Research On Vibration Fuzzy Image Restoration And Recognition Of The Crane Robot

Vision aided decision can significantly improve the working reliability of the crane robot lifting system, in the process of crane operation, how to deal with environmental image blur caused by vibration, is the key to improve the crane robot image identification and positioning accuracy. Against the application background of crane robot visual identification system, the vibration blurred image acquisition, processing, restoration and identification technology are studied. Some achievements are as follows:First, the model known elemental body vibration blurred image acquisition experiment platform is designed and built, which is used for collecting vibration blurred image, and three-dimensional structure model of vibration experiment platform is set up. Secondly, in order to restore the vibration blurred image, reduce or inhibit vibration fuzzy to the greatest extent, improve image quality, the causes that environmental image fuzzy caused by the mechanical vibration are analyzed, and the mechanism of image degradation due to vibration is studied. Considering that the traditional way to restore space-variant blurring image is to use point spread function of the whole image, in the exposure time, movement in the form of invariant can use the same PSF to recover, and in the imaging process image motion caused by vibration is space variant, and the recovery precision is limited. Therefore, the line by line method is put forward to restore the space-variant blurring image, at the same time, according to the relationship between the exposure time and vibration period, the low frequency and high frequency vibration blurred image restoration method are respectively discussed; On the basis of vibration fuzzy restored image, moment invariant neural network fuzzy identification method is applied to realize recognition of the target object. Aiming at crane hook, a variety of image acquisition under different poses of hook are identified. The experimental results show that when the restored image invariant moments are calculated directly, the separation of different target image invariant moment values are poor, and the recognition accuracy is low; The image invariant moment is to be calculated after Sobel edge feature extraction, the separation of invariant moment values are better, the recognition accuracy is effectively improved. Finally, bearing objects are taken as the target interference samples, and the validity of the identification method is proved.