基于智能组件的图像分析系统的研究及应用

The Research and Application of Image Analysis Systems Based on Intelligent Components

针对各种图像分析算法适用于不同场合致使用户难以抉择的问题,提出了一种改进的智能组件模型。该模型具有枚举、适应、传播和淘汰4个算子,通过多个循环周期的迭代计算,达到搜索出给定图像分析场合的最优算法的目的。在此基础上,开发了可快速定制各种图像分析应用程序的智能组件创建工具,并以铜带表面缺陷检测为例,给出了其定制方法。实验结果表明,经过多个周期的计算后,铜带表面缺陷的总体检测率明显得到提高,其方法优于任何单一算法的准确率。

In order to solve the user＇ s dilemma problem while applying algorithms to a specific occasion in image analysis systems as each of them has its own advantage in different case, an improved intelligent component model is propo sed. The model has four operators, namely, Enumerator, Adaptator, Eliminator and Propagator, so that the best algorithm for specific case can be achieved by multiple iterative computes. On this basis, a creator tool of intelligent component model is developed that a variety of image analysis can be quickly customized by it. As an example, the customized approach of the applichtion of copper surface defect detection is presented. The experimental results show that the overall detection accuracy of copper surface defects are significantly improved and are superior to any individual image analysis algorithm after calculating a plurality of cycles.