JP2005202921A - Motion vector detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce erroneous detection of motion vectors more than conventional devices and methods when adding motion vectors detected by a block matching method to candidates of an initial bias vector to detect motion vectors by a gradient method or the like, by applying a motion vector detector and a motion vector detection method to detect motion vectors by the gradient method or the like using, especially, the initial bias vector. <P>SOLUTION: When inter-field differences and motion vectors are discriminated on the basis of prescribed discrimination reference values α, β1, and β2 to find that a motion has quickly changed and an inter-field difference value is large in the case of an already detected initial bias vector INT MVA, a motion vector MVB obtained by the block matching method is set as an initial bias vector V1. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a motion vector detection apparatus, and in particular, can be applied to detection of a motion vector by a gradient method using an initial deviation vector. The present invention relates to a case where the motion is suddenly changed by the determination of the inter-field difference and the motion vector based on a predetermined determination reference value, and the block value when the inter-field difference value is large in the detected initial deviation vector. By setting the motion vector by the matching method to the initial deviation vector, the motion vector detected by the block matching method is added to the initial deviation vector candidate and the motion vector is detected by the gradient method compared to the conventional method. It is possible to further reduce detection errors of motion vectors.

  Conventionally, in high-efficiency encoding processing of video signals, motion vectors are used to improve the efficiency of inter-frame encoding processing. In television signal format conversion, motion using motion vectors is used. By the correction, the jerky movement due to the conversion of the number of fields is corrected to a smooth movement.

  In such motion vector detection, a video signal is blocked into motion vector detection units (that is, macro blocks) of m pixels × n lines (m and n are integers), and a motion vector is detected for each block. For example, there are a pattern matching method disclosed in Japanese Patent Laid-Open Nos. 55-162683 and 55-162684, a gradient method disclosed in Japanese Patent Laid-Open No. 60-158786, a phase correlation method, and the like. It is made to be used.

  Among such motion vector detection methods, for the gradient method, for example, Japanese Patent Application Laid-Open No. 62-206980 proposes a method for detecting a motion vector using an initial deviation vector. Here, the initial deviation vector is a predicted value of a motion vector, and is detected from, for example, a motion vector already detected in a surrounding motion vector detection block. In this motion vector detection method using an initial displacement vector, Japanese Patent Laid-Open No. 4-78286 discloses a method of detecting a displacement vector with reference to the initial displacement vector, and then the initial displacement vector and the displacement vector. A method for detecting a motion vector with a simple configuration by detecting a motion vector by addition to the above is disclosed. According to this detection method, a motion vector can be detected by motion vector = initial displacement vector + deviation vector. For example, when the motion is uniform such as camera pan, the displacement vector is “0”. Thus, only the initial deviation vector has a value.

  However, the method of detecting a motion vector by the gradient method is a method of detecting a motion vector by an arithmetic expression using an image gradient and an inter-frame difference value, and has high detection accuracy but only detection accuracy for several pixels. When the amount of movement suddenly changes, the detection accuracy deteriorates, and there is a problem that it takes several fields to follow the true amount of movement.

  As one method for solving this problem, Japanese Patent Application Laid-Open No. 7-135663 discloses a method of processing a motion vector detected by a block matching method in addition to an initial deviation vector selection target. . In other words, in motion vector detection by the block matching method, there is no correlation in the time axis direction by simply detecting the motion vector by determining the magnitude of the inter-field difference value and inter-frame difference value detected in the motion vector detection range. The motion can also be detected, and thus the motion vector can be detected even when the motion changes suddenly. Thus, if the motion vector detected by the block matching method is added to the initial deviation candidate vector and the motion vector is detected by the gradient method, the motion vector can be detected with high accuracy in response to a sudden change in motion. It is done.

  However, when the motion vector detected simply by the block matching method is added to the initial deviation candidate vector and the motion vector is detected by the gradient method, the motion vector may be erroneously detected. In other words, in motion vector detection by the block matching method, motion vectors are detected by simply determining the magnitude of the inter-field difference value and inter-frame difference value detected in the motion vector detection range in each macro block. This is because a vector is erroneously detected, and a motion vector is detected randomly in an adjacent macroblock.

Such erroneous detection of motion vectors tends to become prominent when the amount of motion is small. However, when the motion vector detected by the block matching method is added to the initial deviation candidate vector and the gradient vector is used to detect the motion vector, when the amount of motion is large, the motion vector can be detected well even for a sudden change in motion. Can do.
Japanese Unexamined Patent Publication No. 55-162683 Japanese Patent Laid-Open No. 55-162684 JP 60-158786 A JP-A-4-78286 JP 7-135663 A

  The present invention has been made in consideration of the above points, and in addition to the motion vector detected by the block matching method in addition to the initial deviation vector candidate, for example, when motion vector detection is performed by the gradient method, the motion is compared with the prior art. It is an object of the present invention to propose a motion vector detection device that can reduce erroneous vector detection.

  In order to solve such a problem, in the first aspect of the invention, an initial deviation vector is detected from the detected motion vector, and the deviation vector is detected using the initial deviation vector to correct the initial deviation vector. The motion vector detection means for detecting a motion vector of a video signal by a block matching method and outputting a motion vector by the block matching method, applied to a motion vector detection device for detecting a motion vector of a sequentially input video signal And a difference value calculation means for the block matching method for detecting an inter-frame difference value or an inter-field difference value for the block matching method by performing motion correction using a motion vector by the block matching method, and an initial bias from the detected motion vector. The calculation means for detecting the candidate of the position vector and the candidate of the initial deviation vector Difference value calculation means for the detected motion vector for detecting the inter-frame difference value or the inter-field difference value related to the detected motion vector after motion correction, and outputting the initial displacement vector candidate as the initial displacement vector And selecting means for selecting and outputting a motion vector based on the block matching method as an initial displacement vector instead of the initial displacement vector candidate when the predetermined condition is satisfied, and satisfying the predetermined condition Is when the inter-frame difference value or the inter-field difference value related to the detected motion vector is larger than a predetermined determination reference value, and the inter-frame difference value or the inter-field difference value related to the block matching method is equal to or less than the determination reference value. To be.

  In the second aspect of the invention, the initial deviation vector is detected from the detected motion vector, the deviation vector is detected using the initial deviation vector, and the initial deviation vector is corrected to be sequentially input. Motion vector detecting means for detecting a motion vector of a video signal, detecting a motion vector of the video signal by a block matching method, and outputting a motion vector by the block matching method; An initial deviation candidate vector is selected from the motion vector, an arithmetic means for detecting an initial deviation vector candidate from the selected initial deviation candidate vector, and an initial deviation vector candidate are output as an initial deviation vector, and predetermined If the above condition is satisfied, the initial motion vector by the block matching method is used instead of the initial deviation vector candidate. Selection means for selecting and outputting as a position vector, and when a predetermined condition is satisfied, the magnitude of the motion vector by the block matching method is not less than the first determination reference value and the initial displacement candidate vector The maximum value of the magnitude is less than or equal to the second determination reference value that is smaller than the first determination reference value.

