CORRELATION EXTRACTING METHOD FOR IMAGE FEATURE QUANTITY AND IMAGE RETRIEVAL DEVICE

摘要

PROBLEM TO BE SOLVED: To determine a rule prescribing the correlation between a physical feature and a sensuous feature mechanically with high reliability by extracting the correlation between both the features by applying multiple regression analysis algorithm. ;SOLUTION: A correlation extraction block 103 consists of a physical feature quantity extraction part 112, a questionnaire result input part 113, and a multiple regression analysis part 114. The physical feature quantity extraction part 112 is an image processor, which extracts various physical quantities by processing a sample image 111, and the questionnaire result input part 113 is a data entry device for inputting evaluated values (sensuous feature quantity), regarding various sensuous words implemented for the sample image 111. The multiple regression analysis part 114 finds a mapping rule from the physical feature quantities and sensuous feature quantities of the sample image 111 by applying the multiple regressive algorithm. This function is actualized by running an application program with program contents on a computer.;COPYRIGHT: (C)1999,JPO