IMAGE PROCESSOR, IMAGE PROCESSING METHOD AND PROGRAM

• Main Authors: MIZUSAWA MASATAKA, ENOKI ATSUYA
• Format: Patent
• Language: English; Japanese
• Published: 11.10.2018
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

PROBLEM TO BE SOLVED: To provide an image processor, an image processing method and a program capable of calculating similarity of different images with high accuracy.SOLUTION: An image processor 100 is provided with: a data acceptance part 10; a feature quantity calculation part 12; an individual similarity calculation part 14; a weight setting part 16; and a total similarity calculation part 18. The feature quantity calculation part 12 calculates local feature point density and a local feature quantity, and calculates one or more kinds of feature quantities except a local feature point from image data on a query image input in the data acceptance part 10. The individual similarity calculation part 14 calculates individual similarity between the query image and each of a plurality of dictionary images for every feature quantity. The weight setting part 16 sets weight of each feature quantity based on the local feature point density of the query image. The total similarity calculation part 18 calculates the total similarity between the query image and each of the plurality of dictionary images based on the individual similarity for every feature quantity and the weight of each feature quantity.SELECTED DRAWING: Figure 1 【課題】異なる画像の類似度を高精度で算出できる、画像処理装置、画像処理方法及びプログラムを提供する。【解決手段】画像処理装置１００は、データ受付部１０、特徴量算出部１２、個別類似度算出部１４、重み設定部１６及び総合類似度算出部１８を備える。特徴量算出部１２は、データ受付部１０に入力されたクエリ画像の画像データから、局所特徴点密度及び局所特徴量を算出し、かつ局所特徴量以外の１種以上の特徴量を算出する。個別類似度算出部１４は、特徴量ごとに、クエリ画像と複数の辞書画像それぞれとの個別類似度を算出する。重み設定部１６は、クエリ画像の局所特徴点密度に基づいて各特徴量の重みを設定する。総合類似度算出部１８は、特徴量ごとの個別類似度と、各特徴量の重みとに基づいて、クエリ画像と複数の辞書画像それぞれとの総合類似度を算出する。【選択図】図１