Persistent feature descriptors for video

    公开(公告)号:US10534964B2

    公开(公告)日:2020-01-14

    申请号:US15419281

    申请日:2017-01-30

    Abstract: Methods and devices for extracting feature descriptors for a video, the video having a sequence of pictures. The method includes identifying a first key picture and a second key picture later in the sequence than the first key picture; extracting a first set of feature descriptors from the first key picture and a second set of feature descriptors from the second key picture; identifying a set of pairs of feature descriptors, where each pair includes one descriptor from the first set and one descriptor from the second set; generating motion information describing the motion field between the first key picture and the second key picture; and filtering the set of pairs of feature descriptors based on correlation with the motion information to produce and output a subset of persistent descriptors.

    Methods and devices for inter-prediction using motion vectors for video coding

    公开(公告)号:US10291928B2

    公开(公告)日:2019-05-14

    申请号:US15402769

    申请日:2017-01-10

    Inventor: Dake He

    Abstract: Methods and devices for video coding using inter-coding and removing redundant motion vectors from a set of candidate motion vectors. The methods may determine that at least some candidate motion vectors in a set of candidate motion vectors in a search range around the motion vector predictor identify candidate blocks in the previously-reconstructed reference picture that are redundant and, based on that determination, eliminate said at least some of the candidate motion vectors to reduce the set to a subset. In some cases, a candidate block is redundant if it is less than a threshold different from another of the candidate blocks identified by one of the candidate motion vectors in the set. A function may be applied to the candidate block and the another candidate block. If the output from the function in the two cases is the same, then the blocks are determined to be redundant.

    Communication with I-Q decoupled OFDM modulation

    公开(公告)号:US10212014B2

    公开(公告)日:2019-02-19

    申请号:US15342865

    申请日:2016-11-03

    Abstract: In some examples, an apparatus includes a transmitter configured to apply input data as a frequency-domain data sequence to modulate a set of subcarriers, and separate the frequency-domain data sequence into a first subsequence of elements and a second subsequence of elements. A first conjugate symmetric subsequence that is based on inserting a zero value into the first subsequence is formed, and a second conjugate symmetric subsequence that is based on inserting a zero value into the second subsequence is formed, where the zero values are inserted at different positions in the first and second subsequences. A time-domain sequence comprising a first component that is a function of the first conjugate symmetric subsequence, and a second component that is a function of the second conjugate symmetric subsequence is generated.

    Scaling in perceptual image and video coding

    公开(公告)号:US09699461B2

    公开(公告)日:2017-07-04

    申请号:US14826265

    申请日:2015-08-14

    Inventor: Dake He

    CPC classification number: H04N19/13 H04N19/126 H04N19/187 H04N19/94

    Abstract: Methods and devices for image and video coding. A block of N coefficients is defined as an input vector of N dimensions having a magnitude value and N−1 angle values. The encoder normalizes and quantizes the angle values to produce quantized angle values, reconstructs the angle values from the quantized angle values and determines a quantizer based on the reconstructed angle values. The determined quantizer is then used to quantize the magnitude value. The quantized angle values and the quantized magnitude value are entropy encoded. The decoder preforms the reverse process, and determines the quantizer for inverse quantizing the quantized magnitude value from the reconstructed angle values.

    Method and system for efficient feature matching

    公开(公告)号:US09600524B2

    公开(公告)日:2017-03-21

    申请号:US14579297

    申请日:2014-12-22

    Inventor: Dake He

    Abstract: A method of non-identical feature matching in a search system, the search system having a set of data points. The method includes determining a threshold error and expanding the set to create an expanded set by including all data points as data elements of the expanded set and, for every data point in the set, finding all data elements within an error distance of that data point and adding those found data elements to the expanded set, wherein the error distance is a function of the threshold error. The method includes creating a summary representation of the expanded set by constructing a binary array using a plurality of hash functions as a bloom filter. The method may further include receiving a query and applying the plurality of hash functions to the query to determine, from the binary array, whether the query matches one of the data elements in the expanded set and, if so, outputting a match.

Patent Agency Ranking