纳米分辨率显微视频的像素级数据挖掘研究

项目名称： 纳米分辨率显微视频的像素级数据挖掘研究

项目编号： No.61502273

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 其他

项目作者： 张睿

作者单位： 衢州学院

项目金额： 20万元

中文摘要： 通过视频研究纳米物质的动态特性是一种新兴的科学研究方法，而研究利用计算机发现纳米物质动力学模式的视频数据挖掘方法具有重要科学意义和应用前景。本项目针对纳米分辨率显微视频的内容特征和数据特点，围绕视频像素级数据挖掘的两个关键基础问题展开研究，主要研究内容及创新点如下：1）从对原子能级跃迁机理的类比研究入手，提出以视频亮度等级为像素值采样单元的亮度等级采样法，建立其物理模型并分析其运行机制；2）将亮度等级采样法与“样本即总体”的大数据统计方法相结合，提出像素点亮度跃迁统计法，对所有像素点样本在各个亮度等级间的量子跃迁概率进行统计，实现对视频隐含信息的有效提取。本项目的研究可望为观察和理解复杂视频数据提供一种新视角，揭示视频数据跟随视场场景变化的内在规律，建立一种适用于高清高内涵显微视频的像素级数据挖掘新方法，为发现研究微观物质运动模式提供新思路，有效推动超高分辨率显微成像领域的信息智能化发展。

中文关键词： 纳米成像；视频数据挖掘；量子跃迁；大数据统计

英文摘要： To study the dynamic characteristics of nanosubstances using video is an emerging research approach, and video data mining that can discover the dynamics pattern of nanosubstances with computers has great value for science and applications. According to the characteristics of nanoscopy videos, this project will focus on two essential issues of pixel-level video data mining. The main contents and major creative ideas of the project include: 1) by comparing the mechanism of atomic energy level transition, an intensity-level sampling method (ISM) in which the intensity levels are taken as sampling units is proposed, and its physical model will be established and the operating mechanism of the model will be analyzed; 2) by combining the ISM with the total population sampling method introduced by big data statistics, an intensity-transition statistical method is proposed, which could extract the hidden information of videos by calculating the probability of pixel quantum transition between intensity levels. Research on this project will lead to a range of promising benefits, such as offering a new prospective of complex video data observation and understanding, revealing the inherent rules that video data changes with scene variations in the field of view, and establishing a new pixel-level data mining method suitable for the high-definition and high-content nanoscopy videos, that can provide a new way of thinking for discover and study the motion patterns of microscopic substances, and effectively promote the development of intelligent information in the field of super-resolution microscopy imaging.

英文关键词： Nanoimaging;video data mining;quantum transition;data big statistics