Xilinx Zynq 7000 SoC Based Character Feature Extraction System for OCR Inspection of Integrated Circuit Chips

This paper designs a feature extraction system for OCR inspection of integrated circuit chips. In this system the following algorithms were implemented: Sobel edge detection and FAST corner detection. The latest Xilinx Zynq-7000 was used to integrate the main part of the system into a single SoC. Parallel arithmetic units and digital modules were implemented on the programmable logic (PL) of Zynq-7000 to decrease system size, while programs running on the processing system (PS) of Zynq-7000 controlled the system work flow and provided display interface using open-source software such as Linux and OpenCV. Meanwhile, industry standard advanced extendable interface (AXI)buses were adopted to encapsulating standardized IP cores and build high speed data exchange bridges between units within Zynq-7000. The experimental results indicate that modern SoC systems are a high-efficiency platform for advanced embedded image processing, analysis and recognition applications.