Research And Application Of Intelligent Edge Detection Based On Hybrid Filtering Algorithm

Visual servo technology enables the robot to have a target recognition function,which can enhance the intelligence and flexibility of the production line,and promote the continuous improvement of production efficiency.Image denoising and edge detection are key points in visual servoing technology,and the quality of the algorithm directly affects the quality and accuracy of image processing.Image noise is the interference signal generated by the environment,transmission path,equipment and other factors during the imaging process.The main types of noise are Gaussian noise and impulse noise.Most current image denoising algorithms deal with a single type of noise and cannot effectively deal with mixed noise.On the other hand,many industrial applications of vision require the detection and extraction of the edge information of the target object.The core of the vision is image edge detection and pose recognition.In the complex environment of the factory,especially when the local illumination is uneven,it is difficult for conventional edge detection algorithms to accurately extract the complete edge information of the target.This paper focuses on the research,improvement and optimization of image mixed noise filtering,local binarization and image edge detection under factory application conditions.The main work of the paper is as follows:In terms of mixed noise filtering,the paper analyzes the advantages and disadvantages of various median algorithms and their improved algorithms.Based on the extreme median and adaptive median algorithm,an improved median filtering algorithm is proposed,which can detect impulse noise points more effectively and filter impulse noise more effectively.And this thesis comprehensively uses the improved median algorithm and wavelet threshold algorithm to denoise mixed noise to achieve the purpose of filtering mixed noise better.In the aspect of local illumination imbalance binarization,this paper compares and analyzes algorithms such as Otsu,Kittler,Niblack,and Sauvola.Based on the improvement of the Niblack algorithm model,this paper proposes an image binarization processing method with fast processing speed and good effect.In the aspect of image edge detection,this paper compares and analyzes the traditional edge detection algorithms in the spatial domain,such as Roberts,Prewitt,Sobel,Laplacian,and Canny operators,and expounds the wavelet modulus maximum edge detection algorithm in the frequency domain.Based on the Canny algorithm and the wavelet modulus maximum edge detection algorithm,this paper proposes an algorithm that combines the threshold idea of the Niblack algorithm and uses its improved algorithm to improve and optimize the algorithm.Finally,through simulation experiments on the improved algorithm,it is proved that the effect of the improved algorithm is better.At the same time,in order to verify the practical application effect of the improved algorithm,the improved algorithm was applied to the recognition and grasping process in the robot grasping conductive rod project,and a good application effect was achieved.