Field of the Invention

The present invention relates to a motion estimation apparatus for use in an image signal encoding system; and, more particularly, to a motion estimation apparatus for detecting a motion vector employing a weight function.

Description of the Prior Art

When an image signal comprising a sequence of image "frames" is expressed in a digital form, substantial amounts of data are generated for transmission, especially in the case of a high definition television system. Since, however, the available frequency bandwidth of a conventional transmission channel is limited, in order to transmit the substantial amounts of digital data through the limited channel bandwidth, it is inevitable to compress or reduce the volume of the transmission data. Among various video compression techniques, a motion compensated interframe coding technique, which utilizes temporal redundancies of the video signals between two adjacent video frames for the compression of the signals, is known to be one of the effective compression techniques.

In the motion compensated interframe coding scheme, current frame data is predicted from previous frame data based on an estimation of the motion between the current and the previous frames. Such estimated motion may be described in terms of two dimensional motion vectors representing the displacement of pixels between the previous and the current frames.

One of the motion vector estimation schemes which have been proposed in the art is the block matching algorithm. According to the block matching algorithm, a current frame is divided into a plurality of equal-sized search blocks. The size of a search block typically ranges between 8x8 and 32x32 pixels. To determine a motion vector for a search block in the current frame, a similarity calculation is performed between the search block of the current frame and each of a multiplicity of equal-sized candidate blocks included in a generally larger search region within a previous frame. An error function employing the MSE (mean square error) is used to carry out the similarity measurement between the search block of the current frame and each of the candidate blocks in the search region. The error function may be expressed as follows.

wherein H x V represents the size of a search block; I(i,j) represents a luminance level of a pixel at a coordinate (i,j) in the search block; and P(i,j) represents a luminance level of a pixel at the coordinate (i,j) in a candidate block.

And a motion vector, by definition, represents the displacement between the search block and a candidate block which yields a minimum error function. The motion vector is then used in a receiver to construct a picture from a previous frame on a block-by-block basis.

Such a motion compensated interframe coding scheme, however, does not take into account the continuity between adjacent blocks due to the use of the error function representing the correlation between blocks. Consequently, the block boundaries may become visible because the boundaries are straight lines and highly structured and the discontinuities are particularly noticeable to the human observer. Such blocking effect occurring at the boundaries between adjacent blocks in a motion compensation process and the block structure artifacts caused by the blocking effect tends to deteriorate the quality of the coded image.

Summary of the Invention

It is, therefore, an object of the invention to provide an improved motion estimation apparatus capable of reducing the blocking effect.

It is another object of the invention to provide an improved motion estimation apparatus for detecting a motion vector maintaining the continuity of boundary areas between adjacent blocks.

In accordance with the invention, there is provided an apparatus which estimates a displacement of a search block in a current frame with respect to each of candidate blocks in a previous frame to produce motion vectors and error functions representing the similarity between the search block and a candidate block. The motion estimating apparatus employs a weight function to weight a number of error functions based on a MSE measurement and selects a weighted error function entailing a minimum error to produce a motion vector corresponding thereto.

Brief Description of the Drawings

The above and other objects and features of the present invention will become apparent from the following description of preferred embodiments given in conjunction with the accompanying drawings, in which:

• FIG. 1 shows a schematic block diagram of a block matching motion estimation apparatus employing a weight function in accordance with the invention;
• FIG. 2 illustrates an exemplary block matching process between a search block having a localized subblock therein and a large search region of the previous frame; and
• FIG. 3 represents a filter window of a 2-dimensional gradient filter to produce another embodiment of the weight function.

Detailed Description of the Preferred Embodiments

Referring to FIG. 1, there is shown a preferred embodiment of a block matching motion estimation apparatus incorporated in a motion estimation and compensation system which is used to achieve a significant data compression by removing the redundancies between successive frames, i.e., a current frame and its adjacent or previous frame. That is to say, there may be differences between the current frame and the previous frame, induced by a displacement or motion of an object; however, such differences may be confined to a relatively small region within a frame. Therefore, it is not necessary to transmit the entire image data of a current frame to a receiver (not shown). Instead, it may suffice to transmit the displacement information, i.e., motion vectors. The receiver then reconstructs the current frame from its previous frame whose image data is stored in a frame memory within the receiver, utilizing the motion vectors.

