一种新的储粮害虫图像边缘检测的元胞自动机法

THE APPLICATION OF CELLULAR AUTOMATON IN IDENTIFICATION OF STORED-GRAIN PEST IMAGE

采用元胞自动机进行虫体图像边缘检测,并与其他几种边缘检测算子的检测结果进行比较。同时以包括元胞自动机在内的6种边缘检测结果为基础,结合RBF神经网络算法和SVM支持向量机算法,对粮虫图像进行了识别。识别结果表明,基于元胞自动机边缘检测的粮虫识别正确率优于其它边缘检测算法。

The damage caused by pest to stored--grain is a main reason for grain postharvest losses. In real engi- neering application with the limitation of circumstance, unclear stored--grain pest images bring great diffi- culty to subsequent pest identification. In this paper, we used cellular automaton（CA） to do a border detec- tion for pest image, and compared detection effects with other many detection operators. At the same time, based on the six border detection methods including CA, We used Radial Basis Function （RBF） neural net- work and Support Vector Machine（SVM） to identify pest image. Experimental results show that, the cor- rect identification rate for stored--grain pest based on border detection using CA is superior to other border detection algorithms.