JP4339237B2 - Sequential scan converter - Google Patents

Description

  The present invention relates to a progressive scan conversion device that converts an interlaced image into a progressive image, and more particularly to a progressive scan conversion device that performs progressive scan conversion by motion compensation.

  A liquid crystal display or plasma display generally displays a progressive image. Accordingly, when displaying an interlaced image on a liquid crystal display or a plasma display, sequential scanning conversion (I / P conversion) for converting the interlaced image into a progressive image is necessary.

  The progressive scan conversion will be described with reference to FIG. In the progressive scan conversion, an interpolated field image in which scanning lines are in an interpolating relationship with the input field image is generated, and a progressive image is generated by combining the two.

  FIG. 5A shows an interlaced field image. In the interlace method, for example, a top field image 101A composed of odd-numbered scanning lines and a bottom field image 102A composed of even-numbered scanning lines are alternately displayed. The top field image 101A and the bottom field image 102A are referred to as one frame. An interpolated field image 101B consisting of even-numbered scanning lines is generated for the top field image 101A consisting of odd-numbered scanning lines, and a progressive image is generated by combining the two. Similarly, an interpolated field image 102B consisting of odd-numbered scanning lines is generated for the bottom field image 102A consisting of even-numbered scanning lines, and a composite image is generated to generate a progressive image. In this way, an interpolated field image in which scanning lines are in an interpolating relationship with respect to all top field images and bottom field images is generated, and a progressive image is generated by synthesizing both.

  As the progressive scan conversion, a motion adaptive type progressive scan conversion and a motion compensation type progressive scan conversion are known.

  The motion adaptation method will be described with reference to FIG. In the motion adaptation method, motion detection is performed by comparing an input field image with a field image in the past of one frame period. In the still image portion, an interpolated field image is generated using an image in the past of one field period. This is called inter-field interpolation. In the moving image portion, an interpolated field image is generated by obtaining an average between adjacent scanning lines in the field image at the same time. This is called intra-field interpolation. For example, when generating the interpolation field image 102B for the bottom field image 102A, intra-field interpolation is performed on the moving object image portion.

  For intra-field interpolation, as described above, the average value of the pixel values of the upper and lower scanning lines is used as the value of the interpolated pixel, or when the field image includes diagonal lines, the diagonal lines are detected and the image quality of the diagonal lines is determined. There has been proposed a method of generating an interpolated pixel so as not to damage the image. In general, intra-field interpolation generates interpolated pixels using the same field image having information at the same time, so the sense of vertical resolution of the generated progressive image is low.

  The motion compensation method will be described with reference to FIG. In the motion compensation method, a motion vector is detected between frames or fields at different times. A motion vector is a vector indicating the movement of an image between frames or fields at different times. An interpolation field image is generated using the motion vector. For example, when generating the interpolation field image 102B for the bottom field image 102A, the moving object images 111A and 113A are detected from the two top field images 101A and 103A, and a motion vector 150 connecting the two is generated. By moving the object image 111A of the top field image 101A by a half vector of the motion vector 150, the object image 112B of the interpolation field image 102B is generated. As a method for detecting a motion vector, block matching is known.

  A motion vector detection method based on block matching will be described with reference to FIG. In block matching, a field image is divided into regions (blocks) composed of a plurality of pixels, blocks are compared between two field images at different times, and the degree of correlation between the two is evaluated. Find the smallest block. A motion vector is obtained from the blocks of the two field images thus obtained. As the degree of correlation, a cumulative value SAD (Sum of Absolute Error) of absolute differences between pixels of two field image blocks is used. That is, the motion vector is obtained from the block having the smallest SAD.

  In order to accurately detect a motion vector, it is necessary to accurately detect a block. However, in block matching, since the block having the smallest SAD is selected, an incorrect block may be interpreted as a correct block when the SAD is zero or small.