  In the invention of claim 3, the initial displacement vector is detected from the already detected motion vector, the displacement vector is detected using the initial displacement vector, and the initial displacement vector is corrected, thereby being sequentially input. Motion vector detecting means for detecting a motion vector of a video signal, detecting a motion vector of the video signal by a block matching method, and outputting a motion vector by the block matching method, and a block matching method Select the initial deviation candidate vector from the detected motion vector and the difference value calculation means related to the block matching method for detecting the inter-frame difference value or the inter-field difference value related to the block matching method. The candidate of the initial deviation vector is selected from the selected initial deviation candidate vector. A calculation unit that outputs, and a difference value calculation unit related to a detected motion vector that detects a difference value between frames or a difference value between fields related to a detected motion vector by performing motion correction with a candidate of an initial deviation vector; Initial deviation vector candidates are output as initial deviation vectors, and when predetermined conditions are satisfied, a motion vector based on the block matching method is selected and output as an initial deviation vector instead of the initial deviation vector candidates. And when the predetermined condition is satisfied, the inter-frame difference value or the inter-field difference value related to the detected motion vector is larger than the predetermined determination reference value and the inter-frame difference value related to the block matching method Alternatively, when the inter-field difference value is equal to or smaller than the determination reference value, and the magnitude of the motion vector by the block matching method is the first. And the determination reference value or more, and so that the maximum value of the magnitude of the initial excursion candidate vector is when: small second criterion value than the first criterion value.

  In the invention of claim 4, the initial displacement vector is detected from the detected motion vector, the displacement vector is detected using the initial displacement vector, and the initial displacement vector is corrected, so that the initial displacement vector is sequentially input. Motion vector detecting means for detecting a motion vector of a video signal, detecting a motion vector of the video signal by a block matching method, and outputting a motion vector by the block matching method; A selection output means for detecting and outputting an initial deviation vector from a motion vector, and a detection means for detecting a macroblock whose movement has changed suddenly from a video signal, and a macro whose movement has changed suddenly based on a detection result by the detection means The block matching method is used to select the initial deviation vector in the selection output means in the block. Add a vector.

  According to a fifth aspect of the present invention, in the configuration of the fourth aspect, the detecting means includes an inter-field difference value obtained by correcting the motion of the immediately preceding field with a motion vector that has already been detected according to the selection of the initial displacement vector, and the immediately preceding field. If the difference value between the difference value between fields detected by the motion vector already detected in the macroblock is larger than a predetermined determination reference value, it is determined that the motion has suddenly changed.

  According to a sixth aspect of the present invention, in the configuration of the fourth or fifth aspect, the detection means obtains the inter-field difference obtained by correcting the motion of the immediately preceding field with the detected motion vector related to the selection of the initial deviation vector. When the difference value between the value and the inter-field difference value detected by the motion vector already detected in the macroblock corresponding to the immediately preceding field is larger than a predetermined determination reference value, it is determined that the motion has suddenly changed.

  According to a seventh aspect of the present invention, in the configuration of the fourth, fifth or sixth aspect, a macroblock having a sudden change in motion is subjected to a motion based on a block matching method as a selection target of an initial deviation vector in the selection output means. A motion vector generated from a motion vector by the block matching method is added together with the vector.

  In the invention of claim 8, in the configuration of claim 7, the motion vector generated from the motion vector by the block matching method is a motion vector obtained by smoothing the motion vector by the block matching method by a two-dimensional low-pass filter. Like that.

  In the invention of claim 9, in the structure of claim 7 or 8, the motion vector generated from the motion vector by the block matching method is smoothed in the time axis direction by the low-pass filter. To be a motion vector.

  In the invention of claim 10, in the structure of claim 7, 8, or 9, the motion vector generated from the motion vector by the block matching method is a motion vector that occupies most in the immediately preceding field. To do.

  According to the configuration of the first aspect, the initial deviation vector is detected from the detected motion vector, the deviation vector is detected using the initial deviation vector, and the initial deviation vector is corrected to sequentially input video. Motion vector detection means for detecting a motion vector of a video signal by a block matching method applied to a motion vector detection device for detecting a motion vector of a signal and outputting a motion vector by the block matching method, and motion by the block matching method Detecting initial deviation vector candidates from the difference value calculation means according to the block matching method for detecting the interframe difference value or the interfield difference value according to the block matching method by performing motion correction with the vector The motion is corrected by the calculation means and the initial deviation vector candidate, and the detected motion is detected. When the difference value calculation means for the detected motion vector for detecting the inter-frame difference value or the inter-field difference value related to the vector and the initial deviation vector candidate are output as the initial deviation vector and the predetermined condition is satisfied And selecting means for selecting and outputting a motion vector by the block matching method as an initial displacement vector instead of the initial displacement vector candidate, the motion vector detected by the block matching method can be used as an initial displacement. In addition to vector candidates, a motion vector can be detected by, for example, a gradient method. When this predetermined condition is satisfied, the inter-frame difference value or inter-field difference value related to the detected motion vector is larger than the predetermined determination reference value, and the inter-frame difference value or inter-field difference related to the block matching method is used. If the value is equal to or less than the criterion value, the motion vector according to the block matching method can be applied to the initial deviation vector when it is determined that the motion vector according to the block matching method is highly reliable. Thus, when a motion vector detected by the block matching method is added to the initial deviation vector candidate and, for example, a motion vector is detected by the gradient method, erroneous detection of the motion vector can be reduced.

  Further, according to the configuration of the second aspect, the initial deviation vector is detected from the detected motion vector, the deviation vector is detected using the initial deviation vector, and the initial deviation vector is corrected to be sequentially input. Motion vector detection means for detecting a motion vector of a video signal by a block matching method by applying to a motion vector detection device for detecting a motion vector of a video signal, and outputting a motion vector by the block matching method; An initial deviation candidate vector is selected from the vector, an arithmetic means for detecting an initial deviation vector candidate from the selected initial deviation candidate vector, and an initial deviation vector candidate are output as an initial deviation vector, If the condition is satisfied, the initial deviation of the motion vector by the block matching method is used instead of the initial deviation vector candidate. If so and selecting means for selecting and outputting a vector, it is possible to detect the motion vector by adding the motion vector detected by block matching method to the candidate of the initial vector for example the gradient method. When this predetermined condition is satisfied, the magnitude of the motion vector by the block matching method is equal to or greater than the first determination reference value, and the maximum value of the initial deviation candidate vector is the first determination reference value. The motion vector according to the block matching method can be applied to the initial deviation vector when a sudden change in motion occurs due to the case where the value is less than the second determination reference value with a smaller value, and thereby the block matching When the motion vector detected by the method is added to the initial deviation vector candidate and, for example, the motion vector is detected by the gradient method, the erroneous detection of the motion vector can be reduced.

