CAJ | 학술논문

本文主要研究了基于动量梯度下降算法的BP神经网络在彩色图像边缘检测中的应用，减少了样本训练的迭代次数，缩短了样本训练时间。同时开发了相应图形用户界面（GUI），实现了将本文的算法与传统边缘检测算法的检测效果对比，显示不同算法的图像边缘检测效果，以及保存处理完成的图片等功能。
본문주요연구료기우동량제도하강산법적BP신경망락재채색도상변연검측중적응용，감소료양본훈련적질대차수，축단료양본훈련시간。동시개발료상응도형용호계면（GUI），실현료장본문적산법여전통변연검측산법적검측효과대비，현시불동산법적도상변연검측효과，이급보존처리완성적도편등공능。
This article mainly studied to use BP neural network to detect the edge of a color image.Using improved learning algorithm to train the network, making samples decreased frequency of training and quickly completed the training steps.This article develop the corresponding graphical user interface(GUI). It can compare this edge detection algorithm and the conventional detection algorithm,show the results of different edge detection algorithm and preserve processed pictures.