基于光电混合系统的家用机器人视觉系统

指出家用机器人比工业机器人将具有更广泛的应用前景。构建一种基于光电混合系统的家用机器人的视觉系统,以联合变换相关器实现目标图像的平移不变性,以综合判别函数和WTA网络构成级联式神经网络实现目标图像的畸变不变性。讨论整个系统的原理及其实施方案,论证其可行性,为家用机器人提供了一种有效的多目标实时模式识别方法。

The Visual System of Household Robot Based on Opto-electric Hybrid System

The housedhold robot shows wider practice situation than industrial robot. This paper presents a visual system of household robot based on optic-electric hybrid system. A union transforms correlator can realize the shift invariant of target image; the synthetic recognition function and Winner Take All network will build cascaded neural network so as to realize distortion invariance of target image. In this paper the principle of the method and its implemental issues are addressed, and it proves the feasibility. This system will certainly provide an effective method for real-time pattern recognition of multi-target.