  FIG. 6A shows a case where a motion vector is accurately detected because a block is accurately detected by block matching. It is assumed that the block 121 in the top field image 101A and the block 123 in the next top field image 103A are accurately detected. That is, it was detected that both blocks correspond from the calculated value of the correlation between the block 121 including the moving object image 111A and the block 123 including the moving object image 113A. Therefore, the motion vector 150 obtained based on both blocks is accurate. When an interpolation field image 102B for the bottom field image 102A is generated using the accurate motion vector 150 in this way, and a progressive image is generated by synthesizing both, an accurate progressive image can be obtained.

  FIG. 6B shows a case where a motion vector is detected incorrectly because a block is detected incorrectly by block matching. Since the block 131 in the top field image 101A and the block 133 in the next top field image 103A are erroneously detected, the motion vector 151 obtained based on both blocks is inaccurate. The block 132 on the interpolation field image 102B designated by the half vector of the motion vector 151 is erroneously determined as a still pixel. When the interpolation field image 102B with respect to the bottom field image 102A is generated using the accurate motion vector 151 in this way, and a progressive image is generated by combining the two, an inaccurate progressive image 202 is obtained.

  In the progressive scan conversion of motion compensation method, the vertical resolution is low, so if the object is moving in the vertical direction, a high-quality progressive image cannot be obtained, but the object moves in the horizontal direction. If it is, a high-quality progressive image can be obtained. However, when an object moving at a constant speed overlaps with a background object, the motion vector obtained by the motion detection described above may be different from the motion of the object on the screen. In such a case, the image quality of the interpolation field image by motion compensation may be worse than that of the interpolation field image by intra-field interpolation processing.

  In order to solve such a problem, there is an example described in Patent Document 1, for example.

  With reference to FIG. 10, an example of the progressive scan conversion apparatus described in Patent Document 1 will be described. The progressive scan conversion apparatus includes a motion vector detection circuit 1001, a motion compensation interpolation circuit 1002, an intra-field interpolation circuit 1003, a vertical bandpass filter 1004, a synthesis circuit 1005, and a double speed conversion circuit 1006. The input video signal is input to the motion vector detection circuit 1001, the motion compensation interpolation circuit 1002, and the intra-field interpolation circuit 1003, respectively. The motion vector detection circuit 1001 detects a motion vector and supplies it to the motion compensation interpolation circuit 1002 and the synthesis circuit 1005. The motion compensation interpolation circuit 1002 performs motion compensation on the video signal of the previous field, for example, according to the motion vector and outputs it. The vertical bandpass filter 1004 extracts a vertical high-frequency component from the motion-compensated previous-field video signal. The synthesizing circuit 1005 synthesizes the vertical high-frequency component and the vertical low-frequency component of the current field obtained from the intra-field interpolation circuit, and outputs it as an interpolation signal for the scanning line.

  The double speed conversion circuit forms a progressive image from the input field image and the interpolated field image output from the synthesis circuit 1005. In this way, even when the motion vector is inaccurate, the image quality of the sequentially scanned image can be maintained.

JP 2001-24987 A

  However, even if the motion compensation is limited to the vertical high-frequency component as in the progressive scan converter of FIG. 10, if the motion vector cannot completely detect the actual motion of the object, the image deterioration due to the compensation is completely prevented. I can't. In particular, when an interlaced field image input to the progressive scan converter has an unnatural motion on the entire screen, the image quality degradation is larger in the interpolation processing by motion compensation than in the intra-field interpolation processing.

  An example of an unnatural movement of the entire input field screen is an interlaced field image converted from a film image by pull-down processing.

  FIG. 3 shows processing for converting a progressive image into an interlaced field image by 2: 2 pull-down processing. In 2: 2 pull-down processing, a progressive screen such as film or computer graphics is converted into an interlaced image having a scanning frequency of twice. For example, in the case of the NTSC system, an interlaced image having a scanning frequency of 60 fields / second is generated from a progressive image having a scanning frequency of 30 frames / second. In the case of the PAL method, an interlaced image having a scanning frequency of 50 fields / second is generated from a progressive image having a scanning frequency of 25 frames / second. As shown in the figure, a top field image is generated from even lines of a progressive image, and a bottom field image is generated from odd lines.