As shown, the current frame signal is provided through a line 12 to a current frame formation section 10. The current frame formation section 10 serves to divide the current frame into a plurality of search blocks of an identical size, each comprising H x V pixels. For the purpose of illustration, it is assumed that H and V are both an equal number of 16 for each search block of pixels in the current frame. Each of the search blocks in sequence is applied to a weight function generator 18 for generating a weight function. The weight function generator 18 defines a localized subblock within the search block, wherein the search block, as shown in FIG. 2, generally depicted as a reference numeral 74, has edge pixels extending along with the edges thereof and internal pixels, excepting the edge pixels, out of the edges, forming the localized subblock 74. And then, a local variance for the localized subblock 74 is derived as the weight function.

The local variance is derived as follows:

wherein var(i,j) represents the local variance of a pixel at a coordinate (i,j) in a localized subblock; SH x SV is the size of a localized subblock; I(i,j) is a luminance level at the pixel coordinate (i,j) in the localized subblock; and "mean" represents the mean luminance level for internal pixels within the localized subblock.

The mean luminance level is defined as follows:

The weight function is then provided to a number of block matching sections, only three 41, 42 and 49 of which are exemplarily shown therein.

Meanwhile, the previous frame stored in a memory (not shown) is fed through a line 13 to a search area formation section 15. The search area formation section 15 defines a generally large search region of the previous frame with an equal size, shape and search pattern, whereby the search or comparison will be carried out.

After the search region is determined at the search area formation section 15, the search region data is also applied to a number of candidate block formation sections, only three 21, 22 and 29 of which are illustratively shown therein. At each candidate block formation section, a candidate block of pixels is generated from the search region as the search block sweeps through the search region starting at the upper left-most position moving horizontally one pixel position at a time, and then vertically down through the search region moving one scan line at a time, until finally reaching the lower right-most position within the search region. All the possible candidate blocks with the size of H x V pixels are formed within the determined search region. And then, the relative displacement between each candidate block and the search block of the current frame is derived and then provided to a multiplexer 60 through lines 31 to 39 as a motion vector of that candidate block. The pixel data of each candidate block is also provided from each of the candidate block formation sections 21 to 29 to each of the block matching sections 41 to 49. At each of the block matching sections 41 to 49, an error function employing a MSE measurement is calculated, using the weight function, between the search block from the current frame block formation section 20 and the candidate block from each of the candidate block formation sections 21 to 29. Conventionally, comparison of luminance level or light intensity is performed between corresponding pixels in the search block and the candidate block to yield the error function for that candidate block. The error function indicates the degree of similarity between the search block and the selected candidate block.

In accordance with the present invention, at each of the block matching sections 41 to 49, the weight function derived from the weight function generator 18 is weighted to the error function in order to detect a motion vector which entails a smooth continuity of adjacent blocks to be reconstructed.

The weighted error function "W-MSE" is defined as follows:

wherein H x V is the size of a search block; I(i,j) is a luminance level at a pixel coordinate (i,j) in the search block; and P(i,j) is a luminance level at the a pixel coordinate (i,j) in a candidate block.

All the weighted error functions from the block matching sections 41 to 49 are supplied to a minimum error detector 50. The minimum error detector 50 compares the weighted error functions to select a weighted error function which has a smallest error.

The minimum error detector 50 outputs a selection signal which indicates the block corresponding to the minimum error function to the motion vector selector 60. The motion vector selector 60, which is of a conventional multiplexer, in response to the selection signal, chooses the displacement vector of the candidate block, which corresponds to the minimum error function. As described above, if a candidate block has the minimum error function, that candidate block will be most similar to the search block; consequently, the displacement vector of the candidate block will be chosen as the motion vector.

FIG. 3 shows a filter window of a 2-dimensional gradient filter employed to derive another embodiment of the weight function.

At the weight function generator 18, the search block is filtered by the 2-dimensional gradient filter to produce a gradient function as the weight function. The gradient function is defined as follows:

wherein ∇(i,j) represents the gradient at a coordinate (i,j) in a search block; I(i+k,j+l) represents a luminance level of a pixel at the coordinate (i+k,j+l); and F(k,l) represents a filter coefficient of a gradient filter.

In a manner similar to the one set forth above, such a weight function will be provided to the blocking matching sections 41 to 49 to weight the error function calculated at the block matching sections 41 to 49 to produce the weighted error function. The weighted error function can be represented as follows:

wherein H x V is the size of a search block; I(i,j) is a luminance level at a pixel coordinate (i,j) in the search block; and P(i,j) is a luminance level at the pixel coordinate (i,j) in a candidate block.

All the weighted error functions derived through the filter will be provided to the minimum error detector 50 to select a motion vector which yields a minimum error, as described above.

While the present invention has been shown and described with reference to the particular embodiments, it will be apparent to those skilled in the art that many changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.