T. Baidyk, E. Kussu (Mexico), and O. Makeyev (Ukraine)
Neural Classifier, Interpolator, Precision Microequipment, Microassembly, Image Recognition, Computer Vision.
The equipment for automatic assembly of microdevices must have high precision because the tolerances of microdevice components are very small. To increase the precision of microequipment we use the feedback based on computer vision principles. We propose image recognition method based on neural network. Two types of neural networks were developed to solve this problem. The first type we call the neural classifier. The second type we call the neural interpolator. The neural classifier and neural interpolator are used to obtain relative positions of micropin and microhole. The neural classifier gives finite number of relative positions. The neural interpolator serves as an interpolation system and gives infinite number of relative positions. The image database of relative positions of microhole and micropin of diameter 1.2 mm was used for neural classifier and neural interpolator tests. This database contains 441 images. Each image differs from neighbor ones by displacements of 0.1 mm along the axes X or Y. We describe and discuss the relative position recognition results for neural classifier and neural interpolator. The experimental results show that neural interpolator gives better results in estimation of the pin-hole relative positions.
Important Links:
Go Back
