position of the interpolation pixel o0 among the interpolation pixels o0 to o4. , O3 is in contact with each other. In this case, among the o0 to o4 in the directionality 8, the candidate value of the interpolation value of o0 at the center is obtained by averaging the pixel values of the four pixels 10, 11, 12, and 13 close to the straight line D8. The candidate values for the interpolation values o2 and o3 are obtained by averaging the pixel values of three pixels near the interpolation position close to the parallel line of the straight line D8 passing through the interpolation position. However, for each of the two pairs of the four pixels 10, 11, 12, and 13 across the o0 position, the difference between the center 4 pixel value and the nearest one near the straight line D8 is calculated, and the absolute value of the difference is th2. If the average is exceeded, linear interpolation is performed on the assumption that the average calculation has a high probability of causing an error. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [10] + y [11] + y [12] + y [13]) / 4
o1 = bo1
o2 = (y [8] + y [10] + y [12]) / 3
o3 = (y [11] + y [13] + y [14]) / 3
o4 = bo4
if (abs (y [10] −y [9])> th2 || abs (y [11] −y [9])> th2)
|| abs (y [12] −y [14])> th2 || abs (y [13] −y [14])> th2)
{o0 = bo0; o2 = bo2; o3 = bo3}
(10) When the directionality 9 is detected as the directionality of the edge In this case, the straight line D9 indicating the directionality 9 shown in FIG. 13 passes through the position of the interpolation pixel o0 among the interpolation pixels o0 to o4. , O3 is in contact with each other. In this case, among the o0 to o4 in the directionality 9 shown in FIG. 13, the pixel values of the four pixels 6, 7, 16, 17 close to the straight line D9 are averaged for the candidate value of the interpolation value o0 in the center. Ask. The candidate values for the interpolation values o2 and o3 are obtained by averaging the pixel values of three pixels near the interpolation position close to the parallel line of the straight line D9 passing through the interpolation position. However, for each of the two pairs of the four pixels 6, 7, 16, 17 sandwiching the o0 position, the difference between the center 4 pixel value and the nearest one near the straight line D9 is calculated, and the absolute value of the difference is th2. If the average is exceeded, linear interpolation is performed on the assumption that the average calculation has a high probability of causing an error. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [6] + y [7] + y [16] + y [17]) / 4
o1 = bo1
o2 = (y [6] + y [14] + y [16]) / 3
o3 = (y [7] + y [9] + y [17]) / 3
o4 = bo4
if (abs (y [6] −y [8])> th2 || abs (y [7] −y [8])> th2)
|| abs (y [16] −y [15])> th2 || abs (y [17] −y [15])> th2)
{o0 = bo0; o2 = bo2; o3 = bo3}
(11) When Directionality 10 is Detected as Edge Directionality In this case, the straight line D10 indicating the directionality 10 shown in FIG. 14 passes through the position of the interpolation pixel o0 among the interpolation pixels o0 to o4. , O4 is almost in contact with each other. In this case, among the o0 to o4 in the directionality 10 shown in FIG. 14, the pixel values of the four pixels 1, 4, 19, and 22 close to the straight line D10 are averaged for the candidate value of the interpolation value o0 in the center. Ask. The candidate values for the interpolation values o1 and o4 are obtained by averaging pixel values of three pixels near the interpolation position close to the parallel line of the straight line D10 passing through the interpolation position. However, for each of the two pairs of the four pixels 1, 4, 19, and 22 across the o0 position, the difference between the center 4 pixel value and the nearest one near the straight line D10 is calculated, and the absolute value of the difference is th2. If the average is exceeded, linear interpolation is performed on the assumption that the average calculation has a high probability of causing an error. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [1] + y [4] + y [19] + y [22]) / 4
o1 = (y [1] + y [8] + y [19]) / 3
o2 = bo2
o3 = bo3
o4 = (y [4] + y [15] + y [22]) / 3
if (abs (y [1] −y [9])> th2 || abs (y [4] −y [9])> th2)
|| abs (y [19] −y [14])> th2 || abs (y [22] −y [14])> th2)
{o0 = bo0; o1 = bo1; o4 = bo4}
(12) When the directionality 11 is detected as the edge directionality In this case, the straight line D11 indicating the directionality 11 shown in FIG. 15 passes through the position of the interpolation pixel o0 among the interpolation pixels o0 to o4. , O4 is in contact with each other. In this case, among the o0 to o4 in the directionality 11 shown in FIG. 15, the candidate values of the interpolation value of o0 in the center are averaged from the pixel values of the four pixels 0, 3, 20, and 23 close to the straight line D11. Ask. The candidate values for the interpolation values o1 and o4 are obtained by averaging the pixel values of three pixels near the interpolation position close to the parallel line of the straight line D11 passing through the interpolation position. However, for each of the two pairs of the four pixels 0, 3, 20, 23 sandwiching the o0 position, the difference between the center 4 pixel value and the nearest one near the straight line D11 is calculated, and the absolute value of the difference is th2. If the average is exceeded, linear interpolation is performed on the assumption that the average calculation has a high probability of causing an error. Therefore, in this case, candidate values for interpolation values of o0 to o4 are calculated by the following formula.

