Abstract: A method for processing a current image of a set of images is proposed, which comprises, at a video processing device comprising a processor: determining a target pixel area in the current image; obtaining a hidden image, comprising a source pixel area, of the set of images; dividing the current image into blocks for an encoding of the blocks according to an encoding sequence; and for a block of the target pixel area: determining a motion vector pointing to a block of the source pixel area in the hidden image corresponding to the block of the target pixel area, and based on the motion vector, encoding the block of the target pixel area according to a temporal correlation prediction mode pursuant to which no pixel residual representing a difference between the block of the target pixel area and the corresponding block of the source pixel area is generated.

Abstract: Systems, methods, and apparatuses are described for compressing digital content. The digital content may comprise a plurality of frames. The plurality of frames may comprise a plurality of crossfade frames. A first boundary frame of the crossfade frames may be determined. A second boundary frame of the crossfade frames may be determined. At least a portion of the crossfade frames may be coded as inter-predicted frames using a weighting factor and based on the first boundary frame or the second boundary frame.

Abstract: A computer-implemented method and a compression unit for performing lossy compression on a block of image data in accordance with a multi-level difference table. The block of image data comprises a plurality of image element values, wherein each level of the multi-level difference table comprises a plurality of entries. An origin value for the block of image data is determined. A level within the multi-level difference table for the block of image data is determined. For each image element value in the block of image data, one of the entries at the determined level within the multi-level difference table is selected. A compressed block of data for the block of image data is formed, wherein the compressed block of data comprises: (i) data representing the determined origin value, (ii) an indication of the determined level, and (iii) for each image element value in the block of image data, an indication of the selected entry for that image element value.

Abstract: A method for visual media processing includes processing a bitstream representation for a conversion between a current video block in a video region of a visual media data and the bitstream representation. The bitstream representation is based on a rule that specifies that one or more syntax elements used in a luma-dependent chroma residual scaling step are selectively included in the bitstream representation. The luma-dependent chroma residual scaling step includes scaling chroma samples based on neighboring reconstructed luma samples during the conversion. The one or more syntax elements are usable to derive a chroma residual scaling factor used for the scaling.

Abstract: An encoded video sequence including a plurality of images may be decoded. First and second adjacent blocks of an image of an encoded video sequence may be provided. A line of pixels including pixels of the first and second blocks may be defined extending across a boundary between the first and second blocks. A first reference value may be calculated based on a first pixel of the pixels from the first block. A second reference value may be calculated based on a second pixel of the pixels from the second block. Filtered pixel values for each pixel of the line of pixels between the first pixel and the second pixel may be calculated using interpolation based on the first and/or second reference values. Filtered blocks may be generated using the filtered pixel values, and a decoded video sequence may be generated based on the filtered blocks.

Abstract: Systems and techniques are described for encoding and/or decoding data based on motion estimation that applies variable-scale warping. An encoding device can receive an input frame and a reference frame that depict a scene at different times. The encoding device can generate an optical flow identifying movements in the scene between the two frames. The encoding device can generate a weight map identifying how finely or coarsely the reference frame can be warped for input frame prediction. The encoding device can generate encoded video data based on the optical flow and the weight map. A decoding device can generate a reconstructed optical flow and a reconstructed weight map from the encoded data. A decoding device can generate a prediction frame by warping the reference frame based on the reconstructed optical flow and the reconstructed weight map. The decoding device can generate a reconstructed input frame based on the prediction frame.

Abstract: Systems, methods, and apparatuses are described for compressing digital content. The digital content may comprise a plurality of frames. The plurality of frames may comprise a plurality of crossfade frames. A first boundary frame of the crossfade frames may be determined. A second boundary frame of the crossfade frames may be determined. At least a portion of the crossfade frames may be coded as inter-predicted frames using a weighting factor and based on the first boundary frame or the second boundary frame.

Abstract: A method decoding a data values set includes: decoding a first portion of each data value from one or more data sets; decoding a second portion depending on integer bits of data values not fully encoded by the data sets, and, if a data value has not been fully decoded by first and second portions, decoding a remaining third portion of the data value; detecting, for a subset of the data values, (i) instances of data values for which a third portion has been encoded and would still have been required had a higher value of n been used, and (ii) instances of data values for which a second portion has been encoded but the value of n was such that the data value could have been fully encoded by first and second portions using a lower value of n; and varying n for use in subsequent data values.

