WO2019081928A1

WO2019081928A1 - Image data encoding and decoding

Info

Publication number: WO2019081928A1
Application number: PCT/GB2018/053078
Authority: WO
Inventors: Stephen Mark Keating; Karl James Sharman; Magali Kimlee Miri Philippe
Original assignee: Sony Corporation; Sony Europe Limited
Priority date: 2017-10-27
Filing date: 2018-10-24
Publication date: 2019-05-02
Also published as: GB2567862A; GB201717685D0

Abstract

An image encoding apparatus comprises an intra-image predictor to predict a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predictor comprising: an oversampling filter to oversample at least some of the reference samples, to generate oversampled reference samples; sample projection circuitry to project at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; and an output generator to generate a prediction of the current sample from the projected reference sample array.

Description

IMAGE DATA ENCODING AND DECODING

BACKGROUND

Field

This disclosure relates to image data encoding and decoding.

Description of Related Art

The "background" description provided herein is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description which may not otherwise qualify as prior art at the time of filing, is neither expressly or impliedly admitted as prior art against the present disclosure.

There are several video data encoding and decoding systems which involve transforming video data into a frequency domain representation, quantising the frequency domain coefficients and then applying some form of entropy encoding to the quantised coefficients. This can achieve compression of the video data. A corresponding decoding or decompression technique is applied to recover a reconstructed version of the original video data.

Current video codecs (coder-decoders) such as those used in H.264/MPEG-4 Advanced Video Coding (AVC) achieve data compression primarily by only encoding the differences between successive video frames. These codecs use a regular array of so-called macroblocks, each of which is used as a region of comparison with a corresponding macroblock in a previous video frame, and the image region within the macroblock is then encoded according to the degree of motion found between the corresponding current and previous macroblocks in the video sequence, or between neighbouring macroblocks within a single frame of the video sequence.

High Efficiency Video Coding (HEVC), also known as H.265 or MPEG-H Part 2, is a proposed successor to H.264/MPEG-4 AVC. It is intended for HEVC to improve video quality and double the data compression ratio compared to H.264, and for it to be scalable from 128 ^χ 96 to 7680 x 4320 pixels resolution, roughly equivalent to bit rates ranging from 128kbit/s to 800Mbit/s.

SUMMARY

The present disclosure addresses or mitigates problems arising from this processing. Respective aspects and features of the present disclosure are defined in the appended claims.

It is to be understood that both the foregoing general description and the following detailed description are exemplary, but are not restrictive, of the present technology. BRIEF DESCRIPTION OF THE DRAWINGS

A more complete appreciation of the disclosure and many of the attendant advantages thereof will be readily obtained as the same becomes better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:

Figure 1 schematically illustrates an audio/video (A/V) data transmission and reception system using video data compression and decompression;

Figure 2 schematically illustrates a video display system using video data decompression;

Figure 3 schematically illustrates an audio/video storage system using video data compression and decompression;

Figure 4 schematically illustrates a video camera using video data compression;

Figures 5 and 6 schematically illustrate storage media;

Figure 7 provides a schematic overview of a video data compression and decompression apparatus;

Figure 8 schematically illustrates a predictor;

Figure 9 schematically illustrates a partially-encoded image;

Figure 10 schematically illustrates a set of possible intra-prediction directions;

Figure 11 schematically illustrates a set of prediction modes;

Figure 12 schematically illustrates another set of prediction modes;

Figure 13 schematically illustrates an intra-prediction process;

Figures 14 and 15 schematically illustrate a reference sample projection process;

Figure 16 schematically illustrates a predictor;

Figures 17 and 18 schematically illustrate a reference sample projection process;

Figures 19 to 21 schematically illustrate a reference sample projection process using oversampling;

Figure 22 schematically illustrates a predictor; and

Figures 23 and 24 are schematic flowcharts illustrating respective methods.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Referring now to the drawings, Figures 1-4 are provided to give schematic illustrations of apparatus or systems making use of the compression and/or decompression apparatus to be described below in connection with embodiments of the present technology.

All of the data compression and/or decompression apparatus to be described below may be implemented in hardware, in software running on a general-purpose data processing apparatus such as a general-purpose computer, as programmable hardware such as an application specific integrated circuit (ASIC) or field programmable gate array (FPGA) or as combinations of these. In cases where the embodiments are implemented by software and/or firmware, it will be appreciated that such software and/or firmware, and non-transitory data storage media by which such software and/or firmware are stored or otherwise provided, are considered as embodiments of the present technology.

Figure 1 schematically illustrates an audio/video data transmission and reception system using video data compression and decompression.

An input audio/video signal 10 is supplied to a video data compression apparatus 20 which compresses at least the video component of the audio/video signal 10 for transmission along a transmission route 30 such as a cable, an optical fibre, a wireless link or the like. The compressed signal is processed by a decompression apparatus 40 to provide an output audio/video signal 50. For the return path, a compression apparatus 60 compresses an audio/video signal for transmission along the transmission route 30 to a decompression apparatus 70.

The compression apparatus 20 and decompression apparatus 70 can therefore form one node of a transmission link. The decompression apparatus 40 and decompression apparatus 60 can form another node of the transmission link. Of course, in instances where the transmission link is uni-directional, only one of the nodes would require a compression apparatus and the other node would only require a decompression apparatus.

Figure 2 schematically illustrates a video display system using video data decompression. In particular, a compressed audio/video signal 100 is processed by a decompression apparatus 1 10 to provide a decompressed signal which can be displayed on a display 120. The decompression apparatus 110 could be implemented as an integral part of the display 120, for example being provided within the same casing as the display device. Alternatively, the decompression apparatus 1 10 maybe provided as (for example) a so-called set top box (STB), noting that the expression "set-top" does not imply a requirement for the box to be sited in any particular orientation or position with respect to the display 120; it is simply a term used in the art to indicate a device which is connectable to a display as a peripheral device.

