摘要
文章研究了图像处理的去除噪声、图像分割、图像增强等多种低层处理的方法,建立用区域增长法进行番茄表面缺陷区域检测,用BP算法训练的多层前馈神经网络对番茄的损伤进行分类。结果表明,番茄损伤检测和分类的准确率不低于90%。
Three methods that include filtering noise and dividing of image and highlighting of image etc. for lower-layer image processing technique were studied in the paper, and distriction increasing for detecting bruise image of tomato was built, and training multilayer feedforward neural networks with BP for detecting tomato bruise and classification was built. The experiment results showed that the rate of testing precision was not less than 90%.
出处
《东北农业大学学报》
CAS
CSCD
2006年第2期215-218,共4页
Journal of Northeast Agricultural University
关键词
计算机视觉
BP算法
番茄
损伤检测
computer vision
BP algorithm
tomato
detection of bruise