Abstract: In embodiments, obtaining a plurality of image sets associated with a geographical region and a time period, wherein each image set of the plurality of image sets comprises multi-spectral and time series images that depict a respective particular portion of the geographical region during the time period, and predicting presence of a crop at particular locations within the particular portion of the geographical region associated with an image set of the plurality of image sets. Determining crop boundary locations within the particular portion of the geographical region based on the predicted presence of the crop at the particular locations, and generating a crop indicative image comprising at least one image of the multi-spectral and time series images of the image set overlaid with indication of crop areas, wherein the crop areas are defined by the determined crop boundary locations.

Abstract: A method embodiment commences upon identifying an on-premises storage area such as a file or virtual disk that is composed of a set of data blocks. A history of data access patterns over the data blocks is determined by observing accesses to the data blocks over time. When offsite storage is demanded (e.g., such as to a cloud-based storage facility), one or more storage objects are formed such that the blocks in a first one of the one or more storage objects comprise the blocks that have been measured to have the highest access rates (e.g., most recently accessed or most frequently accessed). Metadata that serves as a map is stored with the storage objects. A process or thread can access the storage objects in expectation that the most recently accessed or most frequently accessed blocks will be available in the first ones of the one or more storage objects.

Abstract: An embodiment may include a display processor, memory to store a 2D frame corresponding to a projection from a 360 video, and a quality selector to select a quality factor for a block of the 2D frame based on quality information from neighboring blocks of the 2D frame, including blocks which are neighboring only in the 360 video space. The system may also include a range adjuster to adjust a search range for the 2D frame based on a search area of the 2D frame, a viewport manager to determine if a request for a viewport of the 2D frame extends beyond a first edge of the 2D frame and to fill the requested viewport with wrap-around image information, and/or a motion estimator to estimate motion information based on both color information and depth information. Other embodiments are disclosed and claimed.

Abstract: An apparatus and method for improving the quality of user experience and reducing bandwidth consumption in Virtual Reality systems that optimize bandwidth by restricting the transmitted stream to only what is in the field of view of a virtual reality headset are disclosed. The actual content streamed may vary from interactive games to pre-recorded or live video.

Abstract: Methods and apparatus are provided for adaptive filtering of prediction pixels for chroma components in video encoding and decoding. An apparatus includes a video encoder for encoding picture data for at least a portion of a picture. The video encoder includes an adaptive filter for performing filtering of prediction data used to generate an intra prediction for a chroma component of the portion of the picture. The filtering is adaptive with respect to at least one of filter parameters for the filtering and whether the filtering is used or bypassed.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media directed to operations to facilitate generation of time-lapse videos. In accordance with embodiments described herein, frames of a photographic input are analyzed to detect activity occurring across frame pairs. The photographic input, such as video input, is input for which a time-lapse video is to be generated. Activity detected across frame pairs is used to automatically select a plurality of the frames for use in generating the time-lapse video. At least a portion of the frames selected in accordance with the activity detected across frame pairs is used to generate the time-lapse video.

Abstract: Embodiments of the present disclosure provide a method for selecting a public land mobile network PLMN, and an apparatus and a device for same. The method includes: during a process of selecting a network by a mobile terminal, if it is learned by means of determining that a registered public land mobile network RPLMN and a home public land mobile network HPLMN correspond to the mobile terminal are unavailable, acquiring a PLMN network in a PLMN list pre-stored in the mobile terminal, to perform a registration, wherein the PLMN list includes network information of one or more other PLMN networks that are selected and successfully registered.

Abstract: Method and system for adaptive image compression using block characteristics are provided. For adaptive image compression, blocks in an image are quantized and compressed using a quantization table that was used in quantization of the blocks. An image compression system checks a block-characteristics quantization table corresponding to a specific block in the image to be compressed in order to quantize the specific block, and performs calculation using a value of a first element of the quantization table corresponding to a specific element of the specific block and a value of a second element of the checked block-characteristics quantization table corresponding to the specific element in order to quantize the specific element.

Abstract: Method and apparatus for transmitting and receiving a signal in a broadcasting communication system are provided. Data of an input stream is divided for transmission through a frame. A header that includes information about the input stream is constructed. The frame is constructed using the header and the divided data.

