JP2010004255A - Interpolation frame creating apparatus, interpolation frame creating method and interpolation frame creating program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an interpolation frame creating apparatus which forms a motion vector to be used for creating an interpolation frame into the one suited for creating an interpolation frame indicating a more local movement. <P>SOLUTION: The reliability of a motion vector assigned to a block and that thereof on the stationary side in the case assuming that the block is stationary are calculated, and the vector is corrected by weighing according to these ratios. Thus, the contribution of correction of the motion vector on the stationary side with reliability higher than or equivalent to that of the motion vector is reduced among peripheral motion vectors including the motion vector to be corrected, thereby making it possible to obtain the motion vector. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an interpolation frame creation device, an interpolation frame creation method, and an interpolation frame creation program that improve the temporal resolution of a moving image.

  The frame interpolation technique is a technique for increasing the temporal resolution of a moving image by creating an interpolation frame and interpolating between the display times of two frames preceding and following the display time.

  There is a method of dividing an interpolation frame into rectangular areas called blocks (hereinafter referred to as interpolation blocks), assigning motion vectors (hereinafter referred to as interpolation MV) to each interpolation block, and creating an interpolation frame.

  In Patent Document 1, as a method of deriving an interpolation MV, a vector indicating local motion is assigned by assigning a motion vector that has been corrected by smoothing based on a motion vector assigned to surrounding blocks as an interpolation MV. Assigning techniques have been proposed. However, in an image in which a telop such as a character moves in the background, a block including a part of the character on the block boundary has a high correlation with the background. Therefore, there is a high possibility that a motion vector in a block including a part of characters on the block boundary is erroneously derived.

As a method of correcting the derived vector, there is a method as described in Non-Patent Document 1. Non-Patent Document 1 is a technique for applying a weighted vector median filter process by giving a weight to a highly correlated vector. However, when most of the peripheral blocks are still areas, the correlation between images indicated by the zero vectors assigned to the peripheral blocks is high. In such a case, in the method of Non-Patent Document 1, there is a high possibility that the motion vector assigned to the interpolation block including a part of the character on the block boundary is still a zero vector even by the correction process, and the image is broken. It will occur.
JP 2006-279917 A L. Alparone, M .; Barni, F.A. Bartolini, V.M. Capellini “Adaptive weighted vector-media filters for motion-fields smoothing”, ICASSP, pp 2267-2270, 1996.

  The present invention has been made to solve the above-described problem. The reliability of a motion vector assigned to an interpolation block and the reliability of a predetermined vector are calculated for the interpolation block, and the ratio between the two reliability levels is calculated. An object of the present invention is to provide an interpolation frame creation apparatus, method, and program capable of creating an interpolation image using a highly reliable vector by performing weighted vector correction processing accordingly.

  In order to solve the above-mentioned problem, the present invention provides an interpolation frame creating apparatus for interpolating an interpolation frame between a first frame and a second frame of an input moving image, and a plurality of the interpolation frames are divided. Motion estimation means for allocating a first motion vector connecting the first region in the first frame and the second region in the second frame to the target interpolation block to be processed among the interpolation blocks; A first reliability calculating means for obtaining a first reliability representing a level of correlation between the first area and the second area of the target interpolation block; and the first motion in the target interpolation block. When a predetermined vector is assigned instead of a vector, a high correlation between the area in the first frame indicated by the start point of failure prevention and the area indicated by the end point of the predetermined vector in the second frame The first 2nd reliability calculation means for obtaining a reliability of 2, and the first motion based on the first and second reliability of each of the target interpolation block and the peripheral blocks that are the surrounding interpolation blocks Correction means for obtaining a second motion vector obtained by correcting a vector, and creation means for creating the interpolation frame based on the pixel values of the first frame and the second frame using the second motion vector And an interpolation frame creation device characterized by comprising:

  Also provided are an interpolation frame creation method using the same apparatus and an interpolation frame creation program that causes a computer to implement the method.