Figure 3 schematically illustrates an audio/video storage system using video data compression and decompression. An input audio/video signal 130 is supplied to a compression apparatus 140 which generates a compressed signal for storing by a store device 150 such as a magnetic disk device, an optical disk device, a magnetic tape device, a solid state storage device such as a semiconductor memory or other storage device. For replay, compressed data is read from the storage device 150 and passed to a decompression apparatus 160 for decompression to provide an output audio/video signal 170.

It will be appreciated that the compressed or encoded signal, and a storage medium such as a machine-readable non-transitory storage medium, storing that signal, are considered as embodiments of the present technology. Figure 4 schematically illustrates a video camera using video data compression. In Figure 4, an image capture device 180, such as a charge coupled device (CCD) image sensor and associated control and read-out electronics, generates a video signal which is passed to a compression apparatus 190. A microphone (or plural microphones) 200 generates an audio signal to be passed to the compression apparatus 190. The compression apparatus 190 generates a compressed audio/video signal 210 to be stored and/or transmitted (shown generically as a schematic stage 220).

The techniques to be described below relate primarily to video data compression and decompression. It will be appreciated that many existing techniques may be used for audio data compression in conjunction with the video data compression techniques which will be described, to generate a compressed audio/video signal. Accordingly, a separate discussion of audio data compression will not be provided. It will also be appreciated that the data rate associated with video data, in particular broadcast quality video data, is generally very much higher than the data rate associated with audio data (whether compressed or uncompressed). It will therefore be appreciated that uncompressed audio data could accompany compressed video data to form a compressed audio/video signal. It will further be appreciated that although the present examples (shown in Figures 1-4) relate to audio/video data, the techniques to be described below can find use in a system which simply deals with (that is to say, compresses, decompresses, stores, displays and/or transmits) video data. That is to say, the embodiments can apply to video data compression without necessarily having any associated audio data handling at all.

Figure 4 therefore provides an example of a video capture apparatus comprising an image sensor and an encoding apparatus of the type to be discussed below. Figure 2 therefore provides an example of a decoding apparatus of the type to be discussed below and a display to which the decoded images are output.

A combination of Figure 2 and 4 may provide a video capture apparatus comprising an image sensor 180 and encoding apparatus 190, decoding apparatus 1 10 and a display 120 to which the decoded images are output.

Figures 5 and 6 schematically illustrate storage media, which store (for example) the compressed data generated by the apparatus 20, 60, the compressed data input to the apparatus 110 or the storage media or stages 150, 220. Figure 5 schematically illustrates a disc storage medium such as a magnetic or optical disc, and Figure 6 schematically illustrates a solid state storage medium such as a flash memory. Note that Figures 5 and 6 can also provide examples of non-transitory machine-readable storage media which store computer software which, when executed by a computer, causes the computer to carry out one or more of the methods to be discussed below. Therefore, the above arrangements provide examples of video storage, capture, transmission or reception apparatuses embodying any of the present techniques.

Figure 7 provides a schematic overview of a video data compression and decompression apparatus.

A controller 343 controls the overall operation of the apparatus and, in particular when referring to a compression mode, controls a trial encoding processes by acting as a selector to select various modes of operation such as block sizes and shapes, and whether the video data is to be encoded losslessly or otherwise. The controller is considered to part of the image encoder or image decoder (as the case may be). Successive images of an input video signal 300 are supplied to an adder 310 and to an image predictor 320. The image predictor 320 will be described below in more detail with reference to Figure 8. The image encoder or decoder (as the case may be) plus the intra-image predictor of Figure 8 may use features from the apparatus of Figure 7. This does not mean that the image encoder or decoder necessarily requires every feature of Figure 7 however.

The adder 310 in fact performs a subtraction (negative addition) operation, in that it receives the input video signal 300 on a "+" input and the output of the image predictor 320 on a "-" input, so that the predicted image is subtracted from the input image. The result is to generate a so-called residual image signal 330 representing the difference between the actual and projected images.

One reason why a residual image signal is generated is as follows. The data coding techniques to be described, that is to say the techniques which will be applied to the residual image signal, tend to work more efficiently when there is less "energy" in the image to be encoded. Here, the term "efficiently" refers to the generation of a small amount of encoded data; for a particular image quality level, it is desirable (and considered "efficient") to generate as little data as is practicably possible. The reference to "energy" in the residual image relates to the amount of information contained in the residual image. If the predicted image were to be identical to the real image, the difference between the two (that is to say, the residual image) would contain zero information (zero energy) and would be very easy to encode into a small amount of encoded data. In general, if the prediction process can be made to work reasonably well such that the predicted image content is similar to the image content to be encoded, the expectation is that the residual image data will contain less information (less energy) than the input image and so will be easier to encode into a small amount of encoded data.

The remainder of the apparatus acting as an encoder (to encode the residual or difference image) will now be described. The residual image data 330 is supplied to a transform unit or circuitry 340 which generates a discrete cosine transform (DCT) representation of blocks or regions of the residual image data. The DCT technique itself is well known and will not be described in detail here. Note also that the use of DCT is only illustrative of one example arrangement. Other transforms which might be used include, for example, the discrete sine transform (DST). A transform could also comprise a sequence or cascade of individual transforms, such as an arrangement in which one transform is followed (whether directly or not) by another transform. The choice of transform may be determined explicitly and/or be dependent upon side information used to configure the encoder and decoder.

The output of the transform unit 340, which is to say, a set of DCT coefficients for each transformed block of image data, is supplied to a quantiser 350. Various quantisation techniques are known in the field of video data compression, ranging from a simple multiplication by a quantisation scaling factor through to the application of complicated lookup tables under the control of a quantisation parameter. The general aim is twofold. Firstly, the quantisation process reduces the number of possible values of the transformed data. Secondly, the quantisation process can increase the likelihood that values of the transformed data are zero. Both of these can make the entropy encoding process, to be described below, work more efficiently in generating small amounts of compressed video data.