Abstract: A method and system to index and search item listings in a network-based publication system, such as a network-based marketplace. For example, a listing query may be received from a user, the listing query including a query image. The query image may then be processed to generate a plurality of index sets. Each index set of the plurality of index sets may correspond to a respective portion of the query image and may comprise a set of index values. Each index value of the set of index values may be indicative of an image attribute, such as light intensity, for a respective sub-portion of the corresponding portion of the query image. At least one result image from a database of listing images may be identified by comparing the plurality of index sets for the query image to pre-compiled index sets in an index database. Each pre-compiled index set may be associated with at least one listing image which, in turn, may be associated with an item listing on a network-based marketplace.

Abstract: This disclosure describes techniques for coding transform coefficients associated with a block of residual video data in a video coding process. Aspects of this disclosure include the selection of a scan order for both significance map coding and level coding, as well as the selection of contexts for entropy coding consistent with the selected scan order. This disclosure proposes a harmonization of the scan order to code both the significance map of the transform coefficients as well as to code the levels of the transform coefficient. It is proposed that the scan order for the significance map should be in the inverse direction (i.e., from the higher frequencies to the lower frequencies). This disclosure also proposes that transform coefficients be scanned in sub-sets as opposed to fixed sub-blocks. In particular, transform coefficients are scanned in a sub-set consisting of a number of consecutive coefficients according to the scan order.

Abstract: The present invention relates to a method and an apparatus for encoding/decoding a video using a split layer. The video encoding/decoding method generates an encoded image data by encoding a current block partitioned into a plurality of subblocks, generates an encoded partition information data by encoding partition information of the current block, generates a bitstream including the encoded image data and the encoded partition information data, and then reconstructs the video image from the generated bitstream. According to the present disclosure, when encoding a high resolution video with variable sized blocks, various block shapes may be used for the encoding, and efficient encoding and decoding of the block partition information can improve the compression efficiency.

Abstract: Embodiments of the present invention provide a method and an apparatus for acquiring a weight coefficient of a digital filter so as to enhance a processing effect of images or videos and reduce the complexity of operations. The method includes: extracting global block features and local block features of image blocks, where the image blocks include a first image block and a second image block; acquiring an image block distance fs between the first image block and the second image block according to a global block feature and local block features of the first image block as well as a global block feature and local block features of the second image block; and evaluating a value of e fs/?2.

Abstract: A scoreboard for a video processor may keep track of only dispatched threads which have not yet completed execution. A first thread may itself snoop for execution of a second thread that must be executed before the first thread's execution. Thread execution may be freely reordered, subject only to the rule that a second thread, whose execution is dependent on execution of a first thread, can only be executed after the first thread.

Abstract: A moving picture coding apparatus includes an intra-inter prediction unit which calculates a second motion vector by performing a scaling process on a first motion vector of a temporally neighboring corresponding block, when selectively adding, to a list, a motion vector of each of one or more corresponding blocks each of which is either a block included in a current picture to be coded and spatially neighboring a current block to be coded or a block included in a picture other than the current picture and temporally neighboring the current block, determines whether the second motion vector has a magnitude that is within a predetermined magnitude or not within the predetermined magnitude, and adds the second motion vector to the list when the intra-inter prediction unit determines that the second motion vector has a magnitude that is within the predetermined magnitude range.

Abstract: Apparatus and methods of using content information for encoding multimedia data are described. A method of processing multimedia data includes obtaining content information of multimedia data, and encoding the multimedia data so as to align a data boundary with a frame boundary in a time domain, wherein said encoding is based on the content information. In another aspect, a method of processing multimedia data includes obtaining a content classification of the multimedia data, and encoding blocks in the multimedia data as intra-coded blocks or inter-coded blocks based on the content classification to increase the error resilience of the encoded multimedia data. Apparatus that can process multimedia data described in these methods are also disclosed.

Abstract: A moving picture coding apparatus includes an intra-inter prediction unit which calculates a second motion vector by performing a scaling process on a first motion vector of a temporally neighboring corresponding block, when selectively adding, to a list, a motion vector of each of one or more corresponding blocks each of which is either a block included in a current picture to be coded and spatially neighboring a current block to be coded or a block included in a picture other than the current picture and temporally neighboring the current block, determines whether the second motion vector has a magnitude that is within a predetermined magnitude or not within the predetermined magnitude, and adds the second motion vector to the list when the intra-inter prediction unit determines that the second motion vector has a magnitude that is within the predetermined magnitude range.