  According to the configuration of claim 3, the initial deviation vector is detected from the detected motion vector, and the initial deviation vector is detected by using the initial deviation vector to correct the initial deviation vector. Applied to a motion vector detection device that detects a motion vector of a video signal, detects a motion vector of the video signal by a block matching method, and outputs a motion vector by the block matching method, and by a block matching method Select an initial deviation candidate vector from the detected motion vector and difference value calculation means related to the block matching method for detecting the inter-frame difference value or the inter-field difference value related to the block matching method by performing motion correction with the motion vector. , Detecting initial deviation vector candidates from the selected initial deviation candidate vectors A difference value calculating means for the detected motion vector for detecting the inter-frame difference value or the inter-field difference value for the detected motion vector by performing motion correction using the initial deviation vector candidate; Selection of displacement vector candidates as initial displacement vectors, and selection and output of motion vectors by block matching instead of initial displacement vector candidates as initial displacement vectors when predetermined conditions are satisfied The motion vector detected by the block matching method can be added to the initial displacement vector candidate and, for example, the motion vector can be detected by the gradient method. When this predetermined condition is satisfied, the inter-frame difference value or inter-field difference value related to the detected motion vector is larger than the predetermined determination reference value, and the inter-frame difference value or inter-field difference related to the block matching method is used. The value is equal to or smaller than the criterion value, the magnitude of the motion vector by the block matching method is equal to or larger than the first criterion value, and the maximum value of the initial deviation candidate vector is the first criterion. If the value is equal to or smaller than the second determination reference value that is smaller than the reference value, it is determined that the motion vector according to the block matching method is highly reliable, and there is a sudden change in motion. If it occurs, the motion vector according to the block matching method can be applied to the initial displacement vector, thereby detecting the motion vector detected by the block matching method. The appropriate in addition to the candidate initial vector for example the gradient method in the case of detecting a motion vector, it is possible to reduce erroneous detection of a motion vector.

  According to the configuration of claim 4, the initial deviation vector is detected from the detected motion vector, the deviation vector is detected using the initial deviation vector, and the initial deviation vector is corrected to be sequentially input. Motion vector detection means for detecting a motion vector of a video signal by a block matching method by applying to a motion vector detection device for detecting a motion vector of a video signal, and outputting a motion vector by the block matching method; A selection output means for detecting and outputting an initial deviation vector from the vector; and a detection means for detecting a macroblock whose movement is suddenly changed from the video signal, and a macroblock whose movement is suddenly changed based on a detection result by the detection means. Thus, the motion vector based on the block matching method is selected as the initial deviation vector selection target in the selection output means. If the motion is suddenly changed and the motion vector already detected by the gradient method is applied to the initial deviation vector, the motion vector by the block matching method is changed to the initial bias when the sudden motion change cannot be followed. It is possible to detect a motion vector by setting the position vector, thereby reducing false detection of the motion vector as compared with the prior art.

  According to the configuration of claim 5, in the configuration of claim 4, the detection means includes the inter-field difference value obtained by correcting the motion of the immediately preceding field with the already detected motion vector related to the selection of the initial displacement vector, and the immediately preceding macro. If the difference value between the field-detected motion vectors in the block and the field-to-field difference value is greater than a predetermined criterion value, and if it is determined that the motion has suddenly changed, the motion vector by the block matching method is used as the initial displacement vector. It is possible to set the motion vector by the block matching method as the initial deviation vector when it is preferable to detect the motion vector by setting to, thereby reducing the detection error of the motion vector compared to the conventional case. it can.

  According to the configuration of claim 6, in the configuration of claim 4 or claim 5, the detection means obtains the inter-field difference value obtained by correcting the motion of the immediately preceding field with the detected motion vector related to the selection of the initial deviation vector. If the difference between the difference value between the fields detected by the motion vector detected in the macroblock corresponding to the immediately preceding field is larger than a predetermined criterion value, By detecting a sudden change in motion, it is possible to reduce false detection of motion vectors.

  Further, according to the configuration of claim 7, in the configuration of claim 4, 5 or 6, the motion vector by the block matching method is used as a selection target of the initial deviation vector in the selection output means in the macro block whose motion is suddenly changed. At the same time, if a motion vector generated from a motion vector obtained by the block matching method is added, a motion vector suitable for the initial displacement vector can be set as a selection target to detect the initial displacement vector.

  According to the configuration of claim 8, in the configuration of claim 7, the motion vector generated from the motion vector by the block matching method is a motion vector obtained by smoothing the motion vector by the block matching method by a two-dimensional low-pass filter. In this case, specifically, it is possible to detect an initial deviation vector by setting a motion vector suitable for the initial deviation vector as a selection target.

  Further, according to the configuration of claim 9, in the configuration of claim 7 or claim 8, the motion vector generated from the motion vector by the block matching method is smoothed in the time axis direction by the low-pass filter. Even if it is a motion vector, specifically, a motion vector suitable for the initial displacement vector can be set as a selection target, and the initial displacement vector can be detected.

  According to the configuration of claim 10, in the configuration of claim 7, 8, or 9, the motion vector generated from the motion vector by the block matching method is a motion vector that occupies most in the immediately preceding field. Specifically, the initial deviation vector can be detected by setting a motion vector suitable for the initial deviation vector as a selection target.

  According to the present invention, erroneous detection of a motion vector can be reduced as compared with the prior art.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings as appropriate.

(1) Configuration of Embodiment FIG. 2 is a block diagram showing a motion vector detection apparatus according to Embodiment 1 of the present invention. This motion vector detection device 1 is applied to, for example, a format conversion device, a coding device based on MPEG (Moving Picture Experts Group), a noise reduction device, etc., and detects a motion vector V from a luminance signal Y constituting a video signal. Output. In the detection of the motion vector, not only when the luminance signal Y is used, but the color difference signals RY and BY can also be used.

  The motion vector detection device 1 detects an approximate motion vector from the initial displacement vector V1, detects the displacement vector ΔV by the gradient method using the detected motion vector, and corrects the initial displacement vector V1. Thus, the motion vector V is detected.

