自动视觉检查（句法表示）系统涉及从规范、要检查的组件和图像结构的规则生成中获取信息。这些规则直接从计算机辅助设计（CAD）文件和操作员与图像的交互中获得，存储在一个知识库中，通用解析器使用该知识库来评估由VLSI电路元件的显微图像生成的图像原语集。解析器标记可能存在的缺陷，以供检查器验证。新规则可以随时添加到知识库中。该设备与放大和旋转无关。原型开发有望改善半导体制造工艺，缩短新工艺和技术的开发时间，降低成本，提高检测质量。

THE AUTOMATED VISUAL INSPECTION (SYNTACTIC REPRESENTATION) SYSTEMS INVOLVES ACQUISITION FROM PRODUCTION OF RULES ABOUT SPECIFICATIONS, COMPONENTS TO INSPECT, AND THE STRUCTURE OF THE IMAGE. THESE RULES, OBTAINED DIRECTLY FROM COMPUTER-AIDED DESIGN (CAD) FILES AND INTERACTION OF THE OPERATOR WITH THE IMAGE, ARE STORED IN A KNOWLEDGE BASE WHICH IS USED BY A GENERAL-PURPOSE PARSER TO EVALUATE SETS OF IMAGE PRIMITIVES PRODUCED FROM MICROSCOPIC IMAGES OF VLSI CIRCUIT COMPONENTS. THE PARSER FLAGS POSSIBLE DEFECTS TO BE VERIFIED BY THE INSPECTOR. NEW RULES CAN BE ADDED TO THE KNOWLEDGE BASE AT ANY TIME. THE FACILITY IS MAGNIFICATION AND ROTATION INDEPENDENT. IMPROVED SEMICONDUCTOR MANUFACTURING PROCESSES, SHORTENED DEVELOPMENT TIME FOR NEW PROCESSES AND TECHNIQUES, AND LOWER COST OF HIGHER INSPECTION QUALITY ARE EXPECTED TO RESULT FROM THE PROTOTYPE DEVELOPMENT.