Development of predictive safety system for child welfare facility based on behavior analysis by image technology

基于图像技术行为分析的儿童福利设施预测安全系统开发

• 批准号: 23601014
• 负责人: HANAZAWA Akitoshi
• 金额: $ 3.24万
• 依托单位: Kyushu Institute of Technology
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2011
• 资助国家: 日本
• 起止时间: 2011 至 2013
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-23601014/
• 关键词: 画像処理; 行動検出; 年齢識別; 人物検出; 安全システム; 画像認識; 見守り; 子供; セキュリティー

We developed basic elemental image technology and behavior detection technology, which are necessary for the construction of the system that detects children's dangerous behaviors. Followings were developed mainly. (1) human detection. (2) children's gait analysis. (3) human behavior detection by infra-red depth image sensor. (1) For human detection, we tried to improve accuracy and speed of detection. (2) For children's gait analysis, we investigated differences in gait pattern between children and adults when they walk over obstacles, and we found several features that can be used to distinguish children from adults. (3) For human behavior detection by infra-red depth image sensor, we developed technology to detect the position and posture of a human in a room by detecting the position of human head.

我们开发了基本的元素图像技术和行为检测技术，这是构建儿童危险行为检测系统所必需的。主要发展了以下几个方面。(1)人工检测。(2)幼儿步态分析。(3)红外深度图像传感器人体行为检测。(1)对于人的检测，我们试图提高检测的准确性和速度。(2)对于儿童的步态分析，我们研究了儿童和成人在跨越障碍物时步态模式的差异，发现了几个可以用来区分儿童和成人的特征。(3)对于红外深度图像传感器的人体行为检测，我们开发了通过检测人的头部位置来检测人在房间中的位置和姿势的技术。

项目成果

Introduction of Spatial Limitation to Human Detection by Co-occurrence with Shape and Color Self- Similarity Features

通过形状和颜色自相似特征的共现引入人体检测的空间限制

• 发表时间: 2012
• 作者: Hidenori Matsui;Akitoshi Hanazawa.
• 通讯作者: Akitoshi Hanazawa.

Smile Intensity Measurement for Dementia Elders Aiming at Estimation of Emotional Recovery by Learning Therapy

测量痴呆症长者的微笑强度，以评估学习疗法的情绪恢复情况

• 发表时间: 2013
• 作者: Shohei Iwamoto;Akitoshi Hanazawa;新小田春美・池田洋子・境トモエ・福本弘子;Akitoshi Hanazawa
• 通讯作者: Akitoshi Hanazawa

深度画像を用いた高齢者の危険行動検知

使用深度图像检测老年人的危险行为

• 发表时间: 2012
• 作者: 立場将太;花沢明俊
• 通讯作者: 花沢明俊

局所的色相関分布に対する勾配方向ヒストグラムによる人物検出

使用梯度方向直方图进行局部颜色相关分布的人物检测

• 发表时间: 2012
• 作者: 穴井 謙;西森 亜弥;新小田 春美;松村遼,花沢明俊
• 通讯作者: 松村遼,花沢明俊

Combinational Use of Two Different Normalization Methods Improves Accuracy of Human Detection Based on Shape-Based Features

两种不同归一化方法的组合使用提高了基于形状特征的人体检测的准确性

• 发表时间: 2013
• 作者: Shohei Iwamoto;Akitoshi Hanazawa.
• 通讯作者: Akitoshi Hanazawa.