【作者】 任飞；

【导师】 秦志光；

【作者基本信息】 电子科技大学 ， 信息与通信工程， 2013， 博士

【摘要】 随着“三网融合”的推进和“宽带提速”的实施，智能电视成为继3D电视以来的行业新热点，智能电视软件平台成为继智能手机软件平台之后的又一研究热点，其中Android-Based和Web-Based的智能电视平台更受产业界的关注。然而，原生的Android平台不支持数字电视广播业务，其多媒体系统也不支持电视的应用特性，已成为基于Android研发智能电视平台的首要问题；另外，随着智能电视的内容与应用日渐丰富，过载的信息使人机交互过余复杂，海量内容与用户个性化需求难以匹配，以及开放平台引起的安全隐患，已经成为智能电视发展中的突出问题。本文针对上述问题，以四川省战略性新兴产业关键技术产业化重大专项“面向三网融合的新型数字家庭的智能终端设备研发及产业化”项目为依托，主要研究了：TV与多媒体的融合技术、基于自然语音的智能电视Web交互技术、基于视频图像对象识别的上下文感知关键技术、智能电视的系统安全模型与技术架构。本文的主要工作及创新点如下：1.提出了智能电视的三网融合终端多媒体架构模型。基于对原生Android的多媒体框架和数字电视的广播框架深入分析，以及基于对TV的多媒体特性的分析，提出了一种多媒体与TV的终端多媒体架构模型，基于这个架构的TV中间件和电视多媒体引擎方案，以及相应的资源管理方案，实现了广播与互联网的功能融合。2.提出了全语音交互的Web多业务架构模型。基于智能电视对Web应用(WebApp）承载的需求，提出了融合的智能电视Web应用架构，利用JavaScript扩展机制，对数字电视等TV本地应用，以及语音交互能力进行了Web扩展，将基于语音的自然人机交互技术整合到Web框架中，使智能电视成为一个支持全语音交互的Web多业务终端。3.提出了用户行为上下文感知总体架构，以及基于视频图像对象识别的上下文感知模型及识别算法。为感知电视用户行为，向用户提供个性化服务，研究并设计了智能电视终端与云端协同的用户上下文感知总体架构，通过对视频图像中特定对象识别的理论建模，设计了动态背景的图像对象轮廓的训练及识别算法。实验表明该模型及算法具有较高的对象识别能力，为基于用户上下文的个性化服务奠定了良好的基础。4.提出了智能电视软件平台的安全架构模型。针对Android的系统安全问题，研究了Android-Based智能电视平台的安全架构，提出了智能电视软件平台安全模型和实现架构。基于所提出的安全模型和实现架构，重点研究了系统安全加载、端到端的应用安全保障、视频内容防篡改、CA防破解等技术方案，相关技术成果已被视像电子行业协会智能电视标准组等行业标准作为优选技术实现方案采纳。更多还原

【Abstract】 With the implementation of the "Tri-network convergence" and "promotingbroadband speed" projects, the “smart TV” has become the most popular keywords after3DTV in TV industry. Moreover, smart TV platform becomes one hot research topicafter the smart phone platform. In all research of smart TV software platform, theAndroid-Based and Web-Based smart TV platform attract more attention. However,the original android system cannot support digital television broadcasting service, andits multimedia system cannot support the special requirements of television. These havebecome the primary problem of android-based smart TV platform development. Inaddition, along with the content becoming richer and the emergence of tremendoussmart TV applications, overloaded information makes the human-computer interactionbecome more complex, and it is more difficult to match the massive content with theindividual needs. With the smart TV platform being opened, security issues becomeanother focus of concern.Aiming at these problems, based on the industrialization of key technologies ofSichuan strategic emerging industries major project for “Development andindustrialization for new digital home smart terminals of Tri-network convergence”, themain research of this thesis includes: TV and multimedia convergence technology,voice-based interactive Web technology, context-aware key technology based on thevideo image object recognition, smart TV system security model and technicalarchitecture.The main works and innovations of this thesis are as follows:1. Propose a multimedia terminal architecture model for smart TV. Based on thethorough analysis of the original Android multimedia framework and the digital TVbroadcasting framework, along with the multimedia characteristics of TV, this thesisproposes a terminal architecture model which integrated TV and multimedia, and aseries of solutions for TV middleware, multimedia engine and corresponding resourcemanagement. This thesis implements functional converging between broadcasting andinternet. 2. Propose a web multi-service architecture model based on voice interactive.Aiming at web applications hosting needs, this thesis proposes the smart TV webarchitecture, and extends the TV native applications and voice interaction capabilitiesinto web architecture by JavaScript extensions mechanism. The web architecture makesthe smart TV become a full voice-interactive web multi-service terminal.3. Propose the overall architecture of context-aware based on user behavior, as wellas context-aware model and algorithms for video-based image object recognition. Inorder to perceive the user behavior and to provide personalized services, this thesisresearches and designs a kind of smart TV context-aware architecture based oncollaborating between terminal and cloud-service. Moreover, according to thetheoretical modeling on the specialty image recognition, an object’s outline training andrecognition algorithms of the dynamic background are designed. Experiments show thatthe model and the object recognition algorithm reserve higher capacity. This researchlays a good foundation for the future personalized service study in terms of userbehavior context.4. Propose a security architecture model for smart TV platform. In order to solvethe system security concerns in the open platform, the security architecture for theandroid-based smart TV platform is desired. Thus, a complete security model for smartTV platform is submitted. Moreover, based on the security model and achieved schema,more emphasis is given on the system security loading, end to end security guaranteeapplication, anti-tampered video content and anti-cracked CA. The related technologyresearch is adopted as an optimization technology solution by smart TV standard groupof the CVIA (CHINA VIDEO INDUSTRY ASSOCIATION).更多还原