A data scanning process is applied by a scan unit 360. The purpose of the scanning process is to reorder the quantised transformed data so as to gather as many as possible of the non-zero quantised transformed coefficients together, and of course therefore to gather as many as possible of the zero-valued coefficients together. These features can allow so-called run-length coding or similar techniques to be applied efficiently. So, the scanning process involves selecting coefficients from the quantised transformed data, and in particular from a block of coefficients corresponding to a block of image data which has been transformed and quantised, according to a "scanning order" so that (a) all of the coefficients are selected once as part of the scan, and (b) the scan tends to provide the desired reordering. One example scanning order which can tend to give useful results is a so-called up-right diagonal scanning order.

The scanned coefficients are then passed to an entropy encoder (EE) 370. Again, various types of entropy encoding may be used. Two examples are variants of the so-called CABAC (Context Adaptive Binary Arithmetic Coding) system and variants of the so-called CAVLC (Context Adaptive Variable-Length Coding) system. In general terms, CABAC is considered to provide a better efficiency, and in some studies has been shown to provide a 10- 20% reduction in the quantity of encoded output data for a comparable image quality compared to CAVLC. However, CAVLC is considered to represent a much lower level of complexity (in terms of its implementation) than CABAC. Note that the scanning process and the entropy encoding process are shown as separate processes, but in fact can be combined or treated together. That is to say, the reading of data into the entropy encoder can take place in the scan order. Corresponding considerations apply to the respective inverse processes to be described below. The output of the entropy encoder 370, along with additional data (mentioned above and/or discussed below), for example defining the manner in which the predictor 320 generated the predicted image, provides a compressed output video signal 380.

However, a return path is also provided because the operation of the predictor 320 itself depends upon a decompressed version of the compressed output data.

The reason for this feature is as follows. At the appropriate stage in the decompression process (to be described below) a decompressed version of the residual data is generated. This decompressed residual data has to be added to a predicted image to generate an output image (because the original residual data was the difference between the input image and a predicted image). In order that this process is comparable, as between the compression side and the decompression side, the predicted images generated by the predictor 320 should be the same during the compression process and during the decompression process. Of course, at decompression, the apparatus does not have access to the original input images, but only to the decompressed images. Therefore, at compression, the predictor 320 bases its prediction (at least, for inter-image encoding) on decompressed versions of the compressed images.

The entropy encoding process carried out by the entropy encoder 370 is considered (in at least some examples) to be "lossless", which is to say that it can be reversed to arrive at exactly the same data which was first supplied to the entropy encoder 370. So, in such examples the return path can be implemented before the entropy encoding stage. Indeed, the scanning process carried out by the scan unit 360 is also considered lossless, but in the present embodiment the return path 390 is from the output of the quantiser 350 to the input of a complimentary inverse quantiser 420. In instances where loss or potential loss is introduced by a stage, that stage may be included in the feedback loop formed by the return path. For example, the entropy encoding stage can at least in principle be made lossy, for example by techniques in which bits are encoded within parity information. In such an instance, the entropy encoding and decoding should form part of the feedback loop.

In general terms, an entropy decoder 410, the reverse scan unit 400, an inverse quantiser 420 and an inverse transform unit or circuitry 430 provide the respective inverse functions of the entropy encoder 370, the scan unit 360, the quantiser 350 and the transform unit 340. For now, the discussion will continue through the compression process; the process to decompress an input compressed video signal will be discussed separately below.

In the compression process, the scanned coefficients are passed by the return path 390 from the quantiser 350 to the inverse quantiser 420 which carries out the inverse operation of the scan unit 360. An inverse quantisation and inverse transformation process are carried out by the units 420, 430 to generate a compressed-decompressed residual image signal 440. The image signal 440 is added, at an adder 450, to the output of the predictor 320 to generate a reconstructed output image 460. This forms one input to the image predictor 320, as will be described below.

Turning now to the process applied to decompress a received compressed video signal 470, the signal is supplied to the entropy decoder 410 and from there to the chain of the reverse scan unit 400, the inverse quantiser 420 and the inverse transform unit 430 before being added to the output of the image predictor 320 by the adder 450. So, at the decoder side, the decoder reconstructs a version of the residual image and then applies this (by the adder 450) to the predicted version of the image (on a block by block basis) so as to decode each block. In straightforward terms, the output 460 of the adder 450 forms the output decompressed video signal 480. In practice, further filtering may optionally be applied (for example, by a filter 560 shown in Figure 8 but omitted from Figure 7 for clarity of the higher level diagram of Figure 7) before the signal is output.

The apparatus of Figures 7 and 8 can act as a compression (encoding) apparatus or a decompression (decoding) apparatus. The functions of the two types of apparatus substantially overlap. The scan unit 360 and entropy encoder 370 are not used in a decompression mode, and the operation of the predictor 320 (which will be described in detail below) and other units follow mode and parameter information contained in the received compressed bit-stream rather than generating such information themselves.

Figure 8 schematically illustrates the generation of predicted images, and in particular the operation of the image predictor 320.

There are two basic modes of prediction carried out by the image predictor 320: so- called intra-image prediction and so-called inter-image, or motion-compensated (MC), prediction. At the encoder side, each involves detecting a prediction direction in respect of a current block to be predicted, and generating a predicted block of samples according to other samples (in the same (intra) or another (inter) image). By virtue of the units 310 or 450, the difference between the predicted block and the actual block is encoded or applied so as to encode or decode the block respectively.

(At the decoder, or at the reverse decoding side of the encoder, the detection of a prediction direction may be in response to data associated with the encoded data by the encoder, indicating which direction was used at the encoder. Or the detection may be in response to the same factors as those on which the decision was made at the encoder).

Intra-image prediction bases a prediction of the content of a block or region of the image on data from within the same image. This corresponds to so-called l-frame encoding in other video compression techniques. In contrast to l-frame encoding, however, which involves encoding the whole image by intra-encoding, in the present embodiments the choice between intra- and inter- encoding can be made on a block-by-block basis, though in other embodiments the choice is still made on an image-by-image basis.