    According to the present invention, a motion vector with higher reliability can be assigned to an interpolation block, and a high-quality interpolation frame can be created.

  Hereinafter, embodiments of the present invention will be described.

(First embodiment)
The first embodiment will be described below. Of the two input images in which the display times of the input moving images are continuous, the frame with the previous display time is described as the reference frame, and the subsequent frame is described as the target frame. An example in which an interpolation frame is generated between the reference frame and the target frame will be described below.

FIG. 1 is a diagram showing the display timing of the input image and the display timing of the interpolation frame. Of the input image, the frame displayed at time t is the reference frame I _src , and the frame displayed at time t + 1 is the target frame I _dst . The interpolated frame I _int is a frame that is interpolated and displayed at t + Δt (where 0 <Δt <1) between these display times.

  FIG. 2 is a block diagram of the interpolation frame creation apparatus according to the first embodiment.

  The interpolated frame creation apparatus of this embodiment includes a motion estimation unit 101, a frame memory 102, a motion vector correction unit 103, an interpolation image generation unit 104, and a switching unit 105.

The motion estimation unit 101 divides the interpolation frame I _int into rectangular blocks of M ₁ × M ₂ (M ₁ , M ₂ ) pixels set in advance, and assigns motion vectors to the interpolation blocks. Hereinafter, a method for deriving the motion vector u (i) using the block matching method, which is one method of motion estimation, will be described.

FIG. 3 is a diagram for explaining a method of assigning a motion vector to the interpolation block i. This vector x (i) is a vector indicating the upper left position of the interpolation block. The motion vector u (i) of the interpolation block i is obtained based on Expression (1).

Here, B (i) is B (i) = {(i + (x, y)) ^T | 0 ≦ x <M ₁ , 0 ≦ y <M ₂ } M ₁ × M ₂ A set of vectors is shown. I _dst (x) indicates a pixel value at a position x in the target frame. I _src (x) represents a pixel value at a position x in the reference frame. The motion search area is denoted by W.

For the interpolated block i, a block matching error based on the difference square sum criterion between the block in the target frame I _dst indicated by the vector s and the reference block in the reference frame I _src is obtained as a correlation value. A block in the target frame I _dst (hereinafter referred to as the first block B ₁ (i)) and a block in the reference frame I _src when the vector s is changed in the search region W and the matching error is minimized. (Hereinafter referred to as the first block B ₂ (i)) is assigned to the interpolation block i as a motion vector u (i).

Also, u (i) may be derived based on the sum of absolute values of the difference values of the pixel values between blocks shown in the following equation (2).

As a method for estimating a motion vector, other optical flow methods, Pel-recursive, Bayesian methods, and the like may be used instead of the above-described block matching method.

  The frame memory 102 acquires and stores an input image.

  The motion vector correction unit 103 acquires the motion vector u (i) for each interpolation block obtained by the motion estimation unit 101, calculates the motion vector reliability and the failure prevention vector (predetermined vector) reliability, and calculates the motion vector u. The correction process (i) is performed. A reliability calculation method and vector correction processing will be described later. In the present embodiment, a case will be described in which the failure prevention vector (predetermined vector) is a zero vector in all interpolation blocks. This failure prevention vector may be an average vector of motion vectors for one frame obtained by the motion estimation unit 101. The corrected motion vector, motion vector reliability, and failure prevention vector reliability are output to the interpolated image generation unit 104.

  The interpolated image generation unit 104 acquires the motion vector, motion vector reliability, and failure prevention vector reliability from the motion vector correction unit 103, performs interpolation frame generation processing, and outputs the interpolation frame to the switching unit 105. A method by which the interpolation image generation unit 104 generates an interpolation frame will be described later.

  The switch 105 connects the switch of the switching unit 105 to the frame memory 102 side at the timing of displaying the input image, acquires the input image from the frame memory 102, and outputs it. Further, at the timing of displaying the interpolation frame, the switch of the switching unit 105 is connected to the interpolation frame creation unit 104 side, the interpolation frame is obtained from the interpolation frame creation unit 104, and the display image is output.