Abstract: A deblocking filter includes a color resolution expansion unit, an edge filtering unit and a color resolution contraction unit. The color resolution expansion unit increases a color resolution of input video data to generate first video data. The edge filtering unit performs a blurring operation on a block edge of the first video data based on block size information to generate second video data. The color resolution contraction unit reduces a color resolution of the second video data to generate output video data, a color resolution of the output video data being the same as the color resolution of the input video data.

Abstract: Methods and apparatuses to encode multimedia data enabling, among others, for example, improved decoded video quality, improved error recovery capabilities and/or improved decoding efficiency are described. A method according to the application includes one or more of the following: initializing a pixel-level reference counter for a current frame, performing a prediction of a next frame referring to the current frame, incrementing the pixel-level reference counter for each pixel of the current frame that is referenced during the prediction of the next frame, readjusting a macroblock bandwidth map of the current frame, repeating the steps of initializing, performing and readjusting for each next frame in a plurality of next frames referring to the current frame, and encoding the current frame based at least in part on the readjusted macroblock bandwidth map of the current frame.

Abstract: Image data is divided into blocks, each block has M×N pixels, the number of colors of each block undergoes reduction from M×N to each number of one to (M×N?1). Differences are acquired between pixel values for each block after color reduction, and pixel values for each block before first color reduction. A first table in which thresholds are associated with either the amount of data to be reduced or the amount of data to be remained, is acquired based on the acquired difference. A threshold at which the amount of data after compression drops to or below a target data amount is specified, based on the first table, and compressed data is generated based on image data after second color reduction using the specified threshold.

Abstract: According to some embodiments, an external memory unit stores rows of macroblocks representing an image. A set of macroblocks may be transferred from the external memory unit to a local buffer, the set of macroblocks including fewer macroblocks than a row. A macroblock may then be decoded based on information in the local buffer to generate an image signal.

Abstract: A method for encoding pictures within a groups of pictures using prediction, where a first reference picture from a group of pictures and a second reference pictures from the subsequent group of pictures are used in predicting pictures in the group of pictures associated with the first reference picture. A plurality of anchor pictures in the group of pictures associated with the first reference picture may be predicted using both the first and second reference pictures to ensure a smooth transition between different groups of pictures within a video frame.

Abstract: A method for identifying state of macro block of de-interlacing computing and an image processing apparatus are provided, the method is as follows. A video frame is divided into a plurality of regions, where each of the regions includes a plurality of macro blocks. Then, a basic threshold corresponding to each of the regions is provided according to a position of each of the regions in the video frame, and a first macro block is identified to be a first type macro block or a second type macro block according to the basic threshold corresponding to one of the regions where the first macro block of the macro blocks locates. Then, a corresponding de-interlacing computing step is performed on the first macro block according to an result that the first macro block is identified as the first type macro block or the second type macro block.

Abstract: In one embodiment, a method that provides plural representations of a single video signal that comprises a successive sequence of pictures, one or more of the plural representations including a respective sequence of latticed pictures, each latticed picture in the one or more plural representations originating from a corresponding respective picture of the video signal, the order of successive latticed pictures in the one or more of the plural representations of the video signal corresponding to the order of successive pictures in the video signal; processes the plural representations based on a predetermined encoding strategy, the predetermined encoding strategy targeting an appropriate respective amount of bits to each of a plurality of the processed latticed pictures, each of the plurality of the processed latticed pictures having a respective picture importance; and provides the plurality of processed latticed pictures in plural successive, non-overlapping, ordered segments in a single video stream.

Abstract: An encoder is operable to encode input data to generate corresponding encoded output data. The encoder includes data processing hardware. The encoder compresses content associated with blocks or packets, so that the encoded output data is smaller in size than the input data.

Abstract: A method and apparatus are provided for decoding a signal representative of an image sequence. The method includes extracting motion tubes from the signal, each of the tubes being defined by at least the following information cues: a block of reference pixels comprising reference texture information cues; start-of-tube and end-of-tube instants; and transition information cues. The method further includes processing the transition information cues and combining the information cues associated with the current blocks of at least two of the tubes overlapping in an overlap zone of the current image.