Motion-compensated prediction is an example of inter-image prediction and makes use of motion information which attempts to define the source, in another adjacent or nearby image, of image detail to be encoded in the current image. Accordingly, in an ideal example, the contents of a block of image data in the predicted image can be encoded very simply as a reference (a motion vector) pointing to a corresponding block at the same or a slightly different position in an adjacent image.

A technique known as "block copy" prediction is in some respects a hybrid of the two, as it uses a vector to indicate a block of samples at a position displaced from the currently predicted block within the same image, which should be copied to form the currently predicted block.

Returning to Figure 8, two image prediction arrangements (corresponding to intra- and inter-image prediction) are shown, the results of which are selected by a multiplexer 500 under the control of a mode signal 510 (for example, from the controller 343) so as to provide blocks of the predicted image for supply to the adders 310 and 450. The choice is made in dependence upon which selection gives the lowest "energy" (which, as discussed above, may be considered as information content requiring encoding), and the choice is signalled to the decoder within the encoded output data-stream. Image energy, in this context, can be detected, for example, by carrying out a trial subtraction of an area of the two versions of the predicted image from the input image, squaring each pixel value of the difference image, summing the squared values, and identifying which of the two versions gives rise to the lower mean squared value of the difference image relating to that image area. In other examples, a trial encoding can be carried out for each selection or potential selection, with a choice then being made according to the cost of each potential selection in terms of one or both of the number of bits required for encoding and distortion to the picture.

The actual prediction, in the intra-encoding system, is made on the basis of image blocks received as part of the signal 460, which is to say, the prediction is based upon encoded- decoded image blocks in order that exactly the same prediction can be made at a decompression apparatus. However, data can be derived from the input video signal 300 by an intra-mode selector 520 to control the operation of the intra-image predictor 530.

For inter-image prediction, a motion compensated (MC) predictor 540 uses motion information such as motion vectors derived by a motion estimator 550 from the input video signal 300. Those motion vectors are applied to a processed version of the reconstructed image 460 by the motion compensated predictor 540 to generate blocks of the inter-image prediction.

Accordingly, the units 530 and 540 (operating with the estimator 550) each act as detectors to detect a prediction direction in respect of a current block to be predicted, and as a generator to generate a predicted block of samples (forming part of the prediction passed to the units 310 and 450) according to other samples defined by the prediction direction.

The processing applied to the signal 460 will now be described. Firstly, the signal is optionally filtered by a filter unit 560, which will be described in greater detail below. This involves applying a "deblocking" filter to remove or at least tend to reduce the effects of the block-based processing carried out by the transform unit 340 and subsequent operations. A sample adaptive offsetting (SAO) filter may also be used. Also, an adaptive loop filter is optionally applied using coefficients derived by processing the reconstructed signal 460 and the input video signal 300. The adaptive loop filter is a type of filter which, using known techniques, applies adaptive filter coefficients to the data to be filtered. That is to say, the filter coefficients can vary in dependence upon various factors. Data defining which filter coefficients to use is included as part of the encoded output data-stream.

The filtered output from the filter unit 560 in fact forms the output video signal 480 when the apparatus is operating as a decompression apparatus. It is also buffered in one or more image or frame stores 570; the storage of successive images is a requirement of motion compensated prediction processing, and in particular the generation of motion vectors. To save on storage requirements, the stored images in the image stores 570 may be held in a compressed form and then decompressed for use in generating motion vectors. For this particular purpose, any known compression / decompression system may be used. The stored images are passed to an interpolation filter 580 which generates a higher resolution version of the stored images; in this example, intermediate samples (sub-samples) are generated such that the resolution of the interpolated image is output by the interpolation filter 580 is 4 times (in each dimension) that of the images stored in the image stores 570 for the luminance channel of 4:2:0 and 8 times (in each dimension) that of the images stored in the image stores 570 for the chrominance channels of 4:2:0. The interpolated images are passed as an input to the motion estimator 550 and also to the motion compensated predictor 540.

The way in which an image is partitioned for compression processing will now be described. At a basic level, an image to be compressed is considered as an array of blocks or regions of samples. The splitting of an image into such blocks or regions can be carried out by a decision tree, such as that described in Bross et al: "High Efficiency Video Coding (HEVC) text specification draft 6", JCTVC-H1003_dO (November 201 1), the contents of which are incorporated herein by reference. In some examples, the resulting blocks or regions have sizes and, in some cases, shapes which, by virtue of the decision tree, can generally follow the disposition of image features within the image. This in itself can allow for an improved encoding efficiency because samples representing or following similar image features would tend to be grouped together by such an arrangement. In some examples, square blocks or regions of different sizes (such as 4x4 samples up to, say, 64x64 or larger blocks) are available for selection. In other example arrangements, blocks or regions of different shapes such as rectangular blocks (for example, vertically or horizontally oriented) can be used. Other non- square and non-rectangular blocks are envisaged. The result of the division of the image into such blocks or regions is (in at least the present examples) that each sample of an image is allocated to one, and only one, such block or region.

The intra-prediction process will now be discussed. In general terms, intra-prediction involves generating a prediction of a current block of samples from previously-encoded and decoded samples in the same image.

Figure 9 schematically illustrates a partially encoded image 800. Here, the image is being encoded from top-left to bottom-right on a block by block basis. An example block encoded partway through the handling of the whole image is shown as a block 810. A shaded region 820 above and to the left of the block 810 has already been encoded. The intra-image prediction of the contents of the block 810 can make use of any of the shaded area 820 but cannot make use of the unshaded area below that.

