多模态智能审讯技术的原理与实战化应用路径研究

Research on the Principle of Multimodal Intelligent Interrogation Technology and Its Practical Application Path

多模态智能审讯技术通过非接触式生理检测、面部表情与肢体动作分析、眼动检测等多源异构数据融合算法,实现审讯中对犯罪嫌疑人心理的智能化分析和评估以及谎言甄别。将犯罪嫌疑人心理分析和评估意见输入审讯专家系统与知识库中大量的专家知识和经验进行碰撞输出审讯策略建议,为侦查人员提供专家服务。研发多模态智能审讯技术需要在审讯室采集真实的数据,构建一个数据量较大的、可靠的数据库,并充分挖掘和利用不同模态数据,构建多模态智能审讯技术与审讯专家系统同步化的人机结合的审讯技术,为审讯助力。多模态智能审讯技术作为审讯技术的典型代表,改变了传统的以人力资源为主的经验模式,无疑给审讯带来了崭新的活力和深刻的变革。

Multimodal intelligent interrogation technology realizes intelligent analysis and evaluation of suspect's psychology and lie detection during interrogation through non-contact physiological detection,facial expression and limb movement analysis,eye movement detection and other multi-source heterogeneous data fusion algorithms.Input the psychological analysis and evaluation opinions of suspect into the interrogation expert system,collide with a large number of expert knowledge and experience in the knowledge base,and output interrogation strategy suggestions,so as to provide expert services for investigators.The development of multimodal intelligent interrogation technology requires collecting real data in the interrogation room,constructing a large and reliable database,and fully mining and utilizing different modal data to build a human-machine integration interrogation technology that synchronizes multimodal intelligent interrogation technology with interrogation expert systems,to assist in interrogation.As a typical representative of interrogation technology,multimodal intelligent interrogation technology has changed the traditional experience mode mainly based on human resources,undoubtedly bringing new vitality and profound changes to interrogation.