  In the 2: 2 pull-down process, since both the top field image 101A and the bottom field image 102A are generated from the same progressive image 101, the adjacent field images 101A and 102A may have information of the same time. For this reason, when a progressive image is generated by progressive scan conversion from an interlaced image generated by 2: 2 pull-down processing, the image quality of the generated progressive image is degraded.

  For example, consider a case where an interpolation field image 103B for the top field image 103A is generated from two bottom field images 102A and 104A by motion compensation. The interpolation field image 103B is generated as having intermediate time information between the two bottom field images 102A and 104A. On the other hand, the top field image 103A has information at the same time as the bottom field image 104A. Therefore, the progressive image generated by combining the top field image 103A and the interpolation field image 103B is a combination of information at different times, and the image deteriorates.

  In particular, when the input field image includes a moving object, a large difference occurs between the pixel values of the even lines and the odd lines that are vertically adjacent to each other, and a phenomenon called visual combing occurs.

  FIG. 4 shows processing for converting a progressive image into an interlaced field image by 3: 2 pull-down processing. In 3: 2 pull-down processing, a progressive screen such as film or computer graphics is converted into an interlaced image having a scanning frequency of 2.5 times. For example, in the case of the NTSC system, an interlaced image having a scanning frequency of 60 fields / second is generated from a progressive image having a scanning frequency of 24 frames / second. As shown in the drawing, two field images and three field images are alternately generated from one progressive image. For example, a top field image 101A is generated from an even line of one progressive image 101, a bottom field image 102A is generated from an odd line, and two top field images 103A, 105A and an odd line are generated from an even line of the next progressive image 103. One bottom field image 104A is generated.

  When an interpolated field image is generated by motion compensation from a field image obtained by 3: 2 pull-down processing, and a progressive image is generated by synthesizing both, an image similar to the case of a field image obtained by 2: 2 pull-down processing is obtained. Deterioration issues arise.

  Another example in which interlaced field images exhibit unnatural motion is an image that is displayed in slow motion by repeatedly displaying each field image captured at a normal field frequency. In this process, the number of repeated display of the field image is determined by the slow motion playback speed. In such an image, when the same field image is repeatedly displayed, the motion between the field images becomes zero. Therefore, the motion between adjacent field images is the same as the above-described 2: 2 pull-down processing and 3: 2 pull-down processing. The amount becomes irregular. As a result, the progressive image generated by the interpolation field image by motion compensation may be combed.

  Another example in which an interlaced image exhibits unnatural motion is an image that has been subjected to scan rate conversion processing. For example, a moving image converted between NTSC and PAL may have irregular motion between converted field images due to the influence of the scan rate conversion process. As a result, the progressive image generated by the interpolation field image by motion compensation may be combed.

  An object of the present invention is to provide an apparatus that performs progressive scan conversion by motion compensation that can avoid image quality degradation even when an image showing such an unnatural motion is input.

  According to the present invention, the progressive scan conversion apparatus uses a field memory that delays an input field image, a motion detection unit that detects a motion vector and correlation information from two field images having different display times, and the motion vector. A motion compensation unit for generating an interpolation field image by motion compensation, an intra-field interpolation image generation unit for generating an intra-field interpolation field image from an input field image, an interpolation field image by the motion compensation and the intra-field interpolation field image A blend unit that generates an interpolated field image by blending at a predetermined blend rate, a progressive image generating unit that generates a progressive image by combining the interpolated field image generated by the blend unit and the input field image, and an input F A reliability determination unit that determines reliability of the interpolation field image based on motion compensation using the correlation information, and a reliability determination unit that determines reliability of the field image, the interpolation field image based on motion compensation, and the correlation information. The blend ratio of the blend portion is controlled according to the determination result.

