Abstract: The time and space computing complexity of the traditional Generalized Hough Transform(GHT) is too high for real-time application, so a GHT is proposed in this paper. The computing complexity is reduced in this modified Hough algorithm. The basic idea of this algorithm is combining multi-scale analysis and GHT. To verify this algorithm, chip detection in Automatic Optical Inspection(AOI) system is performed. Test results prove to be good.

Key words: Generalized Hough Transform(GHT), chip detection, object recognition

摘要： 传统的广义霍夫变换空间复杂度及时间复杂度都很高，不适用于实时的应用。针对该问题，提出一种基于广义霍夫变换的芯片检测算法，降低了计算复杂度。该算法的主要思想是将多尺度分析与广义霍夫变换相结合。将该算法应用到自动光学检测系统的芯片检测中，取得了较好的检测结果。

关键词: 广义霍夫变化, 芯片检测, 物体识别

CLC Number: 

• TP391.41