Research On Image Edge Detection Algorithm Based On Median Filter

In industrial vision,the image generation process will be affected by many aspects such as the environment and signal transmission,resulting in noise disturbance in the final image output result and distortion caused by the loss of real information.In order to restore the real information of the image for subsequent image analysis,it is necessary to remove the interference noise in the image by the method of image filtering.The currently proposed image filtering algorithms are often limited to only one type of noise model removal,and in addition to the filtering algorithm for the Gaussian noise model,there are fewer filtering algorithms and denoising for models such as salt and pepper noise and Poisson noise also is the poor effect.Image denoising is a basic task in image processing,and edge extraction after removing noise interference is the basis and key to highquality image analysis and measurement.This paper focuses on the difficulty of edge extraction in the digital image processing step.From the essence of image edge extraction,the filtering algorithm of salt and pepper noise in filtering is explored.The classic edge detection operator is improved,and it is found that the second-order differential operator pulls the superior performance of the Laplacian noise reduction model,combined with morphological filtering on the basis of the improved median filtering algorithm,realizes the enhancement and effective extraction of edges.The innovation of this paper is mainly to propose a second-order edge operator combined with morphological filtering enhancement.Before the edge enhancement,a new median filter with multi-window threshold adaptation is used to filter out salt and pepper noise.After many tests in the data set pictures and continuous comparison with other classic algorithms,it is concluded that the algorithm in this paper is suitable for edge extraction in the background of high-density salt and pepper noise in industry.The simulation results prove that the method has more accurate results in higher density or complex industrial scene applications.The article finally built a machine vision measurement system.The verification of the actual measurement results also shows that this method can accurately and effectively detect the size of the parts,greatly reducing the probability of product misjudgment,and providing new ideas for the improvement of industrial inspection lines.