  That is, in the motion vector detection device 1, the pre-filter 2 is a two-dimensional low-pass filter, which receives the luminance signal Y and suppresses high frequency components, thereby preventing erroneous detection of motion vectors due to noise. In addition, when the luminance signal Y is based on an interlaced video signal, the pre-filter 2 executes so-called center-of-gravity correction processing, so that when detecting a motion vector between consecutive fields, the lines differ between the fields. This prevents the deterioration of the motion vector detection accuracy. The pre-filter 2 is not limited to a low-pass filter, and a band-pass filter can be applied.

  The 1-field delay circuit 3 receives the luminance signal Y from the pre-filter 2 and outputs it after being delayed by a period of 1 field.

  The motion vector detection circuit 4 sequentially detects and outputs the motion vector MVB of the luminance signal Y output from the pre-filter 2 based on the luminance signal Y1 output from the one-field delay circuit 3 by the block matching method. . In this embodiment, macroblocks are sequentially set in the luminance signal Y output from the pre-filter 2, and the motion vector MVB is detected in units of macroblocks. This macroblock is composed of 8 pixels × 8 lines. It is designed to be set to size.

  The initial deviation vector selection circuit 5 inputs initial deviation candidate vectors VK1 to VK8 from the already detected motion vectors held in the memory circuit 8, and receives the motion vector MVB from the motion vector detection circuit 4. The initial deviation vector V1 is selected from these motion vectors VK1 to VK8 and MVB and output.

  The gradient method computing circuit 6 detects the displacement vector ΔV by the gradient method with reference to the initial displacement vector V1 output from the initial displacement vector selection circuit 5. That is, the gradient method computing circuit 6 performs motion correction on the luminance signal Y1 output from the one-field delay circuit 3 using the initial displacement vector V1, and the luminance signal Y output from the prefilter 2 The deviation vector ΔV is detected by calculation using the gradient method between the two.

  The adding circuit 7 corrects the initial displacement vector V1 by adding the displacement vector ΔV to the initial displacement vector V1, and outputs the result as a motion vector V.

  The memory circuit 8 records and holds the motion vector V output from the adder circuit 7. Further, from the held motion vector V, the gradient vector calculation circuit 6 uses the held motion vector V as a part of the initial deviation candidate vector in correspondence with the macro block of the motion vector detection target used for the calculation of the displacement vector ΔV. Output as. In this embodiment, eight types of motion vectors VK1 to VK8 are output as a part of the initial deviation candidate vectors from the detected motion vectors of the previous field and the current field.

  FIG. 1 is a block diagram showing the initial deviation vector selection circuit 5 together with its peripheral circuits. The initial deviation vector selection circuit 5 detects an initial deviation vector candidate from the already detected motion vector V, and selects a motion vector suitable for detecting a motion vector from the candidate vector and a motion vector by block matching. To set the initial deflection vector. That is, in the initial deviation vector selection circuit 5, the memory circuit 11 records and holds the luminance signal Y1 output from the one-field delay circuit 3, and the held luminance signal Y1 is an area having a size corresponding to the macroblock. Output in units. In this process, the memory circuit 11 holds the address by shifting the read address by the amount of the motion vector MVB detected by the motion vector detection circuit 4 with respect to the write address used for writing the luminance signal Y1. The luminance signal Y1 is output, and the luminance signal Y1 held thereby is corrected by the motion vector MVB and output.

  The difference calculation circuit 12 is in units of macro blocks of the luminance signal Y output from the prefilter 2 and between the luminance signal Y1 output from the memory circuit 11 and the luminance signal Y output from the prefilter 2. The field difference value DFD B is calculated and output. In this embodiment, the inter-field difference value DFD B is calculated by calculating a difference value of pixel values between corresponding pixels and adding the absolute value of the difference values for each macroblock. It is designed to be applied. Note that a square sum of such difference values may be applied to the inter-field difference value DFD B.

  The memory / difference calculation circuit 14 detects and outputs the motion vector INT MVA most suitable for the initial displacement vector from the initial displacement candidate vectors VK1 to VK8 output from the memory circuit 8. That is, the memory / difference calculation circuit 14 records and holds the luminance signal Y1 output from the one-field delay circuit 3, and reads out the held luminance signal Y1 in units of areas corresponding to macroblocks. By processing with the luminance signal Y output from the filter 2, the motion vector INT MVA which is a candidate of the initial displacement vector is detected from the detected motion vector V and output.

  Specifically, the memory / difference calculation circuit 14 sets initial deviation candidate vectors VK1 for each write address used for writing the luminance signal Y1 in units of macroblocks of the luminance signal Y output from the prefilter 2. The luminance signal Y1 held by the address control in which the read address is deviated is read by ˜VK8, and the luminance signal Y1 held thereby is subjected to motion correction by the initial deviation candidate vectors VK1 to VK8, respectively. Further, between the read luminance signal Y1 and the luminance signal Y output from the prefilter 2, an inter-field difference value is calculated for each of the initial deviation candidate vectors VK1 to VK8, and the inter-field difference value is calculated. The smallest initial deviation candidate vector is selected and output by the motion vector INT MVA. Further, the inter-field difference value DFD A related to the motion vector INT MVA is also output. Note that, even in the inter-field difference values related to the initial deviation candidate vectors VK1 to VK8, although the difference absolute value sum is applied, the difference between the pixel values corresponds to the inter-field difference value related to the motion vector MVB. You may make it apply the square sum of a value.

  The difference comparison circuit 13 outputs a motion vector MVB by the block matching method output from the motion vector detection circuit 4 and a maximum value MAX VK (VK1 to VK1) of the motion vectors VK1 to VK8 that have been detected and output from the memory / difference calculation circuit 14. VK8) is determined according to a predetermined criterion, and the initial deviation vector V1 is selectively output.

That is, the difference comparison circuit 13 has a motion vector MVB magnitude of the block matching method that is equal to or greater than a predetermined determination reference value β1 (MVB ≧ β1), and the maximum magnitude MAX VK of the magnitudes of the motion vectors VK1 to VK8 that have already been detected. (VK1 to VK8) is equal to or smaller than a predetermined determination reference value β2 (MAX VK (VK1 to VK8) ≦ β2), and the inter-field difference value DFD B according to the block matching method is a predetermined determination reference value α. If the inter-field difference value DFD A related to detection is smaller than this determination reference value α (DFD
A>α> DFD B), the motion vector MVB by the block matching method is selected as the initial deviation vector V1. Here, the determination reference values β1 and β2 satisfy β1> β2. If this condition is not satisfied, the detected motion vector INT MVA is set as the initial deviation vector V1.