In some examples, the image is encoded on a block by block basis such that larger blocks (referred to as coding units or CUs) are encoded in an order such as the order discussed with reference to Figure 9. Within each CU, there is the potential (depending on the block splitting process that has taken place) for the CU to be handled as a set of two or more smaller blocks or transform units (TUs). This can give a hierarchical order of encoding so that the image is encoded on a CU by CU basis, and each CU is potentially encoded on a TU by TU basis. Note however that for an individual TU within the current coding tree unit (the largest node in the tree structure of block division), the hierarchical order of encoding (CU by CU then TU by TU) discussed above means that there may be previously encoded samples in the current CU and available to the coding of that TU which are, for example, above-right or below- left of that TU.

The block 810 represents a CU; as discussed above, for the purposes of intra-image prediction processing, this may be subdivided into a set of smaller units. An example of a current TU 830 is shown within the CU 810. More generally, the picture is split into regions or groups of samples to allow efficient coding of signalling information and transformed data. The signalling of the information may require a different tree structure of sub-divisions to that of the transform, and indeed that of the prediction information or the prediction itself. For this reason, the coding units may have a different tree structure to that of the transform blocks or regions, the prediction blocks or regions and the prediction information. In some examples such as HEVC the structure can be a so-called quad tree of coding units, whose leaf nodes contain one or more prediction units and one or more transform units; the transform units can contain multiple transform blocks corresponding to luma and chroma representations of the picture, and prediction could be considered to be applicable at the transform block level. In examples, the parameters applied to a particular group of samples can be considered to be predominantly defined at a block level, which is potentially not of the same granularity as the transform structure.

The intra-image prediction takes into account samples coded prior to the current TU being considered, such as those above and/or to the left of the current TU. Source samples, from which the required samples are predicted, may be located at different positions or directions relative to the current TU. To decide which direction is appropriate for a current prediction unit, the mode selector 520 of an example encoder may test all combinations of available TU structures for each candidate direction and select the prediction direction and TU structure with the best compression efficiency.

The picture may also be encoded on a "slice" basis. In one example, a slice is a horizontally adjacent group of CUs. But in more general terms, the entire residual image could form a slice, or a slice could be a single CU, or a slice could be a row of CUs, and so on. Slices can give some resilience to errors as they are encoded as independent units. The encoder and decoder states are completely reset at a slice boundary. For example, intra-prediction is not carried out across slice boundaries; slice boundaries are treated as image boundaries for this purpose.

Figure 10 schematically illustrates a set of possible (candidate) prediction directions. The full set of candidate directions is available to a prediction unit. The directions are determined by horizontal and vertical displacement relative to a current block position, but are encoded as prediction "modes", a set of which is shown in Figure 1 1. Note that the so-called DC mode represents a simple arithmetic mean of the surrounding upper and left-hand samples. Note also that the set of directions shown in Figure 10 is just one example; in other examples, a set of (for example) 65 angular modes plus DC and planar (a full set of 67 modes) as shown schematically in Figure 12 makes up the full set. Other numbers of modes could be used.

In general terms, after detecting a prediction direction, the systems are operable to generate a predicted block of samples according to other samples defined by the prediction direction. In examples, the image encoder is configured to encode data identifying the prediction direction selected for each sample or region of the image.

Figure 13 schematically illustrates an intra-prediction process in which a sample 900 of a block or region 910 of samples is derived from other reference samples 920 of the same image according to a direction 930 defined by the intra-prediction mode associated with that sample. The reference samples 920 in this example come from blocks above and to the left of the block 910 in question and the predicted value of the sample 900 is obtained by tracking along the direction 930 to the reference samples 920. The direction 930 might point to a single individual reference sample but in a more general case an interpolated value between surrounding reference samples is used as the prediction value. Note that the block 910 could be square as shown in Figure 13 or could be another shape such as rectangular.

Figures 14 and 15 schematically illustrate a previously proposed reference sample projection process.

In Figures 14 and 15, a block or region 1400 of samples to be predicted is surrounded by linear arrays of reference samples from which the intra prediction of the predicted samples takes place. The reference samples 1410 are shown as shaded blocks in Figures 14 and 15, and the samples to be predicted are shown as unshaded blocks. Note that an 8x8 block or region of samples to be predicted is used in this example, but the techniques are applicable to variable block sizes and indeed block shapes.

As mentioned, the reference samples comprise at least two linear arrays in respective orientations with respect to the current image region of samples to be predicted. For example, the linear arrays may be an array or row 1420 of samples above the block of samples to be predicted and an array or column 1430 of samples to the left of the block of samples to be predicted.

As discussed above with reference to Figure 13, the reference sample arrays can extend beyond the extent of the block to be predicted, in order to provide for prediction modes or directions within the range indicated in Figures 10-12. Where necessary, if previously decoded samples are not available for use as reference samples at particular reference sample positions, other reference samples can be re-used at those missing positions. Reference sample filtering processes can be used on the reference samples.

A sample projection process is used to project at least some of the reference samples to different respective positions with respect to the current image region, in the manner shown in Figures 14 and 15. In other words, in embodiments, the projection process and circuitry operates to represent at least some of the reference samples at different spatial positions relative to the current image region, for example as shown in Figures 14 and 15. Thus at least some reference samples may be moved (for the purposes at least of defining an array of reference samples from which samples are predicted) with respect to their relative positions to the current image region. In particular, Figure 14 relates to a projection process performed for modes which are generally to the left of the diagonal mode (18 in Figure 1 1) mainly modes

2... 17, and Figure 15 schematically illustrates a reference sample projection carried for modes 19...34, namely those generally above the block to be predicted (to the right of the diagonal mode 18). The diagonal mode 18 can be assigned to either of these two groups as an arbitrary selection, such as to the group of modes to the right of the diagonal. So, in Figure 14, when the current prediction mode is between modes 2 and 17 (or their equivalent in a system such as that of Figure 12 having a different number of possible prediction modes), the sample array 1420 above the current block is projected to form additional reference samples 1440 in the left hand column. Prediction then takes place with respect to the linear projected array 1450 formed of the original left hand column 1430 and the projected samples 1440. In Figure 15, for modes between 18 and 34 of Figure 1 1 (or their equivalent in other sets of prediction modes such as those shown in Figure 12), the reference samples 1500 in the left hand column are projected so as to extend to the left of the reference samples 1510 above the current block. This forms a projected array 1520.

