Application Research Of Digital Image Processing Technology In Rolling Deformation Measurement

When the roll angle is measured by moiré fringes,it is necessary to separately collect the moiré fringe images before and after the torsional deformation.In practice,this is generally completed by an image acquisition device such as CCD array.However,due to the limitations of the external environment,the cost of the image acquisition device and the manufacturing process,most of the images are of poor quality: the image is blurred,have no clear edge,and with a lot of noise mixed in.These factors greatly lower the resolution of images,and thus affect the subsequent measurement accuracy.In order to improve measurement accuracy,it is important to improve image resolution.In order to improve the measuring precision of roll angles,this topic starts from the moiré fringe images,and its research content can be mainly divided in two aspects:1.The super-resolution-based image processing algorithm of moiré fringe is completed: firstly,the moiré fringe image is preprocessed,and the noise in the image is removed by using frequency domain filtering and singular value decomposition;then the learning-based super-resolution algorithm is used to form the training sample set,and a super-resolution algorithm for moiré fringe images is designed.Finally,a post-processing algorithm of deblocking is designed to filter the block effect introduced by super-resolution reconstruction.To confirm the effect of the algorithm,an experiment aimed to comparing the results of different super-resolution algorithms is designed.The result shows that the whole image processing algorithm of moiré fringe has certain superiorities in both subjective and objective indicators.2.The method of calculating the roll angle using moiré fringe is completed: the moire fringe images collected are often not ideal parallel fringes,and sometimes are even bend.If the traditional processing method is used to extract and fitting the skeleton of the moiré fringe directly,we can't actually get the correct and reliable stripe parameters.In order to solve this problem,a hypothesis is put forward,the hypothesis assumes that the moiré fringes with bending and non-parallel phenomena retain some properties of their corresponding ideal stripes.According to this hypothesis,the media parallel line group is introduced.After the image thinning,the intersection coordinates of the two sets of stripes are used to transform the approximation problem of the moiré fringes before and after the rolling into an optimization problem,and to obtain the stripe parameters which can optimize the limitation factor.Multiple sets of moiré fringe images were collected and the experiment was carried out to verify the algorithm.The result of the experiment indicates that the algorithm can effectively reduce the error caused by directly fitting the skeletons of the stripes and the measurement error can be controlled within 2".