JPH077721A - Movement corrected video signal processing system - Google Patents

Abstract

PURPOSE: To generate a further exact moving vector by providing a means for generating plural moving vectors, means for detecting whether those moving vectors are larger or smaller than a prescribed value, and means for limiting an output picture. CONSTITUTION: A movement processor generates plural moving vectors indicating picture movement between a pair of input pictures of an input video signal for using it for the interpolation of the output picture of an output video signal. Then, whether each moving vector is larger or smaller than a prescribed size is detected by an amplitude detector 400. Also, at the time of interpolating an output picture, the use of the moving vector detected as that larger than the prescribed size is limited by a moving vector use limiter 450. A wide area vector detector 330 sets a corresponding false signal flag. The contribution of an input local moving vector whose reliability flag indicates that the vector fails in a reliability test to a wide band vector can be prevented by preventing the increment of a frequency array 420 to a long vector.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion compensation video signal processing system (apparatus and method). The motion compensation video signal processing device is used for applications such as conversion of a television standard system, conversion of a film standard system, and conversion between a video standard system and a film standard system.

[0002]

2. Description of the Related Art For example, British Patent Application No. GB-A-2 231 749
As described in the specification, in a motion-compensated television standard converter, a plurality of pairs of consecutive input images are processed to represent a plurality of pairs of input images representing motion of the images. A set of motion vectors is generated. This process is performed on separate blocks of the image. As a result, each motion vector represents motion between images of their respective block contents.

[0003]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a motion compensation video signal processing method and apparatus capable of generating a more accurate motion vector.

[0004]

In the motion vector estimation process, the correlation surface representing the spatial correlation between the blocks of two input images is examined in order to detect a plurality of maximum correlation points. (The correlation surface represents the difference between the two input images. Therefore, the maximum correlation point is actually the minimum point on the correlation surface, referred to here as such.) Is detected, a motion vector is generated from the minimum spatial position in the correlation surface. The rest of the correlation surface is tested to see if its minimum represents a significant correlation peak. If the minimum value passes this test, it is treated as "valid",
The confidence flag associated with that motion vector is set.

After the individual motion vectors, termed "local" motion vectors, have been derived for the blocks of the input image, those local motion vectors are provided to a motion vector reducer. The motion vector reducer 220 selects from among a set of vectors known as a zero motion vector, the local motion vector of the block, the local motion vector of the block adjacent to the block in the input image, and the “wide area” motion vector. A set of selected motion vectors is assigned to each block. A global motion vector is derived for each image by ranking all (valid) local motion vectors for this image in decreasing frequency. Then, from the most common of the local motion vectors, a number of unique motion vectors to be used as wide area motion vectors are selected. Here, suppression is added to ensure that the global motion vectors differ from each other by at least a minimum amount.

The sets of motion vectors for all blocks of the input image are then provided to a motion vector selector. The purpose of the motion vector selector is to assign a single motion vector selected from a set of motion vectors given for the corresponding block of the input image to each pixel of a block of the output image. It is in.
When this motion vector is used to interpolate the output pixels, the set of motion vectors provided to the motion vector selector is tested by performing a correlation test on the pixel blocks pointed to by each of the set of motion vectors. It As a result, the "best" motion vector is sorted for each pixel from the set of motion vectors of the corresponding block. The motion vector sorted for each pixel is then added to the motion compensation interpolator. The interpolator uses these motion vectors and
The pixels of the output image from the pair of input images are interpolated according to the temporal shift of the output image from the pair of input images.

Therefore, the purpose of the motion vector reducer is to ideally select the motion vector for each block in order to select the most suitable motion vector for each output pixel corresponding to each block. It is to give to the motion vector selector. The wide area motion vector represents, as a wide area, a motion that is frequently detected in the image. Therefore, it is ideal to include these vectors in the set of vectors from which the most suitable vector is selected by the motion vector selector.

It has been found that fast moving parts of a video image have less spatial detail than static or slow moving parts. This is done during vector screening (which relies on detecting the correlation between parts of the same object in two consecutive input fields), even if the motion vector that accurately represents the fast motion of that object can be estimated. Correlation tests mean unreliable.

Since the present invention is used for interpolation of an output image of an output video signal, means for generating a plurality of motion vectors representing image motion between a pair of input images of an input video signal, each motion vector being a predetermined motion vector. A motion compensation video signal processing device comprising means for detecting whether or not it is larger than a size, and means for limiting use of a motion vector detected as being larger than a predetermined size when interpolating an output image. give.

Therefore, even if large magnitude (long) motion vectors (ie, vectors representing rapid image motion) are generated, they are purposely used to interpolate the output image. Limited. this is,
This is because the lack of spatial detail in fast moving objects (eg, due to camera integral blur) implies the following:

A) The vector selection process applied to rapidly moving objects can be unreliable. b) For a rapidly moving object, the eye is subject to abrupt changes (judde) caused by the lack of motion-compensated interpolation.
r) cannot be detected. Therefore, this abrupt change is less disturbing than motion-corrected interpolation using an erroneously selected vector.