  The determination reference values β1 and β2 are provided for the X component and the Y component, respectively. In the determination based on the determination reference values β1 and β2 of the motion vector MAX VK (VK1 to VK8) and MVB, the X component (VX ) And Y component (VY) are executed according to the determination criteria for each component, and when the above-mentioned determination criteria are satisfied for the component in any direction, the motion vector MVB by the block matching method is used as the initial displacement vector V1. Has been made to choose. Note that, as the determination reference values β1 and β2 and the determination reference value α, values set by an operator's operation are applied, or fixed values set in advance are applied. In the determination based on the determination reference values β1 and β2 of the motion vector MAX VK (VK1 to VK8) and MVB, the horizontal component VX and the vertical component VY of the motion vector are expressed by a relational expression of MV = (VX2 + VY2) 1/2. It is also possible to detect the magnitudes MV of the motion vectors MAX VK (VK1 to VK8) and MVB, respectively, and determine the magnitude MV based on the determination reference values β1 and β2.

  Therefore, in this motion vector detection device 1, the motion vector MVB by the block matching method is equal to or greater than a predetermined determination reference value β1 (MVB ≧ β1), and the maximum value MAX VK (VK1 to VK1) of the motion vectors VK1 to VK8 already detected. When VK8) is equal to or less than a predetermined determination reference value β2 (MAX VK (VK1 to VK8) ≦ β2), the motion is small up to the previous field, the motion in the current field is large, and the motion suddenly changes. It can be judged.

  On the other hand, when the inter-field difference value DFD B according to the block matching method is equal to or smaller than the predetermined determination reference value α and the already-detected inter-field difference value DFD A is larger than the determination reference value α, the block matching method It can be said that the motion correction by the motion vector MVB according to the method has a smaller inter-field difference, and the motion correction by the motion vector MVB according to the block matching method can perform the motion correction with higher reliability.

  Accordingly, the difference comparison circuit 13 is related to the block matching method when the motion is suddenly changed and it can be determined that the motion correction by the motion vector MVB according to the block matching method can be corrected with higher reliability. The motion vector MVB is set as the initial deviation vector, and in other cases, the detected motion vector is set as the initial deviation vector.

(2) Operation of Embodiment In the above-described configuration, in the motion vector detection device 1, the luminance signal Y of the video signal to be processed (FIG. 2) is the initial value after the frequency characteristic is corrected by the prefilter 2. The displacement vector selection circuit 5 detects the initial displacement vector V1, and the displacement vector ΔV is detected by the gradient method processing in the gradient method computing circuit 6 using the initial displacement vector V1. In the luminance signal Y, the displacement vector ΔV is added to the initial displacement vector V1 to detect the motion vector V. The motion vector V is recorded in the memory circuit 8 and applied to, for example, field interpolation. .

  The motion vector recorded in the memory circuit 8 is an initial displacement vector selection circuit in which a plurality of motion vectors VK1 to VK8 are selected from the memory circuit 8 as candidates for the initial displacement vector V1 in accordance with the subsequent macroblock to be detected. 5, the inter-field difference value is detected for each of the motion vectors VK1 to VK8 in the memory / difference calculation circuit 14 (FIG. 1) in the initial deviation vector selection circuit 5, and the inter-field difference value is the largest. A small motion vector is selected as the initial deflection vector V1.

  Thereby, in this motion vector detection apparatus 1, when the motion is small, the motion vector V can be detected with high detection accuracy by effectively using the correlation between successive fields. However, in this state, when the motion changes suddenly, the motion vector V cannot be detected so as to follow this motion, and the motion vector V is erroneously detected.

  That is, for example, from the state where the object has been moving from left to right with a size of 10 pixels per field, the movement suddenly changed, and from right to left it moved with a size of 5 pixels per field. In this case, in the detected motion vector VX applied to the initial deviation vector, VX = 10. In this case, in the detection of the motion vector by the gradient method in which the motion vector of VX = 10 is applied to the initial displacement vector, the motion of 15 pixels / 1 field must be detected as the displacement vector. However, in the gradient method, since the range of the motion vector that can be detected is narrow (in practice, it is a range of 0 to 3 pixels), the motion vector is erroneously detected for such an object. Several field hours are required to finally follow the movement.

  On the other hand, the block matching method detects a motion vector regardless of the correlation between successive fields, and within a motion vector search range, even if the motion changes suddenly, the motion can be moved within one field. Can be detected. As a result, even when the motion changes rapidly as described above, the motion vector VX = −5 can be detected in the following field from the motion vector VX = 10.

  However, since the block matching method has a small correlation not only between continuous fields but also with surrounding macroblocks, unless the detected motion vector is smoothed between adjacent macroblocks, the format is changed. It cannot be applied to the interpolation process of the converter. However, in such a smoothing process, the movement of the small object is also smoothed to the movement of the surrounding macroblock, and eventually the movement of such a small object cannot be detected.

  Therefore, in this motion vector detection device 1, a motion vector suitable for detecting a motion vector is selected from the candidate INT MVA of the initial deviation vector V1 based on the already detected motion vector V and the motion vector MVB based on block matching. The initial deviation vector V1 is set, thereby effectively avoiding false detection of motion vectors.

  In other words, in the motion vector detection device 1, the motion vector V by the gradient method is set to the initial deviation vector V1 to detect the motion vector V, and the motion is suddenly changed and already detected. When the inter-field difference value by the initial displacement vector V1 becomes larger than the inter-field difference value by the block matching method, the motion vector detected by the block matching method is used as the initial displacement vector instead of the motion vector according to the already detected method. Thus, erroneous detection of the motion vector V is effectively avoided.

  That is, in the motion vector detection device 1, the motion vector detection circuit 4 detects the motion vector MVB by the block matching method, and the difference calculation circuit 12 detects the inter-field difference value DFD B based on the motion vector MVB. Further, the inter-field difference value DFD A is detected for the initial deviation vector candidates detected by the memory / difference calculation circuit 14 from the initial deviation candidate vectors VK1 to VK8.

  In the motion vector detection apparatus 1, the inter-field difference value DFD B related to the block matching method detected in this way and the inter-field difference value DFD A based on the already detected motion vector are input to the difference comparison circuit 13, where The comparison is performed using a predetermined determination reference value α, and thereby, a case where the inter-field difference value DFD A according to the already detected is larger than the inter-field difference value DFD B by the block matching method is detected. In this difference comparison circuit 13, the maximum values MAX VK (VK1 to VK8) of the motion vectors VK1 to VK8 already detected and the motion vector MVB based on the block matching method are respectively determined based on the determination reference values β1 and β2, and thus the motions up to that point are determined. Is small and the movement suddenly increases as a case where the movement changes suddenly.

  Thus, in the motion vector detection device 1, the motion is suddenly changed based on the determination results based on the determination reference values α, β1, and β2 in the difference comparison circuit 13, and the initial displacement vector V1 that has already been detected is used. When the inter-field difference value is large, the motion vector MVB by the block matching method is set as the initial deviation vector.

