欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
通过提取芯片图像中暗点、边缘、块数、面积和亮点等5种与芯片位置无关而且相互独立、易于提取的图像特征,建立正态分布模型,利用新识别出的损坏或缺陷芯片自动修正模型参数,提高模型的准确度.基于最小风险贝叶斯模式识别构造出各种损坏和缺陷芯片的分类器,对污损、烧蚀、碎裂和电极缺失芯片的正确识别率可以达到90%以上.
英文摘要：
Based on five kinds of image feature such as dark pixel number,edge pixel number,blocks number,area and bright pixel number which are extracted from chip image irrelevant to chip position,the normal distribution model is established.Newly identified auto-modified model parameters for damaged or defective chips are employed to promote its accuracy.The classifier for various damaged and defective chips based on the minimum risk Bayesian classifier is established to identify stained,ablated,disintegrated and electrode-lost chips with accuracy higher than 90%.
同期刊论文项目
同项目期刊论文
An optimum design model for textured surface with elliptical-shape dimples under hydrodynamic lubric
A novel method to identify the scaling region for chaotic time series correlation dimension calculat
INFLUENCE OF SAMPLING LENGTH AND SAMPLING INTERVAL ON CALCULATING THE FRACTAL DIMENSION OF CHAOTIC A
期刊信息
位置：