JP4412436B2 - Same image detection method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an identical image detection method for detecting an identical image developed over two or more consecutive fields in an interlaced image.
[0002]
[Prior art]
When performing field interpolation to convert interlaced images to non-interlaced images by interpolating lines in each field of the interlaced image, image motion is detected, and inter-field interpolation is performed based on the result. And intra-field interpolation are adaptively switched and executed.
[0003]
[Problems to be solved by the invention]
However, in an image of a movie film or an image created by CG (computer graphics), as shown in FIG. 3, one image is developed across a plurality of fields. For such interlaced images, rather than performing normal field interpolation based on motion detection, non-interlaced images are detected by detecting consecutive fields in which the same image is developed and combining these fields. Obtaining an image should be more faithful to the original image and provide higher image quality.
[0004]
An object of the present invention is to provide an identical image detection method that can solve the above-described problems in conventional field interpolation and obtain a high-quality non-interlaced image.
[0005]
[Means for Solving the Problems]
The method of the first invention is the same image detection method for detecting the same image developed over two or more consecutive fields in an interlaced image, and (a) within an area set in the current field. For each set of two pixels adjacent in the vertical direction, the values A and A ′ of the two pixels, the pixel value X at a position corresponding to the center of the two pixels in the previous field, and the value in the rear field Based on the pixel value Y at the position corresponding to the center of the two pixels, the difference value dX = median (A, A ′, X) −X and the difference value dY = median (A, A ′, Y) −Y are obtained. (B) a step of counting the number N1 of the sets satisfying abs (dX−dY) <T1 with respect to the reference value T1 (> 0), and (c) a reference value T2 (> 0). And abs (dX) <T2 and abs (dX−dY) ≧ T1 simultaneously with respect to the reference value T1 A step of counting the number N2 of the sets to be satisfied; and (d) the set that simultaneously satisfies abs (dY) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 and the reference value T1. (E) for the number N of all sets of the two pixels in the region, the number N1 is close to the number N under a predetermined criterion, or If N2> N3 and the number difference N2-N3 is close to or at least one of the number differences N-N1-N3 under another predetermined criterion, the current field and the previous field are the same Determining that the image is an image.
[0006]
The method of the second invention is the same image detection method of the first invention, wherein (f) the number N1 is close to the number N under the predetermined criterion or N3> N2 at the same time. When the number difference N3-N2 is near or at least one of the number differences N-N1-N2 under the different predetermined criterion, it is determined that the current field and the subsequent field are the same image. The process is further provided.
[0007]
According to a third invention, in the same image detection method of the second invention, (g) it is determined in the step (e) that the current field and the previous field are the same image, and the step ( The method further includes the step of predicting that the subsequent field and the subsequent field are the same image when it is not determined in f) that the current field and the subsequent field are the same image.
[0008]
According to a fourth aspect of the invention, there is provided an identical image detection method for detecting an identical image developed over two or more consecutive fields in an interlaced image, wherein: (a) within an area set in a current field; For each set of two pixels adjacent in the vertical direction, the values A and A ′ of the two pixels, the pixel value X at a position corresponding to the center of the two pixels in the previous field, and the value in the rear field Based on the pixel value Y at the position corresponding to the center of the two pixels, the difference value dX = median (A, A ′, X) −X and the difference value dY = median (A, A ′, Y) −Y are obtained. (B) a step of counting the number N1 of the sets satisfying abs (dX−dY) <T1 with respect to the reference value T1 (> 0), and (c) a reference value T2 (> 0). And abs (dX) <T2 and abs (dX−dY) ≧ T1 simultaneously with respect to the reference value T1 A step of counting the number N2 of the sets to be satisfied; and (d) the set that simultaneously satisfies abs (dY) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 and the reference value T1. (E) for the number N of all sets of the two pixels in the region, the number N1 is close to the number N under a predetermined criterion, or If N3> N2 and the number difference N3-N2 is close to or at least one of the number differences N-N1-N2 under another predetermined criterion, the current field and the subsequent field are the same Determining that the image is an image.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a flowchart showing a processing procedure of a field interpolation method according to an embodiment of the present invention. FIG. 2 is an explanatory diagram relating to the processing of FIG. In these drawings, the vertical axis V represents the vertical direction on the screen (direction perpendicular to the line), and the horizontal axis H represents the horizontal direction on the screen (direction parallel to the line). Reference numerals (i−1), (i), and (i + 1) represent identification numbers assigned to the fields of the interlaced image along the time flow. The field (i) is the current field, the field (i-1) corresponds to the previous previous field, and the field (i + 1) corresponds to the next future field.