  According to the present invention, the reliability determination unit controls the output of a motion vector from the motion detection unit according to a reliability determination result.

  According to the present invention, there is further provided a scene change detection unit for detecting a scene change in the input field image, and the reliability determination output from the reliability determination unit for a certain period after the scene change detection unit detects the scene change. Control the result.

  According to the present invention, there is further provided a film detection unit for detecting whether the input field image is a video image, a field image by 2: 2 pull-down processing, or a field image by 3: 2 pull-down processing, and the film detection unit The reliability determination result output by the reliability determination unit is controlled during a period during which 2: 2 pull-down processing or 3: 2 pull-down processing is being determined.

  According to the present invention, it is possible to prevent image quality degradation due to motion compensation when a moving image having an unnatural motion is input to the progressive scan converter.

  A first example of the progressive scan converter of the present invention will be described with reference to FIG. The progressive scan conversion apparatus of this example includes a first field memory 10, a second field memory 11, a motion detection unit 12, a motion compensation unit 13, an intra-field interpolation image generation unit 14, a blending unit 15, and a progressive image generation unit 16. And a reliability determination unit 17.

  First, the input field image 21 is supplied to the first field memory 10 and the motion detection unit 12. The field image 22 from the first field memory 10 is supplied to the second field memory 11 and the intra-field interpolation image generation unit 14.

  Each of the first and second field memories 10 and 11 stores one field image and outputs it for each field period. Accordingly, the first field memory 10 outputs a field image 22 that is one field period past from the input field image 21, and the second field memory 11 outputs two field (one frame) period past from the input field image 21. Field image 23 is output.

  The motion detection unit 12 performs block matching between the input field image 21 and the field image 23 past two fields (one frame), thereby detecting a motion vector 29 and correlation information 28. As described with reference to FIG. 6, block matching is a process of dividing a field image into blocks and comparing two field images for each block, and is a well-known technique, and will not be described in detail here. The motion detection unit 12 outputs the correlation information 28 to the reliability determination unit 17 and outputs the motion vector 29 to the motion compensation unit 13. As the correlation information 28, a sum of absolute values SAD (Sum of Absolute Error) for each pixel of the block indicated by the motion vector may be used. Hereinafter, a case where SAD is used as the correlation information 28 will be described.

  The motion compensation unit 13 performs motion compensation processing using the motion vector 29 from the motion detection unit 12, and generates an interpolation field image 25 that is one field period past from the input field image 21. As an example of motion compensation processing, there is a method of generating an interpolated field image by moving the input field image 23 in the past for two fields (one frame) by a half size of the motion vector 29 for each block. As another example of motion compensation processing, for each block, a field image obtained by moving the input field image 23 in the past of two fields (one frame) by half the size of the motion vector 29, and the input field image 21 are There is a method of generating an interpolated field image by mixing field images shifted in the opposite direction by half the magnitude of the motion vector 29 at a ratio of 1: 1. The present invention does not limit the operation of the motion compensation process.

  The interpolated field image 25 generated by the motion compensation unit 13 is output to the reliability determination unit 17 and the blend unit 15.

  The intra-field interpolated image generation unit 14 receives the field image 22 from the first field memory 10, performs intra-field interpolation, and generates an interpolated field image 24 past one field period from the input field image 21. As described with reference to FIG. 5B, intra-field interpolation is a process for performing scanning line interpolation by calculating an average value of adjacent scanning lines in the same time field image. Yes, not described in detail here.

  The reliability determination unit 17 first determines the reliability of the interpolation field image 25 generated by the motion compensation unit 13, that is, the reliability of the motion vector. For the determination of reliability, the number of combing detections and the correlation information from the motion detection unit 12 are used. First, combing will be described. The reliability determination unit 17 synthesizes the field image 22 from the first field memory 10 and the interpolated field image 25 from the motion compensation unit 13 to generate a progressive image for combing detection. When the reliability of the motion vector 29 is low, combing occurs in the progressive image for combing detection. Therefore, the reliability of the motion vector 29 can be determined by detecting combing from the progressive image for detecting combing.