(3) Advantages of the embodiment According to the above configuration, a motion vector suitable for motion vector detection is selected from the initial deviation vector candidate obtained from the already detected motion vector and the motion vector based on the block matching method. By setting the initial displacement vector, it is possible to reduce erroneous detection of the motion vector as compared with the conventional case.

  More specifically, when the motion has suddenly changed due to the determination of the inter-field difference based on a predetermined determination reference value and the motion vector, and the inter-field difference value is large in the detected initial deviation vector, By setting the motion vector based on the block matching method as the initial deviation vector, it is possible to reduce erroneous detection of the motion vector.

  In the motion vector detection apparatus according to this embodiment, in the difference comparison circuit 13 described above with reference to FIG. 1, whether or not the magnitude of the motion vector MVB by the block matching method is equal to or greater than a predetermined determination reference value β1 and based on the previous detection. It is determined whether the maximum value MAX VK (VK1 to VK8) of the magnitudes of the motion vectors VK1 to VK8 is equal to or less than a predetermined determination reference value β2, and the size of the motion vector MVB by the block matching method is a predetermined determination reference If the maximum value MAX VK (VK1 to VK8) of the detected motion vectors VK1 to VK8 is not less than the predetermined determination reference value β2 and is equal to or greater than the value β1, the initial deviation of the motion vector MVB by the block matching method is determined. The position vector V1 is selected. The motion vector detection apparatus according to this embodiment has the same configuration as that of the motion vector detection apparatus 1 described above with reference to FIGS. 1 and 2 except that the configuration of the difference comparison circuit is different.

In this way, when the motion changes suddenly, the motion vector MVB by the block matching method is selected as the initial displacement vector V1, so that the initial motion can be detected by the block matching method that can detect a good motion vector when the motion amount is large. A deviation vector can be set. As a result, according to the motion vector detection device of this embodiment, the inter-field difference value DFD B related to the block matching method and the inter-field difference value DFD related to the previous detection
Since the configuration related to the process for comparing A with the determination reference value α can be omitted, the configuration of the motion vector detection device can be simplified.

  In the motion vector detection apparatus according to this embodiment, whether or not the inter-field difference value DFD B according to the block matching method is smaller than a predetermined determination reference value α in the difference comparison circuit 13 described above with reference to FIG. It is determined whether or not the inter-field difference value DFD A is greater than the determination reference value α, and the inter-field difference value DFD B according to the block matching method is smaller than the predetermined determination reference value α and the detected inter-field difference When the difference value DFD A is larger than the determination reference value α (DFD A> α> DFD B), the motion vector MVB by the block matching method is selected as the initial deviation vector V1. The motion vector detection apparatus according to this embodiment has the same configuration as that of the motion vector detection apparatus 1 described above with reference to FIGS. 1 and 2 except that the configuration of the difference comparison circuit is different.

  In this way, when it is determined that the reliability of the motion vector MVB by the block matching method is high, the motion vector MVB by the block matching method can be set as the initial deviation vector V1. As a result, according to the motion vector detection device of this embodiment, the magnitude of the motion vector MVB by the block matching method is equal to or greater than the predetermined determination reference value β1 (MVB ≧ β1), and the motion vectors VK1 to VK8 based on the previous detection. Since the maximum value MAX VK (VK1 to VK8) is less than or equal to a predetermined determination reference value β2 (MAX VK (VK1 to VK8) ≦ β2), the configuration relating to the process for determining the motion vector can be omitted. The configuration of the detection device can be simplified.

(1) Configuration of Embodiment FIG. 3 is a block diagram showing an initial deviation vector selection circuit applied to a motion vector detection apparatus according to Embodiment 4 of the present invention. In the motion vector detection apparatus according to this embodiment, the first embodiment has been described above with the exception that the initial displacement vector selection circuit 25 is applied instead of the initial displacement vector selection circuit 5 described above with reference to FIG. The configuration is the same as that of the motion vector detection device 1.

  Here, in the initial deviation vector selection circuit 25 of the motion vector detecting apparatus according to the fourth embodiment, the initial deviation vector selection circuit 26 receives the initial deviation candidate vectors VK1 to VK8 from the memory circuit 8, and these initial deviation vectors. A candidate MV0 (corresponding to INT MVA in the first embodiment) of the initial deviation vector V1 is selected from the rank candidate vectors VK1 to VK8 and output. In other words, the initial deviation vector selection circuit 26 applies initial deviation candidate vectors VK1 to VK8 to the write address used for writing the luminance signal Y1 in units of macroblocks of the luminance signal Y output from the prefilter 2, respectively. Accordingly, the luminance signal Y1 held by the address control in which the read address is deviated is read, and the luminance signal Y1 held thereby is corrected by the initial deviation candidate vectors VK1 to VK8. Further, between the read luminance signal Y1 and the luminance signal Y output from the prefilter 2, an inter-field difference value is calculated for each of the initial deviation candidate vectors VK1 to VK8, and the inter-field difference value is calculated. The smallest initial deviation candidate vector is selected and a candidate MV0 of the initial deviation vector V1 is output. The inter-field difference value DFD0 related to the motion vector MV0 is also output. Note that the difference value between fields related to the initial deviation candidate vectors VK1 to VK8 is 2 of the difference value of each pixel value corresponding to the difference value between fields related to the motion vector MVB although the sum of absolute differences is applied. Multiplication sum may be applied, or the difference value may be simply added.

  The motion amount deviation detection circuit 27 detects a sudden change in motion by comparing the candidate vector MV0 of the initial deviation vector V1 with the motion vector V detected in the immediately preceding macroblock and the corresponding macroblock in the immediately preceding field. . Specifically, the motion amount deviation detection circuit 27 uses the motion vector MV1 detected in the immediately preceding macroblock to correct the motion of the luminance signal Y1 for the immediately preceding macroblock, and performs an interfield difference with the luminance signal Y. The value DFD1 is calculated. Also, using the motion vector MV2 detected in the corresponding macroblock of the immediately preceding field, the luminance signal Y1 is subjected to motion correction and the inter-field difference value DFD2 is calculated with the luminance signal Y for the macroblock of the immediately preceding field. To do.

  Further, when the relational expressions of DFD1-DFD0> α1 and DFD2-DFD0> α2 are both established between these inter-field difference values DFD0 to DFD2, it is determined that the motion has suddenly changed.