  Next, the motion vector correction unit 103 will be described in detail.

  FIG. 4 is a block diagram in which details of the motion vector correction unit 103 of the interpolation frame creation apparatus are extracted.

  The motion vector correction unit 103 includes a motion vector storage memory 201, a motion vector reliability calculation unit 202, a motion vector reliability storage memory 203, a failure prevention vector reliability calculation unit 204, a failure prevention vector reliability storage memory 205, and a switching unit 206. And a reliability ratio motion vector correction unit 207.

  The motion vector storage memory 201 stores a motion vector for each interpolation block.

  The motion vector reliability calculation unit 202 extracts a motion vector u (i) for one frame from the motion vector storage memory 201, and how appropriate these motion vectors are as motion vectors used when interpolating an interpolation frame. The motion vector reliability indicating whether or not it is derived is derived by the following method.

The reliability of the image to be interpolated when the motion vector v is assigned to the interpolation block i is denoted as α (i, v). The reliability α (i, v) is an index indicating the level of pixel correlation between the block in the target frame I _dst and the block in the reference frame I _src connected by the motion vector v. The reliability α (i, v) can be calculated from the equation (3).

Here, ε is a safety constant for preventing positive division by zero. E (i, v) is derived from Equation (4) based on the sum of absolute differences.

Represented by

Note that E (i, v) may be obtained based on the sum of squared differences of pixels shown in Equation (5).

Further, α (i, v) may be derived not from the equation (4) but from the equation (6) using an exponential function.

Here, σ _I represents the standard deviation of noise on the image signal of the input image.

  The motion vector reliability calculation unit 202 interpolates when the motion vector u (i) obtained by the motion estimation unit 101 is assigned to the interpolation block i by substituting u (i) into v in Expression (3). Α (i, u (i)) indicating the reliability of the image is derived.

  The motion vector reliability storage memory 203 stores the motion vector reliability α (i, u (i)) for each interpolation block derived by the motion vector reliability calculation unit 202.

  The failure prevention vector reliability calculation unit 204 calculates a failure prevention vector reliability indicating the reliability of the failure prevention vector which is a predetermined vector. As described above, in this embodiment, the failure prevention vector is set to a zero vector. The failure prevention vector reliability is zero in v of any one of the equations (3), (4), or (6), as in the case of deriving the motion vector reliability α (i, u (i)). Substituting the vector, the reliability α (i, 0) of the bankrupt prevention vector is derived.

  The failure prevention vector reliability storage memory 205 stores the failure prevention vector reliability α (i, 0) for each interpolation block i derived by the failure prevention vector reliability calculation unit 204.

  The reliability ratio motion vector correction unit 207 acquires the motion vector u (i), the motion vector reliability α (i, u (i)), and the failure prevention vector reliability α (i, 0), respectively, and α (i , u (i)) and α (i, 0) are calculated, and a vector filtering process weighted by the ratio is performed. A detailed description of the filter process will be described later.

  Next, with respect to the motion vector (corrected motion vector) obtained by the filter processing, a motion vector reliability (corrected motion vector reliability) is derived by a method similar to the processing performed by the motion vector reliability calculation unit 202, If the corrected motion vector reliability is higher than the motion vector reliability input from the motion vector reliability storage memory 203, the corrected motion vector reliability is written in the motion vector reliability storage memory 203. Also, the corrected motion vector is written into the motion vector storage memory 201.

  The switcher 206 acquires from the reliability ratio motion vector correction unit 207 an index indicating how many remaining motion vector correction processes are performed for a preset number of times, and if itr is non-zero, the motion The vector u (i), the motion vector reliability α (i, u (i)), and the failure prevention vector reliability α (i, 0) are connected to the reliability ratio motion vector correction unit 207 side. On the other hand, when itr is zero, the above u (i), α (i, u (i)), α (i, 0) are connected to the interpolated image generation unit 104 side. The initial value of itr is an integer value that can be set by the user, and is held in the reliability ratio motion vector correction unit 207.