One reason why projection of this nature is carried out is to reduce the complexity of the intra prediction process, in that all of the samples to be predicted are then referencing a single linear array of reference samples, rather than referencing two orthogonal linear arrays.

Figure 16 schematically illustrates a previously proposed prediction circuitry 600 arranged to carry out the projection process of Figures 14 and 15, specifically by providing projection circuitry 1610 configured to carry out a projection process on the reference samples currently selected for a block of region to be predicted. The projected reference samples are stored in a buffer 1620 to be accessed by an intra predictor 1630 to generate predicted samples from the projected reference samples. The projection process is carried out according to the prediction mode associated with the current block to be predicted, using the techniques discussed in connection with Figures 14 and 15.

In some embodiments, the same projection process is carried out in the decoder and in the encoder, so that the predicted samples are the same in each instance.

A potential problem with the previously proposed projection process described with reference to Figure 14 and 15 will now be described with reference to Figure 17 and 18.

For comparison, Figure 17 schematically illustrates the prediction of predicted samples 1700 from reference samples (shaded) 1710 for an upwards-diagonal mode, such as the prediction mode 15 (using the numbering of Figure 1 1). Figure 17 shows the situation in the absence of reference sample projection and Figure 18 shows the same prediction mode 15 being handled using a previously proposed reference sample projection process.

In the case of Figure 17, the prediction of a row 1720 of predicted samples using the prediction mode 15 is based upon seven of the reference samples numbered 0...6 in the row 1710, as indicated by arrows 1730 indicating the location amongst the reference samples on which each predicted sample is based. Typically, sample prediction is carried out at a positional accuracy corresponding to less than one sample width. For example the sample prediction can be carried out by interpolation to a 1/32 sample position accuracy, with respect to the spatial separation of the predicted samples (and of the reference samples in the arrangement of Figure 17).

Referring to Figure 18, again using mode 15 as the example, carrying out the projection process to provide projected samples across the range from the sample 0...6 in the projected array 18 means that only samples 0, 2, 4, 6 are projected. This means that the row of predicted samples 1720 is instead derived only from four samples rather than 7 as in Figure 17.

So, the projection process has resulted in a loss of information from which the sample prediction can take place. Even though, for example, a 1/32 sample position accuracy may be still used in the prediction process carried out on the projected reference samples of Figure 18, the prediction process itself is based upon less information (four reference samples rather than seven reference samples) and so the result may be potentially a less accurate prediction.

Accuracy of prediction is generally advantageous so as to reduce the energy in the residual image, which in turn can mean that less data is required to encode the residual image.

Figure 19 to 21 schematically illustrate the operation of a reference sample projection process using oversampling.

In Figure 19, a sample 1900 to be predicted has an associated prediction mode or direction such as the prediction mode 15. An arrow 1910 indicates the position with respect to an (unprojected) reference sample array on which the predicted sample 1900 will be derived.

Projection is used as discussed above. However, a difference with respect to the previously proposed arrangements already described is that before the prediction process takes place, and before any projection takes place, the reference samples are oversampled or super- sampled to a higher sampling frequency or smaller sample separation. The projection process then takes place on the basis of the oversampled samples to produce a reference sample array 2100 in Figure 21 from which the samples to be predicted 21 10 are derived.

Figure 20 schematically illustrates the oversampling process, for an example super- sampling or oversampling ratio or rate of 4x. Note that in embodiments, a different rate (such as 32x) can be used, and/or (as discussed below) a different rate can be used for reference samples which, by virtue of the projection process, are to be projected to new positions relative to the block or region of samples to be projected, compared to samples which are not to be moved by the projection process.

Figure 20 provides a representation of a top-left corner region of the arrangement of Figure 19, in which reference samples are shown as a horizontal row of reference samples 2000 and a vertical column of reference samples 2010, shown as large crosses (X) in Figure 20. A sample 2030 is considered to be in both groups.

Smaller crosses (x) 2040, 2050 in Figure 20 schematically illustrate the positions of oversampled reference samples, and remaining larger crosses (X) 2060 represent the positions of at least some of the samples to be predicted.

Values for the oversampled samples 2040, 2050 are generated by a super-sampling or oversampling process with respect to the reference samples 2000, 2010 respectively. For example, the values for the oversampled samples 2040, 2050 can be derived by inserting zero values in between values for the reference samples 2000, 2010 and then filtering. By carrying out the oversampling before the projection process, information from all of the reference samples in the array 2000 to be projected can be taken into account in generating the oversampled reference samples which are then projected.

In some examples, the oversampling can be such that the oversampled array 2100 has a sampling frequency of (say) 32 times that of the original reference samples, or another factor can be used, such that the prediction process then becomes the selection of a nearest neighbouring oversampled sample (that is to say, nearest to the position, amongst the reference samples, pointed to by the current mode direction from a predicted sample position, rather than requiring a further interpolation process to be carried out.

Although it is not strictly necessary to oversample those of the reference samples which are not being projected (the samples in the vertical array 2010 in Figure 20), oversampling those reference samples as well can provide lower complexity at the prediction stage by providing a uniform oversampled array 2100 in Figure 21. This can be useful if a nearest neighbour selection process such as that mentioned above is used.