  That is, in the candidate MV0 of the initial displacement vector V1, since it is the initial displacement vector V1 according to the gradient method, the motion vector detected in the immediately preceding macroblock and the macroblock corresponding to the immediately preceding field are detected. When the motion vector and the inter-field difference values DFD0 to DFD2 obtained by the motion correction are compared, if the motion does not change suddenly, the inter-field difference value DFD0 is the macroblock corresponding to the immediately preceding macroblock and the immediately preceding field. This is a decrease compared to the detected inter-field difference values DFD1, DFD2. Therefore, when such a relationship cannot be detected using the inter-field difference values DFD0 to DFD2, it can be determined that the motion has suddenly changed. If the detected motion vector is applied to the initial displacement vector V1, You will not be able to follow sudden changes. Therefore, in this case, when the motion vector MVB by the block matching method is set to the initial deviation vector V1, the motion vector can be accurately detected following the sudden change in motion.

  As a result, in this embodiment, when both of the relational expressions described above are satisfied, the motion amount deviation detection circuit 27 determines that the macroblock has a sudden change in motion and outputs the control signal S1. Here, the control signal S1 is a control signal for adding the motion vector MVB by the block matching method and the motion vector generated by processing the motion vector MVB to the selection target of the initial deviation vector. In this embodiment, when both of the two relational expressions described above are satisfied, the movement suddenly changes. However, when one of the above two relational expressions is satisfied as necessary, It is also possible that the movement suddenly changes. Here, the determination reference values α1 and α2 are set to α1≈α2 and α1 = 128 in a macro block of 8 × 8 pixels.

  The low-pass filter (LPF) 30 is a two-dimensional low-pass filter, and smoothes and outputs a motion vector MVB by a block matching method. Therefore, the motion vector MVB detected by the block matching method has a feature that the correlation between adjacent macroblocks is small. In this embodiment, the motion vector MVB is smoothed by the low-pass filter 30 to improve the detection accuracy of the motion vector V. . Here, in this embodiment, a 5 × 5 tap filter in which tap coefficients are set to 1: 2: 2: 2: 1 in the horizontal direction and the vertical direction is applied.

  The low-pass filter 31 is a low-pass filter in the time axis direction, and smoothes and outputs the current field, the previous field of the current field, and the motion vector MVB of the corresponding three macroblocks related to the previous field of the previous field.

  The representative vector generation circuit 32 detects a motion vector MVB by the block matching method that occupies the largest number of peripheral macroblocks to be processed, and outputs it as a representative vector. In this embodiment, the motion vector occupying the most among the motion vectors detected in the immediately preceding field is set as this representative vector.

  The switching circuit 33 is a motion vector MVB based on the block matching method and a motion derived from the motion vector MVB when both of the two relational expressions described above are satisfied by the control signal output from the motion amount deviation detection circuit 27. The motion vectors output from the low-pass filters 30 and 31 and the representative vector generation circuit 32, which are vectors, are output to the initial deviation vector selection circuit.

  The initial deviation vector selection circuit 34 selects and outputs the initial deviation vector V1 from the motion vector output from the switching circuit 33 and the motion vector MV0 output from the motion amount deviation detection circuit 27. Specifically, the initial displacement vector selection circuit 34 is a motion vector suitable for detecting a motion vector by comparing the difference values between fields for these five motion vectors, as in the difference comparison circuit 13 described above with respect to the first embodiment. Is selected and output by the initial deviation vector.

  According to the above configuration, the initial deviation vector candidate MV0 is selected from the already detected motion vectors, the macro block whose motion is suddenly changed by this candidate is detected, and the initial deviation is detected only in the macro block whose motion is suddenly changed. By setting a motion vector by block matching as a vector selection target, it is possible to detect a motion vector with high accuracy following a sudden change in motion.

  In addition, by detecting a macroblock whose motion is suddenly changed by such an initial deviation vector candidate MV0, specifically, the inter-field difference value DFD0 obtained by motion correction by this candidate and the motion in the immediately preceding macroblock. By detecting a macroblock whose motion has suddenly changed by determining a difference value from the inter-field difference value DFD1 by the correction based on a predetermined determination reference value α1, it is possible to effectively use the correlation between macroblocks that are horizontally continuous. A sudden change in motion can be detected.

  Further, by detecting a macroblock whose motion has changed suddenly by such an initial deviation vector candidate MV0, specifically, the inter-field difference value DFD0 obtained by correcting the motion by this candidate corresponds to the immediately preceding field. By utilizing the correlation between successive fields effectively by detecting a macroblock whose motion has changed suddenly by determining a difference value from the inter-field difference value DFD2 by motion correction in the macroblock by a predetermined determination reference value α2. A sudden change in motion can be detected.

  Further, by adding not only the motion vector MVB by the block matching method but also a motion vector derived from this motion vector MVB to the selection target of the initial deviation vector V1, the detection accuracy of the motion vector can be further improved. it can.

  In other words, the motion vector MVB by the block matching method is smoothed by a two-dimensional low-pass filter and added to the selection target, thereby correcting the disadvantage that the correlation between adjacent macroblocks is small in the motion vector MVB by the block matching method. Thus, the motion vector can be detected.

  In addition to the processing result by the two-dimensional low-pass filter, the motion vector MVB by the block matching method is smoothed by the low-pass filter in the time axis direction and added to the selection target, thereby smoothing by the two-dimensional low-pass filter 30. Depending on the situation, even when a macroblock whose motion has changed rapidly cannot be detected, the motion vector V1 can be detected following the change in motion. Specifically, when the direction of the motion vector changes, such as a rotation pattern, the motion vector smoothed in the time axis direction is closer to the actual motion amount than the processing result by the two-dimensional low-pass filter. Thus, in such a case, the initial displacement vector based on the motion vector smoothed in the time axis direction is selected, and the motion vector can be detected with high accuracy.

  In addition, by adding a motion vector that occupies most of the detected motion vectors as a representative vector to the selection target, for example, in a scene where a new image appears from the background, the representative vector can quickly approach the actual motion amount. In this case, the representative vector is selected and the motion vector can be detected with high accuracy.

  Thus, in this embodiment, it is possible to detect a motion vector with high accuracy following a sudden change in motion.

  In the above-described embodiment, the case where the deviation vector is detected by the gradient method has been described. However, the present invention is not limited to this, and the deviation vector is detected by another method such as a phase correlation method. Also good.

  In the above-described embodiment, the case where the initial deviation candidate vector is selected from the eight types of candidates has been described. However, the present invention is not limited to this, and the case where the initial deviation candidate vector is selected from a plurality of candidates is widely used. Can be applied.

  In the above-described embodiment, the case where the motion vector is detected between the fields is described. However, the present invention is not limited to this, and the motion vector may be detected between the frames. In this case, a 1-frame delay circuit may be applied instead of the 1-field delay circuit 3 so that the luminance signal Y is delayed by 1 frame.

  The present invention relates to a motion vector detection device and a motion vector detection method, and can be applied to, for example, a format conversion device.