In a preferred embodiment, each motion vector comprises a plurality of coordinate axes representing image motion along the respective coordinate axis, and the detection means further comprises each of the coordinate axes. Means for comparing to a respective predetermined threshold value associated with.

In another preferred embodiment, each motion vector comprises a plurality of coordinate axes representing image motion along the respective coordinate axis, and the detecting means further comprises a coordinate value of each motion vector. Means are provided for generating a magnitude value and means for comparing the magnitude value with a predetermined threshold value. Further, the magnitude value described above represents the magnitude of the image motion represented by the motion vector.
In this case, the means for generating the magnitude value are preferably operable to detect the square root of the squared weighted addition of the coordinate values (weighted addition). Weight addition is used to reflect differently scaled motion vector coordinate values in various coordinate directions. Alternatively, each squared coordinate value may be equally weighted.

Coordinate axes exist along various directions with respect to the input image, but preferably the coordinate axes represent vertical and horizontal image movement.

Further, each motion vector represents the image motion of each block of one input image of the pair between one input image of the pair and the other image of the pair, and Means for deriving a set of global motion vectors comprising a plurality of unique motion vectors selected from the most common of the plurality of motion vectors; It is preferable to have means for assigning to each of the blocks a group of motion vectors selected from the motion vector representing the motion and the global motion vector.

[0016] In order to prevent the selection of a global vector that only represents the motion of a small part of an image, each motion vector selected as a global vector may occur at least a predetermined number of times among a plurality of motion vectors. preferable.

In one embodiment for preventing "long" motion vectors from being used as wide area vectors, the limiting means selects motion vectors detected as being larger than a predetermined size as wide area motion vectors. It is preferable to provide a means for preventing this.

In another embodiment, in response to detecting that the motion vector representing the motion of the block is greater than a predetermined magnitude, the assigning means is controlled to represent the zero motion vector and the motion of the block. Limiting means adapted to assign only those motion vectors that are allocated to the block, imposes further restrictions on the use of "long" motion vectors.

In a further preferred embodiment, during interpolation of the output image, a "long" motion is provided by using a limiting means comprising means for preventing the use of motion vectors detected as being larger than a predetermined magnitude. It makes sure that the vector is never used.

Whether each motion vector is greater than a second predetermined magnitude that is greater than the first said predetermined magnitude so that different measurements are made on the motion vector in two different magnitude ranges. It is preferable to provide a means for detecting the motion vector and a means for preventing the use of the motion vector detected as being larger than the second predetermined size when the output image is interpolated.

In another preferred embodiment, the limiting means comprises means associated with each motion vector for setting a flag to indicate whether the motion vector is greater than a predetermined magnitude, the video signal. The processing device further includes means for selecting one motion vector from the plurality of motion vectors for use in interpolating each pixel of the output image, a motion compensated pixel interpolator operable according to at least two operation modes, and A means for selecting one of the operation modes of the motion compensated pixel interpolator for interpolating the pixels of the output image, depending on whether the flag corresponding to the motion vector selected for use in interpolating the pixel is set or not. It is preferably provided.

Pixel interpolators of various configurations are configured to perform different interpolation operations depending on whether or not the flag is set. However, the pixel interpolator does not change temporally of the output image with respect to the pair of input images. According to one of the operating modes to interpolate between respective blocks of the two images dictated by the selected motion vector according to the position-dependent coupling ratio, and according to an equal coupling ratio, It is preferably operable according to another mode of operation in order to interpolate between respective blocks of the two images indicated by the zero motion vector.

As mentioned above, a problem arises when the wide area vector is selected from the vectors of relatively low occurrence frequency. For example, if the majority of the image is stationary, except for small moving objects, then the majority of the motion vectors in that image will be close to zero. The vectors representing the static part of the image differ slightly due to noise or spurious signals, but are not too close to each other to be selected as global vectors. Therefore, the next most commonly occurring vector selected as the wide area vector is the vector derived from that small moving object. If this vector passes vector selection as one of the four vectors associated with static blocks in the image, it is incorrectly selected during vector selection. The use of erroneous vectors in this manner results in the generation of visible miscellaneous components.

From a second perspective, the present invention provides a motion compensated video signal processing apparatus in which a motion vector is generated to represent image motion between a pair of input images of an input video signal. The processor produces a plurality of local motion vectors for representing the image motion of each block of one input image of the pair between the one input image of the pair and the other image of the pair. Means for generating, a means for deriving from the local motion vector a set of global motion vectors comprising a plurality of unique motion vectors selected from the most common local motion vector, and a zero motion vector, local for the block Means for allocating a group of motion vectors selected from the motion vector and the wide area motion vector to each of the blocks are provided, and each wide area vector is assumed to occur at least a predetermined number of times among a plurality of local motion vectors.