[0010]
In FIG. 2, symbols A and A ′ represent pairs of pixels adjacent vertically in the current field (i), and symbol X represents the center of two pixels A and A ′ in the previous field (i−1). A pixel at a corresponding position is represented, and a symbol Y represents a pixel at a position corresponding to the center of the two pixels A and A ′ in the subsequent field (i + 1). In FIG. 2, the parallel strips extending horizontally represent lines (scanning lines) existing in each field.
[0011]
In a color image, the Y component (luminance component) is preferably selected as the pixel value to be calculated, but other components can also be used, and each of the three components can be used individually. It is also possible to perform the same calculation. Hereinafter, in order to simplify the description, a pixel and its value (pixel value) are represented by the same symbol.
[0012]
When the same image detection processing of FIG. 1 is started, first, for each of the two pixel pairs (A, A ′) in the current field (i), the difference value dX = median (A, A ′, X ) −X and the difference value dY = median (A, A ′, Y) −Y are calculated (step S1). Here, the symbol “median” represents an intermediate value. In the next step S2, an area is set in the current field (i), and a reference value T1 (> 0) and a reference value T2 (> 0) are set. The area may be all or part of the current field (i). The area is set in advance, and in step S1, the difference values dX and dY may be calculated only for the group (A, A ′) in the area.
[0013]
The reference values T1 and T2 may be constants, or may be variables such as ¼ × abs (AA ′). Here, the symbol “abs” represents an absolute value. Further, the reference values T1 and T2 may be set to be the same or different from each other. Further, the difference values dX and dY may be further subjected to a filter process such as a low-pass filter, and the value after the filter process may be used in a later step.
[0014]
In step S2, four numbers N, N1, N2, and N3 are counted. The number N is the number of all pairs (A, A ′) in the region. The number N1 is the number of sets (A, A ′) that satisfy abs (dX−dY) <T1 in the region. The number N2 is the number of sets (A, A ′) that simultaneously satisfy abs (dX) <T2 and abs (dX−dY) ≧ T1 in the region. Further, the number N3 is the number of sets (A, A ′) that simultaneously satisfy abs (dY) <T2 and abs (dX−dY) ≧ T1 in the region. In addition, when the area | region is preset and the total number N of the group (A, A ') which belongs to this is known, the total number N does not need to be counted.
[0015]
Next, with respect to the total number N of pairs (A, A ′) in the region, the number N1 is close to the number N under a predetermined criterion, or N2> N3 and at the same time the number difference N2− It is determined whether N3 is close to the number difference N−N1−N3 under another predetermined criterion (step S3). In this determination, if a positive determination result is obtained, it is determined that the current field (i) and the previous field (i-1) are the same image (step S4), and if a negative determination result is obtained. Then, it is determined that the current field (i) and the previous field (i-1) are not the same image (step S5).
[0016]
As the determination based on the above-mentioned predetermined reference, for example, a determination as to whether or not N1 / N ≧ α can be employed using the reference value α (0 <α <1). Further, as a determination based on the above-described another criterion, for example, a determination is made as to whether or not (N2-N3) / (N-N1-N3) ≧ β using a reference value β (0 <β <1). Can be adopted. In this case, the reference values α and β are usually set to values close to 1, such as 15/16 <α and β <1, but the two fields being compared are predicted to be the same image. If so, the reference value β is set to a small value such as β = 1/4. In this way, by changing the determination reference value by prediction, it is possible to prevent omission of detection and erroneous detection of the same image and to improve detection accuracy.
[0017]
Next, with respect to the total number N of the pairs (A, A ′) in the region, the number N1 is close to the number N under the above-mentioned predetermined criterion, or N3> N2 and at the same time the number difference It is determined whether N3−N2 is close to the number difference N−N1−N2 under another predetermined criterion described above (step S6). In this determination, if a positive determination result is obtained, it is determined that the current field (i) and the subsequent field (i + 1) are the same image (step S7), and if a negative determination result is obtained. Then, it is determined that the current field (i) and the subsequent field (i + 1) are not the same image (step S8).
[0018]
Next, it is determined whether a positive determination result is obtained in step S3 and a negative determination result is obtained in step S6 (step S9). If a positive determination result is obtained in this determination, it is predicted that the subsequent field (i + 1) and the subsequent field (i + 2) are the same image (step S10). When the above process is completed, the same image detection process is completed.
[0019]
Through the above processing, it is determined whether or not the current field (i) and the previous field (i-1), and the current field (i) and the subsequent field (i + 1) correspond to consecutive fields that develop the same image. Is done. By performing field interpolation using the result, even when an interlaced image includes an image such as a movie film, the field interpolation is performed by adaptively switching only between inter-field interpolation and intra-field interpolation. Compared with the method, a higher-quality non-interlaced image can be obtained.
