ANOMALY DETECTION DEVICE AND METHOD USING NEURAL NETWORK, AND DEVICE AND METHOD FOR TRAINING NEURAL NETWORK

• Main Authors: YOO, Kwangsun, LEE, Jungi
• Format: Patent
• Language: English; French; Korean
• Published: 30.11.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

An anomaly detection device according to one embodiment may comprise: a receiver for receiving a hyperspectral image; and a processor for extracting, on the basis of a plurality of target partial autoencoders corresponding to a plurality of bands included in the hyperspectral image, a plurality of local features corresponding to the plurality of bands, extracting a global feature on the basis of the plurality of local features through an aggregate autoencoder, and detecting anomalies on the basis of the global feature. Un dispositif de détection d'anomalie selon un mode de réalisation peut comprendre : un récepteur destiné à recevoir une image hyperspectrale; et un processeur destiné à extraire, sur la base d'une pluralité d'autocodeurs partiels cibles correspondant à une pluralité de bandes comprises dans l'image hyperspectrale, une pluralité de caractéristiques locales correspondant à la pluralité de bandes, à extraire une caractéristique globale sur la base de la pluralité de caractéristiques locales par l'intermédiaire d'un autocodeur agrégé, ainsi qu'à détecter des anomalies sur la base de la caractéristique globale. 일 실시예에 따른 이상 탐지(anomaly detection) 장치는 초분광 이미지(hyperspectral image)를 수신하는 수신기, 및 상기 초분광 이미지에 포함된 복수의 대역(band)에 대응하는 복수의 타깃 부분 오토인코더(partial autoencoder)에 기초하여 상기 복수의 대역에 대응하는 복수의 지역 특징(local feature)을 추출하고, 집합 오토인코더(aggregate autoencoder)를 통해 상기 복수의 지역 특징에 기초하여 전역 특징(global feature)을 추출하고, 상기 전역 특징에 기초하여 이상을 탐지하는 프로세서를 포함할 수 있다.