This aspect of the invention uses a wide area motion vector derived from a relatively small number of occurrences by preventing selection of local vectors that occur less than a predetermined number of times as wide area motion vectors. To solve.

Preferably, the derivation means is provided with means for preventing selection of two or more local motion vectors that differ from each other by an amount smaller than a predetermined amount as the wide area vector.

In the preferred embodiment, the predetermined number of occurrences can be adjusted under operator control.

In a simple and advantageous embodiment, the processor is a means for storing a frequency sequence of a plurality of addressable memory locations, a memory location corresponding to each of the plurality of motion vectors. Means are provided for incrementing the frequency value and means for detecting those memory locations having the highest stored frequency value.

The video signal processing device according to the present invention is particularly effective for a television standard conversion device.

Viewed from a third aspect, the present invention provides a motion compensated video signal processing method. Since this method is used to interpolate an output image of an output video signal, a step of generating a plurality of motion vectors for representing an image motion between a pair of input images of an input video signal, each motion vector being a predetermined Detecting if greater than magnitude and limiting the use of motion vectors detected as greater than a predetermined magnitude in interpolating the output image.

Viewed from a fourth aspect, the present invention provides a motion compensated video signal processing method wherein motion vectors are generated to represent image motion between a pair of input images of the input video signal. The video signal processing method comprises a plurality of local motions between one input image of a pair and the other image of the pair to represent image motion of each block of the one input image of the pair. Stage of generating vector,
Deriving from the local motion vector a set of global motion vectors comprising a plurality of unique motion vectors selected from the most common local motion vector, and the zero motion vector, the local motion vector of the block and the global motion vector. There is the step of assigning a group of motion vectors selected from the motion vectors to each of the blocks. Further, each wide area vector is generated at least a predetermined number of times among the plurality of motion vectors.

[0032]

DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 shows a schematic block diagram of a motion compensated television standard converter. This device uses an input interlaced digital video signal 50 (eg 1125/60).
2: 1 high definition video signal (HDVS) is received and output interlaced digital video signal 60 (eg, 1150/5)
0 2: 1 signal) is generated.

The digital video signal 50 is first provided to the input buffer / packer 110. For a conventional definition input signal, the input buffer / packer 110 formats the image data into a high definition (16: 9 aspect ratio) format, filling it with black pixels where necessary. Input buffer / packer 1 for HDVS input
10 merely provides a data buffer function.

Data is supplied from the input buffer / packer 110 to the matrix circuit 120, and the matrix circuit 12
At 0, the colorimetry of the input video signal (if necessary) is
Converted to standard “CCIR recommendation 601” (Y, Cr, Cb) colorimetry.

The input video signal from the matrix circuit 120 is applied to the time base conversion and delay circuit 130. The signal is also provided to a sub (down) sampled time base conversion and delay circuit 180 via a subsampler 170. The time base conversion and delay circuit determines the temporal position of each field of the output video signal and is used to interpolate the output field, so that the two of the input video signals that are temporally closest to the output field are used. Select a field. For each field of the output video signal, the two input fields selected by the time base conversion circuit are appropriately delayed before being provided to the interpolator 140 for interpolating the output field. A control signal t indicating the temporal position of each output field with respect to the two selected input fields.
From the time base conversion and delay circuit 130 to the interpolator 14
Given to 0.

The time base transform and delay circuit 180 operates similarly, except that it uses the spatially subsampled video signal provided by the subsampler 170. The pair of fields corresponding to the pair selected by the time base transform and delay circuit 180 are selected by the subsampled time base transform and delay circuit 180 from the subsampled video signal and used to generate the motion vector. The time base conversion and delay circuits 130 and 180 operate according to sync signals associated with the input video signal, the output video signal, or both. When only one sync signal is provided, the timing of the other field of the two sync signals is deterministically generated in the time base conversion and delay circuits 130, 180. The pairs of fields of the subsampled input video signal selected by the time base conversion and delay circuit 180 are processed by the block matcher 190, the correlation plane processor 20.
0, a vector estimator 210, a vector reducer 220, a vector sorter 230, and a vector post-processor 240. The pairs of input fields are first correlated to represent the spatial correlation between the temporally earlier search block of the two selected input fields and the temporally later (larger) search area of the two input fields. It is provided to the block matcher 190 where the surface is calculated.

