基于KL变换与BP神经网络的农田害虫识别算法

摘要

ABSTRACT:KL transform is a algorithm for reducing the dimension, and BP neural network is an intelligent recognition algorithm. The farmland pests training requires a huge pest image gallery. Facing the huge pest image gallery, the use of high-definition digital photos will be a great challenge for computer’s performance. Therefore, KL orthogonal transform is used to reduce the dimension and to farthest maintain the neural network’s recognition rate based on improving computer’s performance.%KL变换是一种降低维度的算法，BP神经网络是一种智能识别算法，农田害虫的训练需要一个庞大的害虫图片库，面对巨大的图片库，如果使用高清的数码照片训练，是对计算机性能的巨大挑战，所以将KL正交变换用于降低维度，最大限度地在提高计算机性能的基础上维持神经网络的识别率。