论文部分内容阅读
农作物在生长和发育过程中,经常受到各种因素的影响,导致农作物发生病害。准确、快速的处理与识别植物病害图像,并自动可靠的对其进行危害程度测定是进行农作物病害防治的关键。为此研究了利用分形理论对农作物病变叶片自然纹理图像进行处理,利用BP网络来实现自然纹理图像的分类问题。分析各种不同的特征在分类器中的性能,提出了一种纹理特征性能的比较方法。
Crops in the process of growth and development, are often affected by various factors, leading to crop disease. Accurate and rapid processing and identification of plant disease images, and automatically and reliably determine the degree of damage to the crop disease prevention and control is the key. This paper studies the use of fractal theory of crop diseased leaves natural texture image processing, the use of BP network to achieve the natural texture image classification problem. By analyzing the performance of different features in the classifier, a comparison method of texture feature performance is proposed.