  FIG. 5 is a flowchart showing processing of the motion vector correction unit 103.

  First, loop processing that loops for each interpolation block i in the interpolation frame is performed (step 401). In the loop, the motion vector u (i) is acquired for each interpolation block i, and the motion vector reliability calculation unit 202 calculates the motion vector reliability α (i, u (i)) (step 402). The failure prevention vector reliability calculation unit 204 calculates the failure prevention vector reliability α (i, 0) for each interpolation block i (step 403).

  Next, the switching unit 206 determines whether or not itr is non-zero (step 404).

  When itr is not zero (step 404, No), loop processing for looping over the interpolation block i is performed (step 405). In the loop, the reliability ratio motion vector correction unit 207 performs the correction process of the motion vector u (i) for each interpolation block i (step 406). When the loop is completed and correction processing is performed on all the interpolation blocks i in the interpolation frame, itr is decremented (step 407) and the process returns to step 404. When itr becomes zero (step 404, Yes), the filtering process by the reliability ratio motion vector correction unit 207 is finished, and the switching unit 206 determines the corrected motion vector, the failure prevention vector reliability, and the corrected motion vector reliability. The degree is output to the interpolation frame creation unit 104.

  Next, the reliability ratio motion vector correction unit 207 will be described in detail.

FIG. 6 is a block diagram illustrating details of the reliability ratio motion vector correction unit 207 of the motion vector correction unit 103.

  The reliability ratio motion vector correction unit 207 includes a weighted vector filter unit 301 and an update unit 302.

The weighting vector filter unit 301 performs a vector median filter process using the ratio between the input motion vector reliability α (i, u (i)) and the failure prevention vector reliability α (i, 0) as a weight. A vector after correction processing for the motion vector u (i) of the interpolation block i is derived by Expression (7).

The norm shown in Equation (7) is the value shown in Equation (9) when norm = 2. Note that x = (x ₁ , x ₂ ) ^T.

Represents. When norm = 1, the value is as shown in equation (9).

It is.

Further, w (i + s) is derived from Expression (10). This is a weighting coefficient based on the ratio between the motion vector reliability α (i + s, u (i + s)) and the failure prevention vector reliability α (i + s, 0).

  Here, the set N is a set of interpolation blocks in the vicinity including the interpolation block i. For example, the set N represents nine blocks including the interpolation block i and the surrounding interpolation blocks as shown in FIG. ε is a safety constant for preventing positive division by zero.

The weighting vector filter unit 301, for each of a plurality of interpolation blocks included in the motion vector set N assigned to the set N, the motion vector u (i + c) assigned to the interpolation block i + c to be calculated. ) And the norm of the difference vector between the motion vector u (i + s) assigned to the surrounding interpolation block i + s (s∈N) in the set N, and the weighting coefficient shown in the equation (10) Find each multiplied value and find the sum of them. The motion vector u (i + c _min ) assigned to the interpolation block c _min to be calculated that minimizes the obtained value is assigned as the corrected motion vector u ′ (i) of the interpolation block i.

Further, w (s) may be a value obtained by multiplying the weighting coefficient by the ratio of the sum of absolute differences between the target frame I _dst and the I _src pixel value of the reference frame, as shown in Expression (11).

However, w '(s) here is a value defined by Formula (12).

SAD _Δt is defined by the equation (13).

Note that SAD _Δt may be defined by the sum of squared differences of pixel values shown in Expression (14).

  That is, a value obtained by multiplying w ′ (s) calculated by Expression (12) and Expression (10) is used as the weighting coefficient w (i + s).

In the above-described embodiment, the vector filter process is performed based on the equation (7). However, the vector filter process may be performed by an averaging process shown in the equation (15).

