摘要
基于图像识别的国家储备粮仓袋装粮食数量自动监管与稽核系统的技术核心是智能识别各种粮仓场景图像中粮袋的数量,杜绝人为的弄虚作假。边缘检测是袋装粮图像识别的首要问题,在分析了经典的以及在其基础上进行各种改进的Laplace算子缺陷的基础上,提出了1种改进的Laplace算子,该算子通过设置合理的模板参数克服了原有算子的不足,提高了图像边缘检测的精度。实验结果证明,该算子检测效果优于其他模板,并且能够精确地检测出各种类型的边缘信息。
The technique core of national grain reserves packaged grain＇s number automatical supervision and audit system based on image recognition was intelligent recognition all kinds of granary scene images to eliminate man-made fraud.It was well known that the edge detection was a chiefly problem to identify the packaged grain.Based on the analysis the flaw of original Laplace operator and some improved Laplace operator,we proposed an improved Laplace operator.This Laplace operator overcame the flaw of original Laplace operator by means of setting reasonable template parameter;increased the precision of edge detection.The experiment results showed that this Laplace operator was better than other Laplace operators,and could precision detection various kinds edge information.
出处
《系统仿真技术》
2009年第4期255-257,271,共4页
System Simulation Technology
基金
重庆市财政局重点科技资助项目(2007)