  In order to determine combing, for example, in a progressive image for combing determination, a certain number of pixel rows continuous in the vertical direction are extracted, and the absolute value of the difference between pixel values adjacent to each other is compared with a predetermined threshold value. When the absolute values of all the differences are larger than a predetermined threshold, it is determined that there is combing in the pixel column.

  The reliability determination unit 17 correlates the correlation information from the motion detection unit 12, that is, when the cumulative value of SAD or the average value of SAD is greater than the first threshold value, or the combing detected in the progressive image for combing determination When the total number is larger than the second threshold, it is determined that the reliability of the motion vector is low. Note that when the correlation information from the motion detection unit 12, that is, the SAD is large and the number of combing is large, it may be determined that the reliability of the motion vector is low.

  Next, the reliability determination unit 17 generates a blend rate 30 based on the determination result of the reliability of the motion vector, and performs output control 31 of the motion detection unit 12. The blend ratio 30 is a mixing ratio of the interpolation field image 25 from the motion compensation unit 13 to the interpolation field image 24 from the intra-field interpolation image generation unit 14 in the blending unit 15. When the reliability of the motion vector 29 is low, the ratio of the interpolation field image 25 from the motion compensation unit 13 is reduced. For example, the ratio of the interpolation field image 25 from the motion compensation unit 13 is set to 0%. When the reliability of the motion vector 29 is high, the ratio of the interpolation field image 25 from the motion compensation unit 13 is increased. For example, when the blend rate is 20% (1: 4), the interpolation field image 25 from the motion compensation unit 13 and the interpolation field image 24 from the intra-field interpolation image generation unit 14 are mixed at a ratio of 1: 4. .

  The reliability determination unit 17 outputs the blend rate 30 to the blend unit 15. The reliability determination unit 17 outputs the blend ratio 30 at the same timing as the timing of the field image 22 from the first field memory 10. Therefore, the reliability determination in the reliability determination unit 17 is executed at the same timing as the timing of the field image 22 from the first field memory 10.

  The reliability determination unit 17 adjusts the motion vector 29 output from the motion detection unit 12 based on the determination result of the motion vector reliability. When the reliability of the motion vector is low, the motion vector 29 output from the motion detection unit 12 to the motion compensation unit 13 is set to, for example, zero. Alternatively, when the motion vector detected by the motion detection unit 12 is not zero, control is performed so that the motion vector is likely to be zero by adding a constant value to a function for evaluating block matching. Thereby, it is possible to prevent detection of an unnatural amount of movement other than the stationary state.

  The blend unit 15 mixes the interpolation field image 25 from the motion compensation unit 13 and the interpolation field image 24 from the intra-field interpolation image generation unit 14 based on the blend rate 30 from the reliability determination unit 17, and finally An interpolation field image 26 is generated. The blend unit 15 outputs the generated interpolation field image 26 to the progressive image generation unit 16. The interpolated field image 26 output from the blending unit 15 to the progressive image generating unit 16 is a field image that is one field period past from the input field image 21.

  The progressive image generating unit 16 combines the field image 22 from the first field memory 10 and the interpolated field image 26 from the blending unit 15 to form a progressive image 27 and output it. The progressive image 27 is an image past one field period from the input field image 21.

  A second example of the progressive scan converter of the present invention will be described with reference to FIG. The progressive scan conversion apparatus of this example is different from the first example of FIG. 1 in that a film detection unit 18 and a scene change detection unit 19 are additionally provided. Here, a different part from a 1st example is demonstrated.

  The film detection unit 18 inputs an input field image 21, and is an image from an interlaced video image, a field image obtained by performing 2: 2 pull-down processing on a film image, or Whether the field image is obtained by the 3: 2 pull-down process is detected, and the detection result 32 is output to the reliability determination unit 17 and the progressive image generation unit 16.