From the correlation surfaces output by the block matcher 190, the correlation surface processor 200 produces a larger number of interpolated correlation surfaces. The correlation surfaces are then sent to the vector estimator 210. Vector estimator 210
Detects the maximum correlation point in the interpolated correlation plane. (The original correlation surface actually represents the difference between the blocks of the two input fields. This means that the maximum correlation point is actually the minimum value on the correlation surface, and henceforth " Minimum value "). To find the minimum, the additional points on the correlation planes are interpolated, and some compensation is provided for the loss of resolution caused by using the subsampled video signal to generate those planes. give. From the detected minimum value on each correlation surface, the vector estimator 210 generates a motion vector. The motion vector is provided to the vector reducer 220.

The vector estimator also performs a confidence test on each generated motion vector to see if the motion vector is significant at the average data level, and for each motion vector a confidence measure indicating the test result. Attach the flag. The confidence test, known as the "threshold" test, is described in GB-A-2 231 740 and in British Patent Application No. 9,307,442.5 (along with some other features of the apparatus of Figure 1).

The test to detect if each vector is different is also performed by the vector estimator 210. This test examines the correlation surface (away from the exclusion zone near the minimum detected),
Detects the next lowest minimum. If this second minimum is not at the edge of the exclusion zone, the motion vector derived from the original minimum is flagged as potentially spurious.

Vector reducer 220 operates to reduce the choice of possible motion vectors for each pixel in the output field before the motion vectors are provided to the vector selector. The output field is conceptually divided into a plurality of pixel blocks. Each block is
In the output field has a position corresponding to the position of the earliest search block in the selected input field. The motion vector reducer 220 compiles a group of four motion vectors to be associated with each block of the output field. Each pixel in the block is interpolated with the selected one of the four motion vectors of the group.

Vectors flagged as "fake signals" are the same during vector reduction if they are identical to unflagged vectors in nearby blocks or are similar within selectable thresholds. Re-estimated.

As part of its function, the vector reducer 2
20 is "good" without considering the position of the input fields used to obtain these motion vectors.
Count the frequency of occurrence of motion vectors (i.e., motion vectors that have passed the confidence and false signal tests, or are re-estimated as not false signals). The good motion vectors are then ranked in order of decreasing frequency. The most common of good motion vectors that are very different from each other are classified as "wide area" motion vectors. Then passed the trust test 3
One motion vector is selected for each block of output pixels and, along with the zero motion vector, is provided to a vector selector for further processing. These three motion vectors are selected from the following in a predetermined priority order.

(I) Motion vector generated from the corresponding search block ("local" motion vector) (ii) Motion vector generated from surrounding search blocks ("neighboring" motion vector) (iii) Wide area motion vector.

The vector sorter is also selected by the time base conversion and delay circuit 180 and the control signal t.
, Used to calculate the motion vector
Takes one input field as input. These fields are delayed appropriately so that they are presented to the vector sorter at the same time as the vectors derived from them. The vector sorter provides an output composed of one motion vector per pixel in the output field. This motion vector is selected from the four motion vectors provided by the vector reducer 220 for that block. The vector selection process involves detecting the degree of correlation between the test blocks of the two input fields indicated by the motion vector under test. The motion vector with the highest degree of correlation between test blocks is selected for use in interpolating the output pixels. The "motion flag" is also generated by the vector sorter. If the degree of correlation between the blocks indicated by the zero motion vector is greater than the preset threshold, then this flag is set to "static" (no motion).

The vector post-processor 240 reformats the motion vectors selected by the vector selector to reflect the vertical or horizontal scaling of the image and provides the reformatted vectors to the interpolator 140. Using those motion vectors, the interpolator 140 uses the time base conversion and delay circuit 130.
The output field from the corresponding two (non-subsampled) interlaced input fields selected by is interpolated and considers any image motion currently represented by the motion vector supplied to interpolator 140.

If the motion flag indicates that the current output pixel is in the moving part of the image, the pixels from the two selected fields provided to the interpolator are (indicated by the control signal t). (As such), they are combined in relative proportions by the temporal position of the output fields with respect to the two input fields. Therefore, closer input fields are used in a larger proportion. If the motion flag is set to "static", the temporal weighting is fixed at 50% of each input field. The output of interpolator 140 is sent to output buffer 150 for output as a high definition output signal and, using motion flags, to down converter 160 which produces a conventional definition output signal 165.

Downconverter 160 uses a conventional definition device to monitor, emit, and / or record the device's output display (which may be, for example, a high definition video signal). This has several advantages as conventional definition recording devices are very cheap and are much more widespread than high definition devices. For example, simultaneous output of conventional and high definition video signals may be required for each transmission over terrestrial and satellite channels. Further, if the output video signal is recorded directly on film, for example using an electron beam recorder, a down converter causes simultaneous recording on video tape.

The sub-sampler 170 receives the input field to the time base conversion and delay circuit 180 before
Performs horizontal and vertical spatial subsampling of these input video fields received from matrix circuit 120. Horizontal subsampling means that the input field is first prefiltered by a half-bandwidth lowpass filter (in this example, which is 2: 1 horizontal decimation) so that every other video sample along each video line is It is simple in that it is discarded, thereby reducing the number of samples along each video line by half.

