多方法融合技术在多目标位置检测中的应用

提出了将多种技术融合一体定位多目标的方法。首先根据计算机视觉原理，采用正交光轴的双摄像机配置检测系统；然后建立去除复杂实际背景和自适应二值化的动态图象预处理模型，并根据矩法搜索和求取目标质心；再利用神经网络模拟人眼感知事物的功能，确立图象点与空间点的映射关系，并将目标质心象点坐标作为输入节点，采用两个BP网并行处理左右两幅图象，快速求得目标空间位置。实践表明，此方法具有一定的实用性，并且在实际应用中取得了满意的效果。

Detecting Position of Multi Objects by An Integrated Method of Multi Techniques

An integrated method of multi techniques for detecting position of multi objects is proposed in this paper. Firstly, on the basic of computer vision principle, detection system is composed of two-CCD with orthogonal light shift. Secondly, Deleting background model and binarization model are established, and the mass center of object is searched and computed based on method of moment. Thirdly, by the function that neural network can simulate eyes, the relation between image points and special points is built. The mass center in image is input into neural network, two BPs are applied to processing a pair of left-right images, and special position of object is rapidly obtained. Satisfactory result is obtained in the experiments.