Here, T (x) is an operator for counting the number of elements of x. Further, w ″ (c) is a value obtained by normalizing w (c) described in Expression (4) or Expression (8), and is a value represented by Expression (16).

It is.

  Next, the update unit 302 will be described in detail.

  FIG. 8 is a flowchart showing the processing of the update unit 302.

  The update unit 302 acquires the corrected motion vector u ′ (i) from the weighted vector filter unit 301 and determines whether or not it is the same vector as the motion vector u (i) (step 501). If they are the same vector (step 501, Yes), the process is terminated. If they are different vectors (No in step 501), the motion vector reliability α (i, u ′ (i)) of u ′ (i) is calculated (step 502).

  Next, α (i, u ′ (i)) is compared with the motion vector reliability α (i, u (i)) of the motion vector u (i) before correction (step 503). If α (i, u ′ (i))> α (i, u (i)) and the corrected motion vector is more reliable (step 503, Yes), α (i, u ′ ( i)) is overwritten in the motion vector reliability storage memory 203 (step 504). Similarly, u ′ (i) is overwritten on the motion vector storage memory 201 as a motion vector u (i), and the process is terminated. (Step 505). If α (i, u ′ (i)) <α (i, u (i)) (step 503, No), the process is terminated.

Hereinafter, a method in which the interpolation image generation unit 104 creates the interpolation frame I _int will be described in detail.

  The interpolated image generation unit 104 performs motion vector u (i), motion vector reliability α (i, u (i)), and failure prevention vector reliability α (i, i, i) stored in the motion vector storage memory for each interpolation block i. 0) and the pixel value belonging to the block i of the interpolation frame is determined by the weighted average processing using the above two reliability values for the pixel values of the target frame and the reference frame specified from the motion vector and the failure prevention vector. .

  The interpolation pixel at pixel position xεB (i) in the first block is determined as follows.

In this embodiment, since the zero vector is set as the failure prevention vector, the still pixel I _still is generated as shown in Expression (17). Note that x satisfies x∈B (i) as shown in the equation (1).

Here, the reference frame I _src is used as it is, but for example, an average image obtained by averaging pixel values of the reference frame I _src and the target frame I _dst may be used as in Expression (18).

Next, a motion compensation pixel I _mc is generated by shifting the pixel by the motion vector u (i).

Similarly, the average image shown in the equation (20) may be generated as the motion compensation pixel _Imc instead of the equation (19).

The reliability corresponding to the still pixel I _still (x) is the reliability α (i, 0). The reliability corresponding to the motion compensation pixel I _mc (x) is the reliability α (i, u (i)). Considering these α (i, 0) and α (i, u (i)) as weights, the interpolation frame I _int can be generated by the weighted average shown in Expression (21).

Here, when α (i, 0) = 0 and α (i, u (i)) = 0, the weighted average becomes indefinite, so in that case I _int (x) = 0.5 I _mc (x) +0.5 I _still (x). Alternatively, the interpolated frame pixel I _int (x) may be generated as I _int (x) = I _still (x).

The above processing is performed for all the interpolation blocks in the interpolation frame, and an interpolation frame I _int is generated and output.

  As described above, the interpolation frame creating apparatus of the present embodiment performs weighting by the reliability ratio as shown in Expression (10), so that among the motion vectors of the interpolation block to be corrected and the surrounding interpolation blocks, It is possible to reduce the contribution in the correction processing of the motion vector whose reliability on the stationary side is higher than the reliability of the motion vector.

  FIG. 9 shows an input image frame and an interpolation frame in which the character telop ABC is translated in the upward direction on the screen. The broken line represents the boundary when the image is divided into blocks. Moreover, the arrow has shown the motion vector of the block containing the starting point.

