To identify pedestrian attributes with high accuracy using a simple device configuration.SOLUTION: An attribute identification device includes a cluster generation unit 32, which generates a cluster of radar signal points extracted by detecting the peak of the frequency spectrum of radar data, a representative point identifying unit 34, which identifies a representative point of the cluster, a region-of-interest setting unit 41, which sets a region of interest, a signal strength identification unit that calculates signal strength information for each relative velocity value of each of the radar signal points included in the region of interest, a correlation calculating unit 51, which arranges a set of combination data of the relative velocity values and the signal intensity information values for each radar scan of a predetermined length of time in a time series, and then calculates a correlation value in contrast to the set of the combination data arranged in a time series before the shift, while shifting the time series arrangement of the combination data one by one in the direction of the time axis, and a discriminating unit 6, which discriminates the attribute of an object using the correlation value.SELECTED DRAWING: Figure 1
【課題】簡便な機器構成によって歩行者の属性の識別を高精度に行うことを可能にする。【解決手段】レーダデータの周波数スペクトルのピークを検出して抽出されたレーダ信号点のクラスタを生成するクラスタ生成部32と、クラスタの代表点を特定する代表点特定部34と、関心領域を設定する関心領域設定部41と、関心領域に含まれる前記レーダ信号点それぞれの前記相対速度の値別に信号強度情報を計算する信号強度特定部と、所定の時間長さぶんのレーダスキャンごとの相対速度の値と信号強度情報の値との組み合わせデータの集合を時系列に配列した上で、前記組み合わせデータの時系列の配列を時間軸方向に1つずつシフトさせながら、シフト前の、時系列で配列された前記組み合わせデータの集合との対比における相関値を計算する相関計算部51と、相関値を用いて物標の属性を判別する識別部6と、を有する。【選択図】図1
ATTRIBUTE IDENTIFICATION DEVICE AND IDENTIFICATION METHOD
属性の識別装置および識別方法
2021-11-18
Patent
Electronic Resource
Japanese
Synthetic visible imagery for multi-attribute target identification
Tema Archive | 1990
|Vehicle fixed parking position and attribute identification system and method
European Patent Office | 2023
|PEDESTRIAN ATTRIBUTE IDENTIFICATION AND POSITIONING METHOD AND CONVOLUTIONAL NEURAL NETWORK SYSTEM
European Patent Office | 2020
|Pedestrian attribute identification and positioning method and convolutional neural network system
European Patent Office | 2023
|