论文部分内容阅读
为了准确、快速地进行图像测速，介绍了一种前向三层自组织竞争人工神经网络，并将其应用于图像测速技术的粒子轨迹测速中。人工神经网络在图像测速技术中的应用才刚刚起步，该方法利用神经网络智能与快速的特点完成粒子的匹配，从而获得流场的速度。与以往的算法相比，神经网络方法具有检测准确和速度快的优点，但对图像质量要求较高。在恒定流流动表面流速的测量和Ｏｓｓｅｎ涡的模拟中分别使用了神经网络方法，获得了良好的测量结果，并证明了该方法在图像测速技术中的应用潜力。
In order to accurately and quickly image velocimetry, a three-layer forward self-organizing competitive artificial neural network is introduced and applied to particle trajectory velocimetry of image velocimetry. Artificial neural network has just started its application in image velocimetry. This method uses the intelligent and fast characteristics of neural network to complete the matching of particles and obtain the velocity of the flow field. Compared with the previous algorithms, the neural network method has the advantages of accurate detection and fast speed, but requires high image quality. The neural network method was used in the measurement of the flow velocity of constant flow and the Ossen vortex simulation, respectively. Good measurement results were obtained and the potential of this method in image velocimetry was proved.