o0 = (y [0] + y [3] + y [20] + y [23]) / 4
o1 = (y [0] + y [9] + y [20]) / 3
o2 = bo2
o3 = bo3
o4 = (y [3] + y [14] + y [23]) / 3
if (abs (y [0] −y [8])> th2 || abs (y [3] −y [8])> th2)
|| abs (y [20] −y [15])> th2 || abs (y [23] −y [15])> th2)
{o0 = bo0; o1 = bo1; o4 = bo4}
Next, in the above step S4, the pixel values of the central four pixels, which are peripheral pixels, for the candidate values o0, o1, o2, o3, and o4 obtained by any of the processes (1) to (11) above. Are generated as the interpolation values (interpolated pixel values) at the interpolation positions indicated by o0, o1, o2, o3, and o4 in FIG. (Step S5).

  That is, in this step S5, the maximum value max (y [8], y [9], y [14] .y [15]) of the central four pixels and the minimum value min (y [8], y [9], y [14], y [15]), and candidate values limited to values within the range are output as interpolation values. That is, among candidate values o0 to o4, candidate values that are larger than the maximum value are limited to the maximum value and output as an interpolation value, and candidate values that are smaller than the minimum value are limited to the minimum value. And output as an interpolation value. The candidate values within the range of the maximum value and the minimum value are output as they are as an interpolation value.

  Here, the reason for limiting to the value within the range of the maximum value and the minimum value of the pixel values of the above four central pixels is that depending on the surrounding pixel values, it becomes a prominent interpolation value if it is not limited, and noise or aliasing This is because it may appear.

  As described above, according to the present embodiment, using the 24-pixel operator having the arrangement shown in FIG. 3 which is more complicated than the prior art, the image to be increased in resolution with a differential operator having 12 directions (angles). The directionality of the edge of the signal is detected, a candidate value is obtained from a weighted average of a plurality of pixels along the detected directionality of the edge, and among the candidate values, the maximum and minimum values of the central four pixels that are peripheral pixels are obtained. Since the interpolated value is generated as an interpolation value, high-performance oblique line interpolation with few errors can be realized by a relatively simple calculation process.

  The present invention is not limited to hardware image enlargement processing, and can be processed by a computer program. In this case, the computer program may be taken into the computer from a recording medium or may be taken into the computer via a network. In the above embodiment, the number of directions to be detected is “12”, but the present invention is not limited to this number.

It is a block diagram of one embodiment of the present invention. It is a flowchart for operation | movement description of the principal part of one embodiment of this invention. It is a figure which shows the pixel arrangement | positioning in the process unit of 1 time of the interpolation process by FIG. It is a figure which shows the differential operator of directionality 0. It is a figure which shows the differential operator of directionality 1. It is a figure which shows the differential operator of directionality 2. It is a figure which shows the differential operator of directionality 3. It is a figure which shows the differential operator of directionality 4. It is a figure which shows the differential operator of directionality 5. It is a figure which shows the differential operator of directionality 6. It is a figure which shows the differential operator of directionality 7. It is a figure which shows the differential operator of directionality 8. It is a figure which shows the differential operator of directionality 9. It is a figure which shows the differential operator of directionality 10. It is a figure which shows the differential operator of directionality 11.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Video signal processing apparatus 101 1st matrix circuit 102 Y enlargement process part 103 Cb enlargement process part 104 Cr enlargement process part 105 2nd matrix circuit

Claims (1)

In the video signal processing apparatus that interpolates the pixel value of the video signal to increase the resolution,
A directionality detecting means for applying a differential operator having a plurality of directions to pixel values of a plurality of pixels set in advance used for interpolation of the video signal, and detecting a directionality of an edge of the video signal;
When the straight line indicating the directionality of the edge detected by the directionality detection means passes only through the interpolation positions of a plurality of interpolation pixels and does not pass through any of the preset plurality of pixel positions, the interpolation is performed. A candidate pixel value is generated by linear interpolation of pixels around the interpolation pixel among the plurality of pixels, and the straight line passes only through the position of the first interpolation pixel at the center position among the plurality of interpolation pixels. When the candidate value of the first interpolation pixel is generated by averaging pixel values of two or more pixels in the vicinity of the straight line among the plurality of pixels, the position through which the parallel lines of the straight line pass Are generated by averaging pixel values of two or more pixels in the vicinity of the parallel line among the plurality of pixels, and the straight line does not pass through the candidate value. A candidate value for a third interpolation pixel is A candidate value generation means for generating by linear interpolation of pixels around the third interpolation pixel among the plurality of pixels,
The candidate value generated by the candidate value generating means is a maximum value and a minimum value among the pixel values of the central four pixels adjacent to the top, bottom, left, and right of the position of the first interpolation pixel among the plurality of pixels. A video signal processing apparatus comprising: an interpolation value output means for limiting and outputting the limited candidate value as an interpolation value of an interpolation pixel.

Images