  In the prior art, an interpolation block (hereinafter referred to as a block A ′) that should interpolate an image including a part of “A” in FIG. 9 is assigned a zero vector as a motion vector candidate in the motion estimation process. Since the matching error is small, there is a high possibility that the zero vector is assigned as it is without reflecting a part of the movement of “A”. At this time, if an interpolation frame is created using the zero vector as a motion vector, an interpolation frame in which a part of the character is broken as shown in FIG.

  For the input image shown in FIG. 9, if a correction process for assigning the motion vector of the block B ′ to the block A ′ is possible, a higher quality interpolation frame can be obtained.

  The interpolation frame creation apparatus according to the present embodiment performs the correction process represented by Expression (7) using the reliability ratio of the zero vector that is the failure prevention vector and the assigned motion vector as a weight, as represented by Expression (10). . As a result, in the peripheral blocks other than the block B ′, the failure prevention vector reliability and the motion vector reliability are both high, so that the weighting coefficient shown in the equation (10) is almost 1. On the other hand, in the block B ′, the failure prevention vector reliability decreases and the motion vector reliability increases. Therefore, since the motion vector assigned to the block B ′ is subjected to a correction process with a large weight, the motion vector of the block A ′ is corrected to a more preferable motion vector, and the interpolation frame shown in FIG. More likely to be created.

  On the other hand, in the interpolation block on the right side of the block B ′ (hereinafter referred to as the block C ′), correction processing that is erroneously assigned is performed because the motion vector weight of the block B ′ is large for the same reason as above. There is. However, in the block C ′, there is a ☆ pattern that does not move between frames. In that case, since the reliability of the corrected motion vector is smaller than the motion vector before the correction, it is not updated by the updating unit 302 as the motion vector assigned to the block C ′.

By reducing the contribution in motion vector correction processing that has the same or higher reliability on the stationary side than the reliability of the motion vector among the surrounding motion vectors including the motion vector to be corrected as described above. Therefore, a higher quality motion vector can be obtained.

In this embodiment, the case where the zero vector is set as the failure prevention vector has been described. However, when the entire frame has motion, the average vector of the motion vectors assigned to the interpolation block is used as the failure prevention vector. May be equal.

Further, in the above-described embodiment, the description has been given of the case where, among the interpolation frames to be created, the frame whose display time is earlier is the reference frame, and the frame that is later is the target frame. The target frame may be the target frame.

The figure which shows the display timing of an input image and an interpolation frame. The block diagram which shows the interpolation frame production apparatus of 1st Embodiment. The figure for demonstrating the method of assigning a motion vector to the interpolation block i. The block diagram which shows the detail of the motion vector correction | amendment part 103 of the interpolation frame production apparatus. The block diagram which shows the detail of the reliability ratio motion vector correction | amendment part 207 of the motion vector correction | amendment part 103. FIG. 5 is a flowchart showing processing of a motion vector correction unit 103. Diagram showing interpolation block i and surrounding interpolation blocks The flowchart which shows the process of the update part 302. FIG. The figure which shows the image which the character telop is moving parallel to the screen upper direction.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 ... Motion estimation part 102 ... Frame memory 103 ... Motion vector correction part 104 ... Interpolation image generation part 105 ... Switching part

201 ... motion vector storage memory 202 ... motion vector reliability calculation unit 203 ... motion vector reliability storage memory 204 ... failure prevention vector reliability calculation unit 205 ... failure prevention vector reliability storage memory 206: switching unit 207 ... reliability ratio motion vector correction unit

Claims (7)

