排种器性能检测中种子位置智能检测方法

An Intelligent Detecting Method for Seed Position in Performance Detection of Seedmeter

根据基于计算机视觉的排种器性能检测的特点 ,提出了基于神经网络的种子位置智能检测方法。采用两个光轴互相垂直的摄像机监测投种规律 ,并把以矩法求取的质心图像坐标作为输入节点 ,构造检测种子质心实际坐标的两个神经网络 ,并行处理左右两幅图像 ,从而快速求得种子位置。实践表明 ,此方法不仅能求出种子位置 。

Based on the feature of performance detection of a seedmeter, an intelligent detecting method for seed position was proposed in this paper. Two cameras with orthogonal optical axis were applied to monitoring seed position. Mass center of seed was regarded as detecting attribute, and the image coordinate of mass center calculated by method of moment was input into BP. Further more, two neural networks were constructed to synchronously process left and right images, so object position was gained quickly. The experiments showed that seed position could be obtained and the law of seed dropping was also monitored.