FPGA Based Compact and Efficient Full Image Buffering for Neighborhood Operations

Image processing systems based on neighborhood operations i.e. Neighborhood Processing Systems (NPSs) are computationally expensive and memory intensive. Field Programmable Gate Array (FPGA) based parallel processing architectures accelerate calculations of NPS provided if they have fast external-memory data access by using on-chip data buffers. The conventional data buffers namely full Row Buffers (RBs) implemented with FPGA embedded memory resources i.e. Block RAMs (BRAMs) are resource inefficient. It makes overall NPS implementation on FPGA expensive and infeasible especially for resource-constraint environment. This paper presents compact and efficient image buffering architecture with an additional feature of pre-fetching. Proposed design fits in minimal BRAMs by using small yet efficient Main Control Unit (MCU). Its optimal multi-rated BRAM data accessing technique reduces BRAM cost to provide multiple pixels of pre-fetched data/clock to NPS in a fixed pattern. It controls and synchronizes BRAMs operations to attain throughput of 1 clock/pixel. Thus our buffer architecture with 66% reduction in BRAM requirement as compared to conventional RBs is capable to support buffering for real time systems with high resolution (1080x1920@62fps). Therefore proposed buffer architecture can suitably replace conventional RB in any real me NPS application.