Programmable Hybrid Co-Processor for Real-Time Image Understanding

该小型企业创新研究(SBIR)项目将研究一种新型模块化图像处理架构的可行性。拟议的系统将被设计成与现有的微处理器体系结构兼容，以便在不久的将来进行商业产品开发和进入市场。与现有的图像处理方法相比，该系统的其他优点包括提高处理速度、片上立体声和运动对应计算、低功耗、减小尺寸(独立或MCM封装芯片)和成本。这项工作包括对分析基础的研究，以及对可以在硬件中实现的各种算法的实验和模拟。这一广泛的调查范围将导致更好地了解实际建造图像处理器所涉及的问题，并提高第二阶段原型开发期间的成功水平。

This Small Business Innovation Research (SBIR) Project will investigate the feasibility of a novel modular image processing architecture. The proposed system will be designed to be compatible with existing microprocessor architectures to allow for commercial product development and market entry in the near future. Additional benefits of the system over existing image processing methods are increased processing speed, on-chip stereo and motion correspondence computation, low power consumption, reduced size (stand-alone or MCM packaged chips) and cost. The work involves research in analytical foundations, as well as experimentation and simulation with a variety of algorithms that can be implemented in hardware. This broad scope of investigations will lead to a better understanding of the issues involved in the actual construction of an image processor and enhance the level of success during Phase II prototype development.