Vertical subsampling of the input field is complicated because the input video signals are interlaced. This means that successive lines of video samples in each interlaced field are effectively separated into two video lines, and only one video line of a complete frame from the line in the field preceding or following the line in each field. It means that it is displaced in the vertical direction.

Next, the configuration and operation of the vector reducer 220 will be described in more detail.

A block diagram of the vector reducer 220 is shown in FIG. The motion vectors and their associated flags derived by the vector estimator 210 are passed through the input 300 of the vector reducer 220 to the ridge (ri
dge) is supplied to the minimum value re-estimator 310. Ridge minimum re-estimator 310 detects if another motion vector along the ridge direction has a coordinate coordinate of 90 degrees to the ridge direction and, if so, assigns that coordinate to the ridge vector. And “correcting” the motion vector derived from the elongated (“ridge”) minimum by resetting the “ridge flag” associated with that vector. Then, the ridge minimum value re-estimator 310
Provides those vectors and associated flags to the false signal vector verifier 320 and global vector detector 330. The false signal vector verifier 320 may legitimately re-identify vectors that were potentially indicated as false signals, and those vectors and associated flags may be assigned to the global vector limiter 340. Perform a test to see if it forms an input.
Global vector detector 330 derives a set of global motion vectors from the vectors provided thereto. The global vector output by global vector detector 330 forms the second input to global vector limiter 340.

The wide-area vector limiter 340 indicates, for each wide-area vector given thereto, the relation between the wide-area vector and the various blocks of the input image to which the local vector given from the false signal vector checker 320 corresponds. Generate a global mask sequence. The wide area mask sequence generated by the wide area vector limiter 340 is a false signal vector verifier 3.
The local vector and associated flags (properly delayed) from 20 and the global vector from global vector detector 330 are provided at 350. Three
50 assigns to each block of the input image to which the local vector corresponds a plurality of unique vectors supplied to the vector selector via the output block vectors 360 of the vector reducer 220.

The ridge minimum re-estimator 310, the false signal vector verifier 320, the global vector detector 330, and the global vector limiters 340 and 350 are under the control of a microprocessor 370 (shown schematically in dashed lines in the drawing). Works with. The microprocessor 370 is also connected to the global vector detector 330 so that various parameters can be adjusted for any manual intervention in the global vector detection process. The adjusted parameters are
Input by operator, microprocessor 37
0 to the wide area vector detector 330. A bidirectional communication link 380 is provided to connect the microprocessor 370 to yet another similar microprocessor (not shown). This is because the processor wide area (generally shown as 185 in FIG. 1) is actually duplicated, ie two motion processors perform the processing for every other output field. is there. Therefore, the vector reducer 220 shown in FIG. 2 is duplicated and a bidirectional communication link 380 is provided for communicating the two microprocessors that control the vector reduce operation.

Next, the configuration and operation of the components of the vector reducer 220 will be described in detail.

The vectors and flags output by the ridge minimum re-estimator 310 are provided to the global vector detector 330. The purpose of the global vector detector is to derive from the vectors supplied to it a set of global motion vectors consisting of a plurality of unique motion vectors selected from the most common local motion vectors. Generally speaking, the global vector detector 330 ranks the local motion vectors corresponding to a given input image in order of decreasing frequency and selects (in this example) the eight highest common vectors as global vectors. . However, it is advantageous to add various constraints to the process for selecting a global vector from a set of local vectors. In particular,
Even if large (long) motion vectors (ie, vectors that represent rapid image motion) are generated, it is desirable to prevent such long vectors from contributing to the wide area vector. This is because the lack of spatial detail in fast moving objects (eg, due to camera integral blur) means that the operation of the vector classifier (FIG. 1) for fast moving objects cannot be reliable. It has been observed that the rapidly moving parts of the video image have less spatial detail than the stationary or slowly moving parts. This means that even if the motion vector that accurately represents the rapid motion of the object is estimated,
It means that the correlation test performed in the vector classifier, which relies on the correlation detection between parts of the same object in two consecutive input fields, cannot be reliable. It is therefore necessary to prevent such long vectors from being selected as wide area vectors so that large wide area vectors are not incorrectly selected for the selection of image areas that actually have very little motion. Is preferred. This can have a subjective disturbing effect on the output image.

Another constraint added to the global vector detection process is that it prevents a vector being accepted as a global vector if it does not occur at least a predetermined number of times on the input image as global. If the local motion vectors are not sufficiently common, and thus represent the motion of only a very small part of the image, it is possible that this vector may be assigned to other parts of the image during the global vector detection phase. It is preferable to remove.