In an interpolation frame creating apparatus for interpolating an interpolation frame between a first frame and a second frame of an input moving image,
Of the plurality of interpolation blocks obtained by dividing the interpolation frame, the first interpolation in the first frame and the second region in the second frame are connected to the target interpolation block to be processed. Motion estimation means for assigning motion vectors;
First reliability calculation means for obtaining a first reliability indicating a level of correlation between the first area and the second area indicated by the first motion vector assigned to the target interpolation block When,
When a predetermined vector is assigned to the target interpolation block instead of the first motion vector, the predetermined vector indicates a correlation between the area in the first frame and the area in the second frame. A second reliability calculation means for obtaining a second reliability representing the height;
Correction means for obtaining a second motion vector obtained by correcting the first motion vector based on the first and second reliability levels of the target interpolation block and peripheral blocks that are neighboring interpolation blocks;
Creating means for creating the interpolation frame based on the pixel values of the first frame and the second frame using the second motion vector;
An interpolated frame creating apparatus comprising:
The correction means weights the first motion vector of each of the target interpolation block and the peripheral blocks which are the interpolation blocks around the target interpolation block by a ratio between the first reliability and the second reliability. The one of the first motion vectors of the target interpolation block and the peripheral blocks is obtained as the second motion vector of the target interpolation block based on the weighted vector obtained. Interpolation frame creation device.
When the second motion vector is assigned to the target interpolation block instead of the first motion vector, the region in the first frame indicated by the second motion vector and the second frame in the second frame A third confidence level representing the level of correlation with the area is obtained, and when the third confidence level is lower than the first confidence level, the first motion vector is used as the second motion vector. Update means to update
The interpolation frame creation device according to claim 2, further comprising:
The correction means selects (a) one interpolation block as a first interpolation block for the second motion vector of each of the target interpolation block and the peripheral block as the averaging process,
(B) determining a first weighting factor based on a ratio between the first reliability and the second reliability obtained from the second motion vector assigned to the first interpolation block;
(C) obtaining a norm of a difference vector between the first motion vector assigned to the first interpolation block and the first motion vector assigned to an interpolation block other than the first interpolation block;
(D) obtaining a sum of first values obtained by multiplying the norm by the first weighting factor;
4. The first motion vector assigned to the first interpolation block that minimizes the sum of the first values is assigned as the second motion vector of the target interpolation block. The interpolation frame creation apparatus described.
The interpolation frame creation apparatus according to any one of claims 1 to 4, wherein a zero vector is set as the predetermined vector.
In an interpolation frame creation method for interpolating an interpolation frame between a first frame and a second frame of an input moving image,
Of the plurality of interpolation blocks obtained by dividing the interpolation frame, the first motion connecting the first region in the first frame and the second region in the second frame to the target interpolation block to be processed. Assigning a vector;
Obtaining a first confidence level indicating a level of correlation between the first region and the second region of the target interpolation block;
When a predetermined vector is assigned to the target interpolation block instead of the first motion vector, the region in the first frame indicated by the start point of the first failure prevention and the predetermined vector in the second frame Obtaining a second confidence level indicating the level of correlation with the region indicated by the end point of the vector;
Obtaining a second motion vector obtained by correcting the first motion vector based on the first and second reliability of each of the target interpolation block and peripheral blocks which are neighboring interpolation blocks;
Using the second motion vector to create the interpolated frame based on pixel values of the first frame and the second frame;
An interpolated frame creation method characterized by comprising:
In an interpolation frame creation program for interpolating an interpolation frame between a first frame and a second frame of an input moving image,
Of the plurality of interpolation blocks obtained by dividing the interpolation frame, the first motion connecting the first region in the first frame and the second region in the second frame to the target interpolation block to be processed. A motion estimation function that assigns vectors;
A first reliability calculation function for obtaining a first reliability representing a level of correlation between the first area and the second area of the target interpolation block;
When a predetermined vector is assigned to the target interpolation block instead of the first motion vector, the region in the first frame indicated by the start point of the first failure prevention and the predetermined vector in the second frame A second reliability calculation function for obtaining a second reliability indicating the degree of correlation with the region indicated by the end point of the vector;
A correction function for obtaining a second motion vector obtained by correcting the first motion vector based on the first and second reliability of each of the target interpolation block and peripheral blocks which are neighboring interpolation blocks;
A creation function for creating the interpolation frame based on pixel values of the first frame and the second frame using the second motion vector;
An interpolated frame creation program characterized by causing a computer to realize the above.

Images