Note that depending upon the prediction mode or angle in use, the projection process may be such that projected samples 2020 (collectively the samples 2000 and the oversampled samples 2040 in this example) are projected to a region 2120 in Figure 21 , in a similar way to Figure 18 where seven samples were projected to the space normally occupied by four samples, the extent or length of the region 2120 can be less than that of the region 2020. To ensure a uniform oversampled sample frequency in the projected array 2100, the oversampling process can take this difference into account so as to generate all of the oversampled samples in the array 2100 to have the same (for example 32x) oversampling rate. For example, the top row 21 10 could be oversampled or super-sampled at an oversampling rate of 32x to form the samples 2020 and then subsampled as part of the projection process to form the samples 2120. Alternatively if the projection were in respect of (purely as an example) a prediction direction of 30 degrees, the row of samples 2020 could be generated by super-sampling at a rate of 32tan(30)= 18.47 times, such that the projection process is a 1 : 1 mapping.

In practice, as long as the super-sampling rate is sufficiently high then little accuracy would be lost by super-sampling by a fixed rate and sub-sampling for the projection. In other words, therefore, the samples to be projected (the top row 2020 in Figure 20, or the left column in the case of modes 2-17) can be oversampled at a predetermined oversampling factor and then interpolated or subsampled to form the oversampled reference samples 2120 in Figure 21.

Figure 22 schematically illustrates an intra-image prediction circuitry or predictor 2200 (for use in, or forming part of, an image data encoding and/or compression apparatus or an image data decoding and/or decompression apparatus), to predict a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples. The intra-image prediction circuitry comprises an

oversampling filter 2210 to oversample at least some of the reference samples, to generate oversampled reference samples; sample projection circuitry 2220 to project at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; a buffer 2230 to buffer the projected oversampled samples; and an output generator or intra predictor 2240 to generate a prediction of the current sample from the projected reference sample array. The apparatus can operate under the control of a controller 2250.

As discussed above, the reference samples can comprise at least two linear arrays in respective orientations with respect to the current image region, with the oversampling filter 2210 being configured to oversample reference samples in at least a first linear array of the at least two linear arrays; the projection circuitry 2220 can project the oversampled reference samples in the first linear array at respective array positions aligned with a longitudinal axis of a second linear array of the at least two linear arrays.

The oversampling can be such that instead of an interpolation process needing to be used by the intra predictor 2240, instead the intra predictor 2240 can generate a prediction of the current sample by selecting a nearest one of the oversampled reference samples pointed to, in the prediction direction, from the current sample.

In the examples given above, the current image region is a rectangular image region and the orientations of the at least two linear arrays are horizontal and vertical orientations relative to the current image region. In this case, for a prediction direction nearer to a horizontal than a vertical direction (for example modes 2-17 or their corresponding angles), the

oversampling filter is configured to oversample reference samples in at least a horizontally oriented linear array of the at least two linear arrays; and for a prediction direction nearer to a vertical than a horizontal direction (for example modes 19-34 or their corresponding angles), the oversampling filter is configured to oversample reference samples in at least a vertically oriented linear array of the at least two linear arrays. As discussed above, a diagonal mode (18) can be assigned to an arbitrary one of these groups, for example to the group of 19-34. If a different set of modes is used, such as that shown in Figure 12, corresponding groupings can be provided.

As discussed above, in order to provide an array of projected samples having a consistent oversampled sample rate (as between those which have been projected and those which have not), the oversampling filter is configured to select an oversampling factor according to the prediction direction. In example embodiments this selection can be handled by the controller 2250.

Figure 23 is a schematic flowchart illustrating an image encoding method comprising: intra-image predicting (at a step 2300) a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predicting step comprising:

oversampling (at a step 2310) at least some of the reference samples, to generate oversampled reference samples;

projecting (at a step 2320) at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; and

generating (at a step 2330) a prediction of the current sample from the projected reference sample array.

Figure 24 is a schematic flowchart illustrating an image decoding method comprising: intra-image predicting (at a step 2400) a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predicting step comprising:

oversampling (at a step 2410) at least some of the reference samples, to generate oversampled reference samples;

projecting (at a step 2420) at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; and

generating (at a step 2430) a prediction of the current sample from the projected reference sample array.

The techniques discussed above can apply in isolation to one or more components of a video sampling scheme, for example when luminance and or chrominance and/or components of chrominance samples are differently sampled.

It will be appreciated that the various different techniques described can be combined so that the selection of the set of candidate modes applicable to a sample or to a block or region of samples can take into account any permutation of one or more (being a subset or the whole group) of considerations or aspects discussed above.

In so far as embodiments of the disclosure have been described as being implemented, at least in part, by software-controlled data processing apparatus, it will be appreciated that a non-transitory machine-readable medium carrying such software, such as an optical disk, a magnetic disk, semiconductor memory or the like, is also considered to represent an embodiment of the present disclosure. Similarly, a data signal comprising coded data generated according to the methods discussed above (whether or not embodied on a non- transitory machine-readable medium) is also considered to represent an embodiment of the present disclosure.

It will be apparent that numerous modifications and variations of the present disclosure are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended clauses, the technology may be practised otherwise than as specifically described herein.

Respective aspects and features are defined by the following numbered clauses:

1. An image encoding apparatus comprising:

an intra-image predictor configured to predict a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predictor comprising:

an oversampling filter configured to oversample at least some of the reference samples, to generate oversampled reference samples;

sample projection circuitry configured to project at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; and

an output generator configured to generate a prediction of the current sample from the projected reference sample array.

2. Apparatus according to clause 1 , in which the reference samples comprise at least two linear arrays in respective orientations with respect to the current image region.

3. Apparatus according to clause 2, in which the oversampling filter is configured to oversample reference samples in at least a first linear array of the at least two linear arrays.

4. Apparatus according to clause 3, in which the sample projection circuitry is configured to project the oversampled reference samples in the first linear array at respective array positions aligned with a longitudinal axis of a second linear array of the at least two linear arrays.

5. Apparatus according to any one of clauses 1 to 4, in which the output generator is configured to generation the prediction of the current sample by selecting a nearest one of the oversampled reference samples pointed to, in the prediction direction, from the current sample.

6. Apparatus according to clause 3, in which:

the current image region is a rectangular image region; and

the orientations of the at least two linear arrays comprise horizontal and vertical orientations relative to the current image region.