A schematic block diagram of a global vector detector 330 in which the above suppression is implemented is shown in FIG. The input vector and associated flags are provided to the amplitude detector 400. In the amplitude detector 400, the amplitude of each input vector, that is, the magnitude of the magnitude of the threshold 410 is determined.
To determine if the vector qualifies as a "long" vector. The magnitude threshold 410 comprises thresholds for both horizontal and vertical components of the motion vector. For example, the threshold is set to half the maximum possible range of motion vectors in the horizontal and vertical directions. In this example, the maximum horizontal range of the motion vector is plus or minus 64 pixels, and the maximum vertical range is plus or minus 32 pixels. Therefore, the size threshold 410 is set to plus or minus 32 pixels in the horizontal direction and plus or minus 16 pixels in the vertical direction. If the magnitude of the input vector exceeds the threshold 410, the amplitude detector 400 sets the "long vector" flag (LV).

As an alternative, the threshold 410 is the vector magnitude √ (x ² + y ² ), that is, the square of x and y.
Represents the square root of the sum of the squares of, i.e., the maximum value of the weighted additions to reflect different scaling of horizontal and vertical coordinates (eg, √ (4x ² + y ² ) or √ (x ² + 4y ² )).

The input vector provided to the amplitude detector 400 is also provided to the frequency array 420. Each input of the frequency array 420 is uniquely addressed by the possible values of the local motion vector. In this way,
The local motion vector generated from the current pair of input fields addresses that column and increments the column input corresponding to the value of the motion vector. However, the long vector flag LV derived from the amplitude detector 400 is also provided to the frequency array 420. Frequency array 4
When the long vector flag LV for the motion vector given to 20 is set, this prevents the associated motion vector from incrementing the frequency sequence.

After all the vectors (apart from the long ones) have been counted by the frequent array 420 in this manner, the array scanner 430 scans the frequent array 420 and outputs the eight column inputs showing the highest counts. Identify. In contrast, the motion vectors differ by at least a different predetermined threshold 435. The addresses of these column inputs represent the values of the eight most common local vectors (apart from the long vectors) for the corresponding pair of input images. The count value held by the frequency sequence for each of these eight vectors (ie the frequency of occurrence of the vector) is the corresponding vector
It is provided by the array scanner 430 to the comparator 440 at the same time as it is provided at the 50 inputs. Comparator 4
Reference numeral 40 compares the count or occurrence frequency corresponding to each vector with a predetermined threshold number, that is, a “wide area threshold”. It is provided at the output 460 of the switch 450.
(Appropriate values for the global threshold vary depending on the particular image material being processed. However, as an example:
The global threshold is set to 0.5% of the total number of local motion vectors derived from the input image. ) If the count corresponding to any of the eight largest common vectors does not exceed the global threshold, that is, the vector does not occur frequently enough to qualify as a global vector, then comparator 440 determines that the corresponding vector Generates an output for controlling switch 450 so that is not sent to its output 460 as a global vector. And
That vector is replaced by the zero motion vector. However, assuming all eight highest common vectors pass the wide area threshold test, a set of eight unique wide area motion vectors is output 460 of switch 450.
Will be given to.

It should be appreciated that various other constraints can be added to the global vector detector 330. In particular,
Any input local motion vector for which the corresponding ridge or false signal flag is set, or the confidence flag indicates that the vector has failed the confidence test, is handled in the same way as for long vectors, i.e., such a vector. Preventing frequent array 420 increments prevents contribution to the global vector.

As shown in FIG. 3, the long vector flag LV derived for each of the input local motion vectors is provided to the block vector allocator 350 (FIG. 2) by the global vector detector 330 after an appropriate delay. And is used for the next vector reduction process of the local vector.

As shown in FIG. 2, global vector detector 330 is communicatively coupled to microprocessor 370. This allows the various thresholds used during global vector detection to be modified, if necessary, by manual intervention of the operator. The operator can enter new parameters, for example, via a system controller (not shown). Those parameters are then provided to the block vector allocator 350 via the microprocessor 370.

The wide area vector output by the wide area vector detector 330 is provided to one of the inputs of the wide area vector limiter 340. At the other input of 340, the processed local motion vector and its associated flag are provided from the false signal vector verifier 320.

As shown in FIG. 3, the long vector flag LV derived for each of the input local motion vectors is provided to the block vector allocator 350 (FIG. 2) by the global vector detector 330 after an appropriate delay. And is used for the next vector reduction process of the local vector.

As shown in FIG. 2, global vector detector 330 is communicatively coupled to microprocessor 370. This allows the various thresholds used during global vector detection to be modified, if necessary, by manual intervention of the operator. The operator can enter new parameters, for example, via a system controller (not shown). Those parameters are then provided to the block vector allocator 350 via the microprocessor 370.

The wide area vector output by the wide area vector detector 330 is provided to one of the inputs of the wide area vector limiter 340. At the other input of 340, the processed local motion vector and its associated flag are provided from the false signal vector verifier 320.