  As a general method for detecting whether or not an image is a pull-down process, it is possible to detect whether or not adjacent field images are field images at the same time, and examine the continuous period or period of the field images at the same time. That's fine. For example, as shown in FIG. 3, when two field images of the same time are consecutive, it is determined that the field image is a 2: 2 pull-down process, and as shown in FIG. When three field images continue after two field images, it can be determined that the field image is a 3: 2 pull-down process.

  In order to detect whether or not the image is a pull-down process, it is necessary to evaluate the correlation between adjacent field images over several field periods of the input field image. In particular, in order to avoid erroneous detection and detect with high accuracy, it is necessary to evaluate the correlation between input field images over a longer period. Therefore, the detection result by the film detection unit 18 cannot be performed simultaneously with the input of the input field image.

  FIG. 7 shows the timing of film detection. FIG. 7A schematically shows a state in which an input field image is input every 1/60 seconds. FIG. 7B shows the actual source of the input field image. The input field image up to time t1 indicates that an interlaced video image is the source, and the input field image from time t1 is an image whose source is a film image. That is, the input field image indicates that, at time t1, the field image is switched from the interlace video image to the field image obtained by pulling down the film image.

  FIG. 7C shows a delay in the detection result by the film detection unit 18. As shown in the figure, the film detection unit 18 outputs a detection result that the source of the input field image is a video image during a period 402 until time t2 later than the time t1 when the source of the input field image actually changed. In a period 400 from time t2 to time t3 after time t1, the film detection unit 18 determines the source of the input field image. At time t3, a determination result that the source of the input field image has changed is output. Therefore, the film detector 18 outputs a detection result that the source of the input field image is a film image during a period 401 after the time t3.

  Next, the scene change detection unit 19 will be described. The scene change detection unit 19 detects a scene change point in the input field image, that is, a scene change, and outputs a detection result 33 to the reliability determination unit 17. Generally, when the source of the input field image is switched, the image content changes, and there is a high possibility of determining a scene change. Therefore, the detection of the scene change can be used as one of the judgment materials that the source of the input field image has been switched. As a general method of detecting a scene change, there is a method of evaluating a correlation between adjacent field images and determining a scene change when it is determined that the correlation is smaller than a predetermined threshold.

  Immediately after the source of the input field image is switched, the image quality of the progressive image may be deteriorated. In this example, the following processing is performed in order to prevent the deterioration of the image quality of the progressive image due to the switching of the source of the input field image.

  The scene change detection unit 19 outputs to the reliability determination unit 17 that the scene has been changed for a certain period 403 immediately after detecting the scene change.

  The film detection unit 18 detects the detection result, that is, the source of the input field image is a video image, the 2: 2 pulldown is being detected, the 3: 2 pulldown is being detected, and the input field image is an image by 2: 2 pulldown. A certain input field image is an image by 3: 2 pull-down, and the like are output to the reliability determination unit 17 and the progressive image generation unit 16.

  The reliability determination unit 17 adds motion compensation based on the period 403 immediately after the scene change from the scene change detection unit 19 or the detection result from the film detection unit 18 in addition to the conditions used in the first example. Limit the behavior of

  When the progressive image generation unit 16 outputs that the 2: 2 pull-down image is detected from the film detection unit 18, as shown in FIG. 8, the top field image and the bottom field at the same time are used without using the interpolation field image. A progressive image is constructed from the image. Further, when the progressive image generation unit 16 outputs that the 3: 2 pull-down image is detected from the film detection unit 18, as shown in FIG. 9, the top-field image and the same time without using the interpolation field image are displayed. A progressive image is composed of a bottom field image or a bottom field image and a top field image.

  Thus, in this example, when a video image is switched to an image by pull-down processing and input, progressive scan conversion by motion compensation is performed without degrading the image quality of the progressive image.

  The example of the present invention has been described above, but the present invention is not limited to the above-described example, and various modifications can be made by those skilled in the art within the scope of the invention described in the claims. It will be understood.

