鼠标点击的图形代替快速识别算法

鼠标图形是一种方便快捷的软件操作方法，本文首先对鼠标图形进行分析，提取具有平移和放缩不变性的特征，并使用这些特征作为多层前馈神经网络的输入来对鼠标图形进行识别.采用改进的RPROP算法对神经网络进行训练，实验结果表明，该算法能够识别复杂的鼠标图形，具有准确度高和抗干扰强，识别速度快的特点.

A fast graph substitute for mouse click recognition method

Mouse graph is a kind of convenient and adroit software operating method. First, after analyzed the mouse graph, get the features which are translation invariant and scale invariant. By using these features as the input of multilayer forward neural network, the mouse graph can be recognized. The authors train the neural network with an improved RPROP algorithm. As is with the features of high accuracy and strong autiinterference and fast recognition speed. Using the algorithm, a high recognition rate can be achieved and complicated mouse graph can be recognized. Experimental results demonstrate the algorithm is feasible, robust and applicable.