| 摘要 |
Abstract KERNALIZED CONTEXTUAL FEATURE Disclosed is a method (200) of matching a first image and a second image comprising: (a) 5 segmenting (502) the first image into segments and the second image into segments ; (b) determining (503) a self-similarity matrix encoding similarity between the first segments, and a self-similarity matrix encoding similarity between the second segments; (c) generating (504) a reduced dimension representation of the first matrix, and a reduced dimension representation of the second matrix; (d) determining (505) a weighted feature correlation 10 pattern of the first reduced dimension representation, and a weighted feature correlation pattern of the second reduced dimension representation; (e) forming (206) a first image vector from the first weighted feature correlation pattern, and a second image vector from the second weighted feature correlation pattern; and (f) matching (206) the first image and the second image by determining a distance between the first image vector and the second image vector. P019408_specilodged /221211 201 Start ) 202 Set the retrieved object image as a query image uery images empt . En 3 No Retrieve job images Generate a Toaogetoo weighted feature correlation pattern s the count of jo Qu'ery images zero? image 303 No Compute the similarity matching score between the weighted feature correlation patterns of the job images and 206 query image Identify the job image with the highest similarity matching 207 Yes 214 score - - -208 215 ighest similarity scor 209Noless than a threshold? Yes Asig tac ienifer 216 Assign new track Assin tack denifie ~~identifier to query of the job image to Iimage query image IF 211 /N 1 Store query image PO19408 lodgedspeci / 5855381_1 Fig 222 |