The purpose of the global vector limiter 340 is to make each of the global vectors a block of the input image in or near the area where the same or similar motion as represented by the global motion vector was initially derived. , To relate. This association prevents inappropriate global vectors from being assigned to blocks, that is, one area of the input image is assigned a global vector that represents a significantly different motion of the block in the remote part of the image, Prevent blocking.

The block vector assigner is the amplitude detector 4
It receives the LV flag from 00 and responds in many ways. In the above-described embodiment, the long vector is not already used as the global vector, but the block vector allocator 350 actually uses the flag to prevent the long vector from being used completely. Such a device is represented schematically in FIG.
In that device, the LV flag associated with each motion vector controls the multiplexer 600 to pass either the associated vector or the zero vector to the block vector allocator 35.
Send to 0 input.

In another apparatus, shown schematically in FIG. 5, the vector selection process for a particular block is
It limits the possibility of using long local vectors only for that block or zero motion vector. Figure 5
The LV flag for each block controls the multiplexer 610 to either the three motion vectors assigned to that block during vector reduction, or the sum of the local motion vector and two zero motion vectors, as shown in FIG. Send (to vector sorter).

In another embodiment, shown in FIG. 6, another amplitude detector 620, similar to amplitude detector 400, for comparing each vector to a second, or larger, threshold value 630. Has been adopted. The use of long vectors is generally allowed (although limited as described above), but the use of vectors larger than this second threshold is not allowed. this is,
Only if each vector does not exceed the second threshold value 630, the vector is assigned to the block vector assigner 350.
Done by using a multiplexer 640 to send to.

In yet another embodiment, the LV flag is used to control the operation of interpolator 140.
The device is shown in FIG. In this device, LV
A flag controls 650 to provide the control signal t or a constant value of 0.5 to the interpolator 140. This indicates that the use of temporal weighting (described above) is prohibited once the LV flag is set. This may reduce random image formation, such as visible fluctuations caused by the beat effect observed for moving parts of the "shuttered" image.

[0073]

According to the present invention, a motion compensation video signal processing method and apparatus capable of generating a more accurate motion vector can be obtained.

[Brief description of drawings]

FIG. 1 is a schematic block diagram of a motion compensation television standard conversion apparatus using the present invention.

FIG. 2 is a schematic block diagram showing an example of the motion vector reducer of FIG.

FIG. 3 is a schematic block diagram of the wide area vector detector of FIG.

FIG. 4 is a first variation on the apparatus of FIG. 1 for controlling the use of motion vectors in response to long vector flags.
FIG.

5 is a second variation on the apparatus of FIG. 1 for controlling the use of motion vectors in response to long vector flags.
FIG.

FIG. 6 is a third variation on the apparatus of FIG. 1 for controlling the use of motion vectors in response to long vector flags.
FIG.

7 is a fourth variation on the apparatus of FIG. 1 for controlling the use of motion vectors in response to long vector flags.
FIG.

[Explanation of symbols]

 410 435 Threshold 400 Amplitude detector 420 Frequency array 430 Array scanner 440 Comparator 450 Motion vector usage limiter (switch)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Martin Rex Dricot United Kingdom Hampshire, Basingstoke, Basing, Lingfield Claus 6 (72) Inventor Shima Lab Giversani United Kingdom Hampshire, Basingstoke, Hook, Foracre Copis 5

Claims (20)