[0020]
Further, prediction is made as to whether the subsequent field (i + 1) and the subsequent field (i + 2) correspond to consecutive fields that develop the same image. By using this prediction result for the next identical image determination, the detection accuracy of the same image can be improved.
[0021]
【The invention's effect】
In the method of the first invention, a set of two adjacent pixels satisfying a predetermined condition is counted in an area set in the current field and compared with each other, thereby comparing the current field and the previous field with a simple calculation procedure. It is determined whether or not the field is the same image. By performing field interpolation using this result, a high-quality non-interlaced image can be obtained even when the interlaced image includes an image such as a movie film.
[0022]
In the method of the second invention, since it is also determined whether or not the current field and the subsequent field are the same image, whether or not the subsequent field and the next field are the same image using the result. Can also be predicted, and the next same image determination can be performed using the result. Thereby, the detection accuracy of the same image can be improved.
[0023]
In the method of the third aspect of the invention, since prediction is made as to whether or not the subsequent field and the subsequent field are the same image, the next same image determination can be performed using the result. Thereby, the detection accuracy of the same image can be improved.
[0024]
In the method according to the fourth aspect of the present invention, a set of two adjacent pixels satisfying a predetermined condition is counted in an area set in the current field, and compared with each other, thereby comparing the current field and the subsequent field with a simple calculation procedure. It is determined whether or not the field is the same image. By performing field interpolation using this result, a high-quality non-interlaced image can be obtained even when the interlaced image includes an image such as a movie film.
[Brief description of the drawings]
FIG. 1 is a flowchart illustrating a processing procedure of a method according to an embodiment.
FIG. 2 is an explanatory diagram regarding processing according to FIG. 1;
FIG. 3 is an explanatory diagram showing a problem of a conventional field interpolation method.
[Explanation of symbols]
(i) Current field A, A ', X, Y Pixel dX, dY Difference value N, N1, N2, N3 Number T1, T2 Reference value

Claims (4)

An identical image detection method for detecting an identical image developed over two or more consecutive fields in an interlaced image,
(a) For each set of two pixels adjacent in the vertical direction within the region set in the current field, corresponding to the values A and A ′ of the two pixels and the center of the two pixels in the previous field Based on the pixel value X at the position and the pixel value Y at the position corresponding to the center of the two pixels in the subsequent field, the difference value dX = median (A, A ′, X) −X and the difference value dY = Calculating median (A, A ', Y) -Y;
(b) counting the number N1 of the sets satisfying abs (dX−dY) <T1 with respect to the reference value T1 (>0);
(c) counting the number N2 of the sets that simultaneously satisfy abs (dX) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 (> 0) and the reference value T1;
(d) counting the number N3 of the sets that simultaneously satisfy abs (dY) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 and the reference value T1;
(e) For the number N of all sets of the two pixels in the region, the number N1 is close to the number N under a predetermined criterion, or N2> N3 and the number difference Determining that the current field and the previous field are the same image if N2-N3 is near or at least one of the number differences N-N1-N3 under another predetermined criterion;
An identical image detection method comprising: (f) The number N1 is close to the number N under the predetermined criterion or N3> N2 and at the same time the number difference N3-N2 is the number difference N under the another predetermined criterion. The same image detection method according to claim 1, further comprising a step of determining that the current field and the subsequent field are the same image when it is close to or at least one of -N1-N2.   (g) In step (e), the current field and the previous field are determined to be the same image, and in step (f), the current field and the subsequent field are not determined to be the same image. 3. The same image detection method according to claim 2, further comprising a step of predicting that the subsequent field and the subsequent field are the same image. An identical image detection method for detecting an identical image developed over two or more consecutive fields in an interlaced image,
(a) For each set of two pixels adjacent in the vertical direction within the region set in the current field, corresponding to the values A and A ′ of the two pixels and the center of the two pixels in the previous field Based on the pixel value X at the position and the pixel value Y at the position corresponding to the center of the two pixels in the subsequent field, the difference value dX = median (A, A ′, X) −X and the difference value dY = Calculating median (A, A ', Y) -Y;
(b) counting the number N1 of the sets satisfying abs (dX−dY) <T1 with respect to the reference value T1 (>0);
(c) counting the number N2 of the sets that simultaneously satisfy abs (dX) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 (> 0) and the reference value T1;
(d) counting the number N3 of the sets that simultaneously satisfy abs (dY) <T2 and abs (dX−dY) ≧ T1 with respect to the reference value T2 and the reference value T1;
(e) With respect to the number N of all the sets of the two pixels in the region, the number N1 is close to the number N under a predetermined criterion, or N3> N2 and at the same time the number difference Determining that the current field and the subsequent field are the same image if N3-N2 is near or at least one of the number differences N-N1-N2 under another predetermined criterion;
An identical image detection method comprising:

Images