Abstract: A 3D graphics rendering pipeline is used to carry out data comparisons for motion estimation in video data encoding. Video data for the pixel block of the video frame currently being encoded is loaded into the output buffers of the rendering pipeline. The video data for the comparison pixel blocks from the reference video frame is stored as texture map values in the texture cache of the rendering pipeline. Once the sets of pixel data for comparison have been stored, the rendering pipeline is controlled to render a primitive having fragment positions and texture coordinates corresponding to the data values that it is desired to compare. As each fragment is rendered, the stored and rendered fragment data is compared by fragment compare unit and the determined differences in the data values are accumulated in an error term register.

Abstract: A video signal processing apparatus including: a video signal band measurement unit that measures a signal band of an input video signal; a band level evaluation unit that evaluates a band level of the resultant signal band measured by the video signal band measurement unit; a block size detection sensitivity adjustment unit that adjusts block size detection sensitivities in block size detection based on the band level evaluated by the band level evaluation unit; and a block size detection unit that detects the block sizes of blocks in the input video signal by using the block size detection sensitivities adjusted by the block size detection sensitivity adjustment unit.

Abstract: A flag parameter in a digital image decoding is calculated. For a macroblock consisting of M×N blocks, a first operation is performed on M block along a first edge to obtain M first parameters, and a second operation is performed on N blocks along a second edge to obtain N second parameters. The first and second parameters are stored into corresponding locations in a first and a second buffer array. Then a flag parameter corresponding to a given block is calculated according to corresponding values stored in the first and second buffer arrays. Calculation for all of the M×N blocks is performed in the order that neighboring left and upper blocks next to the give block is processed prior to the given block.

Abstract: The disclosure is directed to techniques for region-of-interest (ROI) video processing based on low-complexity automatic ROI detection within video frames of video sequences. The low-complexity automatic ROI detection may be based on characteristics of video sensors within video communication devices. In other cases, the low-complexity automatic ROI detection may be based on motion information for a video frame and a different video frame of the video sequence. The disclosed techniques include a video processing technique capable of tuning and enhancing video sensor calibration, camera processing, ROI detection, and ROI video processing within a video communication device based on characteristics of a specific video sensor. The disclosed techniques also include a sensor-based ROI detection technique that uses video sensor statistics and camera processing side-information to improve ROI detection accuracy.

Abstract: Methods and apparatus for implementing a reduced cost HDTV/SDTV video decoder are disclosed. The described joint video decoder is capable of decoding HDTV pictures at approximately the resolution of standard definition television pictures and can be used to decode HDTV and/or SDTV pictures. The described video decoder may be used as part of a picture-in-picture decoder circuit for providing picture-in-picture capability without providing multiple full resolution video decoders. The reduction in decoder circuit complexity is achieved through the use of a plurality of data reduction techniques including the use of a preparser, downsampling, and truncating pixel values.

Abstract: A 3D graphics rendering pipeline is used to carry out data comparisons for motion estimation in video data encoding. Video data for the pixel block of the video frame currently being encoded is loaded into the output buffers of the rendering pipeline. The video data for the comparison pixel blocks from the reference video frame is stored as texture map values in the texture cache of the rendering pipeline. Once the sets of pixel data for comparison have been stored, the rendering pipeline is controlled to render a primitive having fragment positions and texture coordinates corresponding to the data values that it is desired to compare. As each fragment is rendered, the stored and rendered fragment data is compared by fragment compare unit and the determined differences in the data values are accumulated in an error term register.

Abstract: Edge information of an input digital image is obtained. Digital image data corresponding to the input digital image is transmitted to a receiver. The edge information is also transmitted to the receiver. An output digital image is obtained at the receiver by processing the digital image data received at the receiver using the edge information received at the receiver. In this way, because the edge information is transmitted separately, as side information, and does not have to be obtained at the receiver, the computational complexity requires of the receiver is low. This enables for example real-time edge-adaptive interpolation to be carried out at the receiver.

Abstract: According to some embodiments, encoded information associated with an image is received at a decoder. The encoded information may be decoded at the decoder to generate full-sized first image pixels representing a full-sized version of the image. Moreover, the full-sized pixels may be scaled at the decoder to generate scaled image pixels representing a scaled version of the image.