[Claims] 1. Means for generating a plurality of motion vectors representing the motion of an image between a pair of input images of an input video signal for use in interpolating an output image of an output video signal, each motion vector being a predetermined A motion-compensated video signal processing device including means for detecting whether or not it is larger than a size, and means for limiting use of a motion vector detected as being larger than a predetermined size when interpolating an output image. 2. The apparatus according to claim 1, wherein each of the motion vectors comprises a plurality of coordinate axes representing an image motion along each coordinate axis, and the detecting means further comprises each of the coordinate axes. , A motion-compensated video signal processing device comprising means for comparing with a respective predetermined threshold value associated with each of the coordinate axes. 3. The apparatus according to claim 1, wherein each of the motion vectors comprises a plurality of coordinate axes representing image motion along each coordinate axis, and the detecting means further comprises: Means for generating a magnitude value from the coordinate values and means for comparing the magnitude value with a predetermined threshold value, the magnitude value being the magnitude of the image motion represented by the motion vector. A motion-compensated video signal processing device characterized in that 4. Apparatus according to claim 3, wherein said means for generating a magnitude value is operable to detect the square root of the weighted sum of squared coordinate values. Video signal processor. 5. A motion-compensated video signal processing device according to claim 2, wherein the coordinate axes represent vertical and horizontal image movements. 6. The apparatus according to claim 1, wherein each of the motion vectors is between a pair of one input image and the other image of the pair. A set of global motion vectors representing the image motion of each block of one of the input images, the device further comprising a plurality of unique motion vectors selected from the most common of the plurality of motion vectors. A means for deriving a motion vector, and means for allocating to each of the blocks a group of motion vectors selected from a zero motion vector, a motion vector representing the image motion of the block and a wide area motion vector, Motion compensation video signal processing device. 7. The motion compensation video signal processing device according to claim 6, wherein each motion vector selected as a wide area vector occurs at least a predetermined number of times among a plurality of motion vectors. . 8. The apparatus according to claim 6 or 7, further comprising means for preventing the limiting means from selecting a motion vector detected as being larger than a predetermined size as a wide area motion vector. A motion compensation video signal processing device characterized in that 9. The apparatus according to claim 6, wherein the limiting means is responsive to the detection that the motion vector representing the motion of the block is larger than a predetermined magnitude. A motion compensation video signal processing device, characterized in that it controls an allocating means to allocate only a zero motion vector and a motion vector representing the motion of a block to the block. 10. The apparatus according to claim 1, wherein the limiting unit prevents use of a motion vector detected as being larger than a predetermined size when interpolating an output image. A motion compensation video signal processing device comprising: 11. The apparatus according to claim 1, wherein it is detected whether each of the motion vectors is larger than a second predetermined magnitude that is larger than the first predetermined magnitude. Means and when interpolating the output image,
A motion compensated video signal processing device comprising means for preventing the use of a motion vector detected as being larger than a second predetermined magnitude. 12. A device according to claim 1, wherein the limiting means is associated with each motion vector to indicate whether the motion vector is greater than a predetermined magnitude. Equipped with means to set the flag,
The signal processing device further comprises means for selecting one motion vector from a plurality of motion vectors for use in interpolating each pixel of the output image; motion compensated pixel interpolator operable according to at least two operation modes; And whether the flag corresponding to the motion vector selected for use in interpolating that pixel has been set, one of the operating modes of the motion compensated pixel interpolator for interpolating the pixel in the output image.
A motion compensation video signal processing device, characterized in that it comprises means for selecting one. 13. The apparatus according to claim 12, wherein the pixel interpolator is dictated by a selected motion vector according to a combination ratio depending on a temporal position of an output image with respect to a pair of input images. A pixel interpolator operable according to one of the operation modes for interpolating between respective blocks of the two images;
Motion compensated video signal processing device, characterized in that it is operable according to another mode of operation, in order to interpolate between respective blocks of two images indicated by zero motion vectors according to an equal coupling ratio. 14. The motion vector is 1 of the input video signal.
In a motion-compensated video signal processing device, such as is generated to represent image motion between a pair of input images, the processing device comprising: between one input image of a pair and the other image of the pair, A means for generating a plurality of local motion vectors for representing the image motion of each block of one of the input images of the pair, comprising a plurality of unique motion vectors selected from the most common local motion vectors Means for deriving a set of global motion vectors from the local motion vectors, and means for allocating to each of the blocks a zero motion vector, a group of motion vectors selected from the local and global motion vectors of the block, Motion compensated video signal characterized in that each wide area vector occurs at least a predetermined number of times among a plurality of motion vectors Management apparatus. 15. The apparatus according to claim 14, wherein the derivation means includes means for preventing selection of two or more local motion vectors different from each other by a smaller amount than a predetermined amount as the wide area vector. A motion compensation video signal processing device characterized by: 16. A motion compensated video signal processing device as claimed in claim 14 or 15, characterized in that the predetermined number of occurrences can be adjusted under the control of an operator. 17. A device according to claim 14, further comprising means for storing a frequency sequence comprising a plurality of addressable memory locations; storing at a memory location corresponding to each of a plurality of motion vectors. Motion compensated video signal processing device, comprising means for incrementing the stored frequency value and means for detecting those memory locations having the largest stored frequency value. 18. A television standard conversion device comprising the device according to claim 1. 19. A step of generating a plurality of motion vectors for representing image motion between a pair of input images of an input video signal for use in interpolation of an output image of an output video signal, each motion vector being predetermined. A motion-compensated video signal processing method, comprising: detecting whether the motion vector is larger than a predetermined size, and limiting the use of a motion vector detected as larger than a predetermined size when interpolating an output image. 20. The motion vector is 1 of the input video signal.
In a motion-compensated video signal processing method as generated to represent image motion between a pair of input images, the processing method comprising: between one input image of a pair and the other image of the pair, Generating a plurality of local motion vectors to represent the image motion of each block of one of the input images of the pair, with a plurality of unique motion vectors selected from the most common local motion vectors Deriving a set of global motion vectors from the local motion vectors, and assigning to each of the blocks a zero motion vector, a group of motion vectors selected from the local and global motion vectors of the block , A motion-compensated video signal characterized in that each wide-area vector is generated at least a predetermined number of times among a plurality of motion vectors Management method.