基于SV-DPI的图像坏元修正FPGA自动化验证  

Automatic Verification of Field Programmable Gate Arrays for Dead Pixel Correction

作　　者：李艳龙 杨琪 王雪峰 LI Yanlong;YANG Qi;WANG Xuefeng(Xi’an Microelectronic Technique Institute,Xi’an 710065,China)

机构地区：[1]西安微电子技术研究所,陕西西安710065

出　　处：《红外技术》2020年第12期1192-1197,共6页Infrared Technology

摘　　要：为实现红外图像坏元修正FPGA(field programmable gate array)的快速验证,提高测试覆盖性,设计了基于SV-DPI(SystemVerilog-direct programming interface)的FPGA自动化验证平台。采用DPI(direct programming interface)编程接口技术,实现了SystemVerilog平台调用C++编程语言,构建了针对红外图像坏元数据的生成和检测修正模型,建立了两种语言在事务级(transaction level)模型的通信。结果表明相对于传统验证方法,该平台结构简单,可以快速实现激励产生、参考模型构建、测试结果自动比对等功能,实现了红外图像坏元检测与修正FPGA的自动化测试,功能覆盖率达到100%,有效缩短FPGA测试平台搭建和调试周期,提高了测试效率和测试质量。To accelerate the simulation speed and improve the coverage of verification for a field programmable gate array(FPGA)implemented with dead pixel correction of an infrared image,an FPGA automatic verification platform based on SystemVerilog-Direct programming interface(SV-DPI)was designed.Using DPI programming interface technology,the C++programming language was invoked by the SV platform.A generator and correction model for dead pixel data of infrared images was built.This established a communication between two languages on the transaction level.The results show that,compared with the traditional verification method,the proposed platform is simple in structure and can quickly generate a test vector,construct a reference model,and check results automatically.It realizes automated verification for an FPGA implemented with dead pixel detection and correction of an infrared image.The function coverage can reach 100%.It effectively shortens the period of construction and debugging for the FPGA verification platform and improves the efficiency and quality of verification.

关 键 词：自动化测试 FPGA验证 红外图像坏元修正 验证平台 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]