发明名称 METHOD AND SYSTEM FOR IMAGE ANALYSIS
摘要 A method for processing an image includes: identifying a group of keypoints in the image; for each keypoint of the group; a) calculating a corresponding descriptor array including a plurality of array elements, each array element storing values taken by a corresponding color gradient histogram of a respective sub-region of the image in the neighborhood of the keypoint; b) generating at least one compressed descriptor array by compressing at least one portion of the descriptor array by vector quantization using a codebook including a plurality of codewords.
申请公布号 US2015104110(A1) 申请公布日期 2015.04.16
申请号 US201314396525 申请日期 2013.04.19
申请人 TELECOM ITALIA S.p. A. 发明人 Balestri Massimo;Francini Gianluca;Lepsoy Skjalg
分类号 H04N19/124 主分类号 H04N19/124
代理机构 代理人
主权项 1. A method for processing an image, comprising: identifying a group of keypoints in the image; for each keypoint of the group a) calculating a corresponding descriptor array including a plurality of array elements, each array element storing values taken by a corresponding color gradient histogram of a respective sub-region of the image in the neighborhood of the keypoint;b) generating at least one compressed descriptor array by compressing at least one portion of the descriptor array by means of vector quantization using a codebook comprising a plurality of codewords; exploiting said at least one compressed descriptor array of the keypoints of said group for analysing the image, wherein the method further comprises: compressing the codebook, said compressing the codebook including: generating a codebook matrix, each row of the codebook matrix being a codeword of the codebook;factorising the codebook matrix so as to obtain the product of at least a first matrix and a second matrix, the energy of the items of the second matrix generally non-increasing as the column indexes of such matrix increase,truncating the first matrix by removing therefrom a first number of last columns;truncating the second matrix by removing therefrom the first number of last columns and the first number of last rows;generating a first further matrix corresponding to the product of the truncated first matrix by the truncated second matrix;quantizing each item of the first further matrix, wherein each item belonging to a column of the first further matrix is quantized using a corresponding number of quantization levels that is lower than or equal to the number of quantization levels used to quantize the items belonging to a preceding column;generating a second further matrix wherein each item of the second further matrix corresponds to an item of the first further matrix, each item of said second further matrix being an index associated to the quantization level assumed by the corresponding quantized item of the first further matrix; storing the codebook by memorizing said indexes of the second further matrix in a memory unit.
地址 Milano IT
您可能感兴趣的专利