Abstract: A method and apparatus estimates the strength of block artifacts in compressed video is described. Block artifacts are associated with the block-based image/video compression schemes. The block artifacts deteriorate the quality of compressed image and video, especially at low bit rates. A deblocking method measures the strength of the block artifacts at each block boundary and adjusts the deblocking parameters accordingly to improve the performance of the overall deblocking process. A method and device to measure the strength of block artifacts based on the global and local edge information of the current picture is provided.

Abstract: Video filtering using a programmable graphics processor is described. The programmable graphics processor may be programmed to complete a plurality of video filtering operations in a single pass through a fragment-processing pipeline within the programmable graphics processor. Video filtering functions such as deinterlacing, chroma up-sampling, scaling, and deblocking may be performed by the fragment-processing pipeline. The fragment-processing pipeline may be programmed to perform motion adaptive deinterlacing, wherein a spatially variant filter determines, on a pixel basis, whether a “bob”, a “blend”, or a “weave” operation should be used to process an interlaced image.

Abstract: An image processing device comprising a process generation section and a plurality of series-connected operation processing units. Each of the series-connected operation processing units receives a process packet output from the process generation section and performs any processing according to an instruction contained in the process packet. The units are divided into three suites and route selection sections are respectively inserted to input side of each of the suites. If the unit which executes a process related to an input process packet is not included in the immediately following one of the suites, the respective route selection sections supply this corresponding input process packet not to the input side of that one of the suites but to the output side of that suite. The process packet moves as bypassing such a suite as not to have the unit that executes a process related to this process data, thereby reducing its processing time and its power dissipation.

Abstract: In order to decode and display a previously compressed digital image portion, a first portion of this image being previously decoded and displayed in a first display window:—a request of a user defining a direction of movement in the image is read (115);—a new display window in the image is determined (119) as a function of this movement;—at least one area to be decoded and a decoding direction is determined (121) from the relative position of the new display window with respect to the first display window and this area is decoded and displayed (125, 127, 129) according to the decoding direction determined.

Abstract: An apparatus and method for coding block boundary detection using interpolated autocorrelation is disclosed. Input video frames are processed to obtain a binary map of possible block boundary pixels including original and rescaled block boundaries. The boundary pixel map is converted to a boundary histogram. Then interpolated autocorrelation is applied to the boundary histogram to detect the block size and block size offset. The interpolated autocorrelation handles both integer sizes and non-integer sizes. A confidence measure is also determined that reflects the accuracy of the detection result. The output for the current video frame is also post processed with information from previous frames to provide better temporal stability.

Abstract: A method for high quality, fast intra coding usable for creating digital video content. The video image is defined by data points representing original colors. Data points are selected based on a color characteristic, and are arranged into groups. At an initial stage (and at subsequent stages): selecting a subset of data points, using a number of algorithms to generate predicted values of the colors represented by the data points, calculating error values and evaluating mathematical functions to produce cost values, identifying a minimum cost value, establishing a threshold based on the minimum cost and a selection factor, comparing each cost value to the threshold value, and identifying a number of probable best algorithms. One of the probable best algorithms is usable for block size selection and to generate final predicted values of the original color represented by the data points.

Abstract: A method for removing blocking artifacts from moving and still pictures, comprising classifying horizontal and vertical boundaries in each picture as blocky or non-blocky; for each blocky boundary, defining an adaptive, picture contentdependent, one-dimensional filtered pixels region of interest (ROI) that crosses the boundary and is bound at each of its ends by a bounding pixel; defining a finite filter having a length correlated with the length of the ROI; defining a filtering pixels expansion that uniquely determines the padding values of the finite length filtered pixels ROI for the finite length filtering; and filtering the ROI pixels using the finite filter and the filtering expansion.

Abstract: A fast motion estimation system that determines a reference macroblock or sub-block combination within a reference frame for a current macroblock in a current frame includes a memory, a reference macroblock search circuit, a sub-block combination search circuit, and a comparator circuit. The reference macroblock search circuit determines a motion vector, multiple difference values, and a cost value for each macroblock within the reference frame according to a fast motion estimation search pattern, and stores the motion vector and the difference values in the memory. The sub-block combination search circuit searches the motion vector and the difference values in the memory for determining a corresponding one of multiple lowest cost sub-block combinations for each of multiple sub-block motion modes. The comparator circuit determines a lowest cost macroblock and selects from among the lowest cost macroblock and the lowest cost sub-block combinations to determine the reference macroblock.