7. Apparatus according to clause 6, in which: for a prediction direction nearer to a horizontal than a vertical direction, the oversampling filter is configured to oversample reference samples in at least a horizontally oriented linear array of the at least two linear arrays; and

for a prediction direction nearer to a vertical than a horizontal direction, the oversampling filter is configured to oversample reference samples in at least a vertically oriented linear array of the at least two linear arrays.

8. Apparatus according to any one of the preceding clauses, in which the oversampling filter is configured to select an oversampling factor according to the prediction direction.

9. Video storage, capture, transmission or reception apparatus comprising apparatus according to any one of the preceding clauses.

10. An image decoding apparatus comprising:

1 1. Apparatus according to clause 10, in which the reference samples comprise at least two linear arrays in respective orientations with respect to the current image region.

12. Apparatus according to clause 11 , in which the oversampling filter is configured to oversample reference samples in at least a first linear array of the at least two linear arrays.

13. Apparatus according to clause 12, in which the sample projection circuitry is configured to project the oversampled reference samples in the first linear array at respective array positions aligned with a longitudinal axis of a second linear array of the at least two linear arrays.

14. Apparatus according to any one of clauses 10 to 13, in which the output generator is configured to generation the prediction of the current sample by selecting a nearest one of the oversampled reference samples pointed to, in the prediction direction, from the current sample. 15. Apparatus according to clause 12, in which:

the current image region is a rectangular image region; and the orientations of the at least two linear arrays comprise horizontal and vertical orientations relative to the current image region.

16. Apparatus according to clause 15, in which:

for a prediction direction nearer to a horizontal than a vertical direction, the oversampling filter is configured to oversample reference samples in at least a horizontally oriented linear array of the at least two linear arrays; and

17. Apparatus according to any one of clauses 10 to 16, in which the oversampling filter is configured to select an oversampling factor according to the prediction direction.

18. Video storage, capture, transmission or reception apparatus comprising apparatus according to any one of clauses 10 to 17.

19. An image encoding method comprising:

intra-image predicting a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predicting step comprising:

oversampling at least some of the reference samples, to generate oversampled reference samples;

projecting at least some of the oversampled reference samples to different respective positions with respect to the current image region, so as to generate a projected reference sample array; and

generating a prediction of the current sample from the projected reference sample array.

20. Computer software which, when executed by a computer, causes the computer to carry out a method according to clause 19.

21. A machine-readable non-transitory storage medium which stores software according to clause 20.

22. A data signal comprising coded data generated according to the method of clause 19.

23. An image decoding method comprising:

intra-image predicting a current sample of a current image region, of a plurality of regions of an image, with respect to one or more corresponding reference samples of the same image at respective positions relative to the current image region, according to a prediction direction between the current sample and a reference position amongst the reference samples, the intra-image predicting step comprising: oversampling at least some of the reference samples, to generate oversampled reference samples;

24. Computer software which, when executed by a computer, causes the computer to carry out a method according to clause 23.

25. A machine-readable non-transitory storage medium which stores software according to clause 24.

26. A video capture apparatus comprising an image sensor and the encoding apparatus of any one of clauses 1-9, decoding apparatus of any one of clauses 10-17 and a display to which the decoded images are output.

Claims

1. An image encoding apparatus comprising:

2. Apparatus according to claim 1 , in which the reference samples comprise at least two linear arrays in respective orientations with respect to the current image region.

3. Apparatus according to claim 2, in which the oversampling filter is configured to oversample reference samples in at least a first linear array of the at least two linear arrays.

4. Apparatus according to claim 3, in which the sample projection circuitry is configured to project the oversampled reference samples in the first linear array at respective array positions aligned with a longitudinal axis of a second linear array of the at least two linear arrays.

5. Apparatus according to claim 1 , in which the output generator is configured to generation the prediction of the current sample by selecting a nearest one of the oversampled reference samples pointed to, in the prediction direction, from the current sample.

6. Apparatus according to claim 3, in which:

the current image region is a rectangular image region; and

7. Apparatus according to claim 6, in which: for a prediction direction nearer to a horizontal than a vertical direction, the oversampling filter is configured to oversample reference samples in at least a horizontally oriented linear array of the at least two linear arrays; and

8. Apparatus according to claim 1 , in which the oversampling filter is configured to select an oversampling factor according to the prediction direction.

9. Video storage, capture, transmission or reception apparatus comprising apparatus according to claim 1.

10. An image decoding apparatus comprising:

1 1. Apparatus according to claim 10, in which the reference samples comprise at least two linear arrays in respective orientations with respect to the current image region.

12. Apparatus according to claim 1 1 , in which the oversampling filter is configured to oversample reference samples in at least a first linear array of the at least two linear arrays.

13. Apparatus according to claim 12, in which the sample projection circuitry is configured to project the oversampled reference samples in the first linear array at respective array positions aligned with a longitudinal axis of a second linear array of the at least two linear arrays.

14. Apparatus according to claim 10, in which the output generator is configured to generation the prediction of the current sample by selecting a nearest one of the oversampled reference samples pointed to, in the prediction direction, from the current sample.

15. Apparatus according to claim 12, in which:

the current image region is a rectangular image region; and

16. Apparatus according to claim 15, in which:

17. Apparatus according to claim 10, in which the oversampling filter is configured to select an oversampling factor according to the prediction direction.

18. Video storage, capture, transmission or reception apparatus comprising apparatus according to claim 10.

19. An image encoding method comprising:

20. Computer software which, when executed by a computer, causes the computer to carry out a method according to claim 19.

21. A machine-readable non-transitory storage medium which stores software according to claim 20.

22. A data signal comprising coded data generated according to the method of claim 19.

23. An image decoding method comprising:

24. Computer software which, when executed by a computer, causes the computer to carry out a method according to claim 23.

25. A machine-readable non-transitory storage medium which stores software according to claim 24.