It is a figure which shows the 1st example of the motion compensation progressive scan conversion apparatus of this invention. It is a figure which shows the 2nd example of the motion compensation progressive scanning conversion apparatus of this invention. It is a figure which shows the method of producing | generating the field image of an interlace system by 2: 2 pull-down process from a film image | video. It is a figure which shows the method of producing | generating the field image of an interlace system from 3: 2 pull-down process from a film image | video. It is a figure for demonstrating intra-field interpolation and motion compensation interpolation. It is a figure which shows motion compensation interpolation. It is a figure which shows an example of film detection operation | movement. It is a figure which shows an example of the progressive scanning conversion process with respect to 2: 2 pull-down image. It is a figure which shows an example of the progressive scanning conversion process with respect to 3: 2 pull-down image. It is the top view which showed the prior art example of the motion compensation progressive scan conversion circuit.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10, 11 ... Field memory, 12 ... Motion detection part, 13 ... Motion compensation part, 14 ... Inter-field interpolation image generation part, 15 ... Blend part, 16 ... Progressive image generation part, 17 ... Reliability determination part, 18 ... Film Detection unit, 19 ... Scene change detection unit, 28 ... Correlation information, 29 ... Motion vector

Claims (4)

A field memory that delays an input field image, a motion detection unit that detects a motion vector and correlation information from two field images having different display times, and a motion that generates an interpolation field image by motion compensation using the motion vector A compensation unit, an intra-field interpolated image generating unit that generates an intra-field interpolated field image from the input field image, and interpolating by blending the interpolated field image by the motion compensation and the intra-field interpolated field image at a predetermined blend rate A blend unit that generates a field image, a progressive image generation unit that generates a progressive image by combining the interpolated field image generated by the blend unit and the input field image, an input field image, and interpolation based on the motion compensation Field image, and, and a reliability determination unit configured to determine reliability of the interpolation field image by the motion compensation using the correlation information,
The motion detection unit performs block matching between the two field images and outputs a sum of absolute differences of pixel values between blocks as the correlation information,
The reliability determination unit has low reliability of the interpolation field image by the motion compensation when the cumulative value of the sum of absolute differences or the average value of the sum of absolute differences is larger than a first threshold value over the entire screen. And
Further, the reliability determination unit generates a progressive image for combing determination by synthesizing the input field image and the interpolation field image by motion compensation, and in the progressive image for combing determination, a constant continuous in the vertical direction is generated. When the absolute value of the difference between adjacent pixel values is larger than a predetermined threshold, it is determined that there is combing in the pixel column, and the number of combing in the progressive image for the combing determination Is more than the second threshold value, it is determined that the reliability of the interpolation field image by the motion compensation is low,
The reliability determination unit controls a blend rate of the blend unit in accordance with a reliability determination result .   2. The progressive scan conversion apparatus according to claim 1, wherein when the reliability determination unit determines that the reliability of the interpolation field image by the motion compensation is low, the motion vector candidate in the motion detection unit is set to only a zero vector. Alternatively, a sequential scan conversion device that performs control so that a zero vector is easily detected by adding a constant value to a function that evaluates block matching for a vector other than the zero vector in motion vector search. 3. The progressive scan conversion device according to claim 1, further comprising a scene change detection unit for detecting a scene change in the input field image, wherein the reliability is maintained for a certain period after the scene change detection unit detects a scene change. The sex determination unit reduces the blend ratio of the interpolated field image with respect to the intra-field interpolated field image or sets it to zero. 3. The progressive scan conversion apparatus according to claim 1, further comprising a film detection unit for detecting whether the input field image is a video image, a field image by 2: 2 pull-down processing, or a field image by 3: 2 pull-down processing. And the reliability determination unit reduces the blend ratio of the interpolated field image with respect to the intra-field interpolated field image during a period in which the film detection unit is determining 2: 2 pulldown processing or 3: 2 pulldown processing, Alternatively, a progressive scan conversion device characterized in that it is zero.

Images