中文摘要：

基于D．R．Chialvo et al．提出的原始模型和V．Ramos et al．的扩展模型，提出了一种用于数字图像边缘检测的基于灰度梯度感知的人工蚁群模型（ACGGP）。模型利用灰度梯度启发信息和信息素轨迹信息共同来指导蚂蚁的行为，这种策略能够有效地减小群体规模进而提高算法的时间性能。此外，还对模型体现出类似于某些人类视觉感知的机制，包括群体短暂记忆、图像主要特征敏感特别是能够良好抑制噪声的自发侧抑制机制等特征进行了阐述。模型在若干幅8bit灰度图像上进行了测试，并和Ramos模型以及经典数字图像边缘检测算法（canny算法）进行了客观比较。

英文摘要：

A novel artificial ant colony model on digital image edge detection is presented base on the model which was first introduced by D.R.Chialvo et al.and later extend by V.Ramos et al.A heuristic information called grayscale grads perception is proposed and used to provide guidance to the ants with the pheromone trails together.This strategy can reduce colony size and hence improve the alogrithm＇s time performance.Some emergent properties of the model are also figured out and studied including spontaneous lateral inhibition which can inhibit noise efficiently and colony short memory as well as main feature sensitivity which are quite similar with some of human＇s visual perception mechanism. The algorithm is demonstrated by applying it into the edge detection of several 8 bit grayscale images and compared objectively with Ramos＇s model and a classical image edge detection algorithm（canny algorithm）.