Abstract: Recurrent Neural Networks（RNN） is a kind of important artificial neural networks with better ability for learning and rate of convergence in comparison with forward neural networks，however，the design of the structure of the hidden-layer is a difficult problem all the time.Echo State Networks has no such problems with special construction，good stability，short-cut learning process.Application of ESN to the edge detection of images has been introduced after the presentation of the structure and method of learning of ESN，resulting well.

Key words: Echo State Networks（ESN）, edge detection, statistical vector

摘要： 循环神经网络（RNN，也称反馈神经网络）是一种重要的人工神经网络，与前馈神经网络相比具有更好的学习能力和更快的收敛速度，但其隐层结构的设计一直是个难点问题。回声状态网络（ESN）有效地解决了上述问题，相比于以前的循环神经网络，其具有结构独特、稳定性好、学习过程简单快捷等特点。介绍了回声状态网络及其学习方法，将其用于图像的边缘检测中，取得了良好的效果。

关键词: 回声状态网络, 边界检测, 统计向量