It is a block diagram which shows the initial deviation vector selection circuit of the motion vector detection apparatus which concerns on Example 1 of this invention. It is a block diagram which shows the motion vector detection apparatus which concerns on Example 1 of this invention. It is a block diagram which shows the initial deviation vector selection circuit of the motion vector detection apparatus which concerns on Example 4 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Motion vector detection apparatus, 3 ... 1 field delay circuit, 4 ... Motion vector detection circuit, 5 ... Initial deflection vector selection circuit, 6 ... Gradient method arithmetic circuit, 7 ... Addition circuit, 8 ... ... Memory circuit, 12 ... Difference calculation circuit, 13 ... Difference comparison circuit, 14 ... Memory / Difference calculation circuit, 25 ... Initial deviation vector selection circuit

Claims (10)

By detecting an initial deviation vector from the already detected motion vector, detecting the deviation vector using the initial deviation vector, and correcting the initial deviation vector, the motion vector of the sequentially input video signal is obtained. In the motion vector detection device to detect,
Motion vector detecting means for detecting a motion vector of the video signal by a block matching method and outputting a motion vector by the block matching method;
A difference value calculation means according to a block matching method for detecting a difference value between frames or a difference value between fields according to the block matching method by performing motion correction by a motion vector according to the block matching method;
Computing means for detecting a candidate for the initial deviation vector from the detected motion vector;
A difference value calculation means for the detected motion vector for detecting the interframe difference value or the interfield difference value for the detected motion vector by performing motion correction with the initial deviation vector candidate;
The initial deviation vector candidate is output as the initial deviation vector, and when a predetermined condition is satisfied, a motion vector obtained by the block matching method is used as the initial deviation vector instead of the initial deviation vector candidate. Selecting means for selecting and outputting,
When the predetermined condition is satisfied,
The inter-frame difference value or inter-field difference value related to the detected motion vector is larger than a predetermined determination reference value, and the inter-frame difference value or inter-field difference value related to the block matching method is equal to or less than the determination reference value This is a motion vector detection device. By detecting an initial deviation vector from the already detected motion vector, detecting the deviation vector using the initial deviation vector, and correcting the initial deviation vector, the motion vector of the sequentially input video signal is obtained. In the motion vector detection device to detect,
Motion vector detecting means for detecting a motion vector of the video signal by a block matching method and outputting a motion vector by the block matching method;
Calculating means for selecting an initial deviation candidate vector from the detected motion vector, and detecting the initial deviation vector candidate from the selected initial deviation candidate vector;
The initial deviation vector candidate is output as the initial deviation vector, and when a predetermined condition is satisfied, a motion vector based on the block matching method is used as the initial deviation vector instead of the initial deviation vector candidate. Selecting means for selecting and outputting,
When the predetermined condition is satisfied,
A second determination in which a magnitude of a motion vector by the block matching method is equal to or larger than a first determination reference value and a maximum value of the initial deviation candidate vector is smaller than the first determination reference value; A motion vector detection device characterized by being a case where the value is below a reference value. By detecting an initial deviation vector from the already detected motion vector, detecting the deviation vector using the initial deviation vector, and correcting the initial deviation vector, the motion vector of the sequentially input video signal is obtained. In the motion vector detection device to detect,
Motion vector detecting means for detecting a motion vector of the video signal by a block matching method and outputting a motion vector by the block matching method;
A difference value calculation means according to a block matching method for detecting a difference value between frames or a difference value between fields according to a block matching method by performing motion correction by a motion vector according to the block matching method;
Calculating means for selecting an initial deviation candidate vector from the detected motion vector, and detecting the initial deviation vector candidate from the selected initial deviation candidate vector;
A difference value calculation means related to a detected motion vector that detects a difference value between frames or a difference value between fields related to a detected motion vector by performing motion correction using the initial deviation vector candidates;
The initial deviation vector candidate is output as the initial deviation vector, and when a predetermined condition is satisfied, a motion vector based on the block matching method is used as the initial deviation vector instead of the initial deviation vector candidate. Selecting means for selecting and outputting,
When the predetermined condition is satisfied,
When the inter-frame difference value or inter-field difference value related to the detected motion vector is larger than a predetermined determination reference value, and the inter-frame difference value or inter-field difference value related to the block matching method is equal to or less than the determination reference value Because
And the magnitude of the motion vector according to the block matching method is equal to or greater than a first determination reference value, and the maximum value of the initial deviation candidate vector is smaller than the first determination reference value. A motion vector detection apparatus characterized by being a case where the value is equal to or less than a determination reference value. By detecting an initial deviation vector from the already detected motion vector, detecting the deviation vector using the initial deviation vector, and correcting the initial deviation vector, the motion vector of the sequentially input video signal is obtained. In the motion vector detection device to detect,
Motion vector detecting means for detecting a motion vector of the video signal by a block matching method and outputting a motion vector by the block matching method;
Selection output means for detecting and outputting the initial deviation vector from the detected motion vector;
Detecting means for detecting a macroblock whose movement is suddenly changed from the video signal;
A motion vector according to the block matching method is added to a selection target of the initial deviation vector in the selection output means in the macro block whose motion has suddenly changed based on a detection result by the detection means. Detection device. The detection means includes
An inter-field difference value obtained by correcting the motion of the immediately preceding field with the detected motion vector related to the selection of the initial deviation vector, and an inter-field difference value detected with the detected motion vector in the immediately preceding macroblock The motion vector detection device according to claim 4, wherein when the difference value is greater than a predetermined determination reference value, it is determined that the motion has suddenly changed. The detection means includes
The inter-field difference value obtained by correcting the motion of the immediately preceding field with the previously detected motion vector related to the selection of the initial deviation vector and the field detected with the motion vector already detected in the corresponding macroblock of the immediately preceding field The motion vector detection device according to claim 4 or 5, wherein when the difference value from the difference value is larger than a predetermined determination reference value, it is determined that the motion has suddenly changed. Adding a motion vector generated from the motion vector based on the block matching method to the selection target of the initial displacement vector in the selection output means in the macro block whose motion has suddenly changed; The motion vector detection device according to claim 4, 5, or 6. The motion vector detection according to claim 7, wherein the motion vector generated from the motion vector by the block matching method is a motion vector obtained by smoothing the motion vector by the block matching method by a two-dimensional low-pass filter. apparatus. The motion vector generated from the motion vector by the block matching method is a motion vector obtained by smoothing the motion vector by the block matching method in a time axis direction by a low-pass filter. The motion vector detection device described. The motion vector detection device according to claim 7, wherein the motion vector generated from the motion vector by the block matching method is a motion vector that occupies most in the immediately preceding field.

Images