Research On Image Edge Detection And Pattern Recognition Technology

In recent years,image recognition technology based on computer vision has attracted great attention.The online detection technology developed from this technology has also been widely used and deeply studied at home and abroad with its unique advantages.At present,on-line inspection of production line products mostly adopts manual visual inspection.This traditional inspection method has some shortcomings,such as time consuming,low detection efficiency and accuracy.Therefore,the research and development of on-line inspection technology and system is of great significance for saving costs,improving product quality and improving production efficiency.n this paper,the key problems of on-line detection technology based on computer vision,such as denoising and its evaluation,image edge detection and feature extraction,feature-based pattern classification,are studied.And the key technologies of this paper are validated against the background of defect detection of mobile phone protective films.The the wavelet threshold denoising algorithm is compared with other denoising algorithms in this paper,and its advantages and disadvantages are analysed in the meantime.Aiming at the problem of global fixed wavelet threshold,an adaptive threshold setting method is proposed,which changes layer by layer.In particular,to overcome the drawbacks of soft and hard threshold processing functions,such as image distortion,a smooth function between the two threshold processing functions is proposed,which can improve the traditional wavelet threshold denoising function.The experimental results show that the improved algorithm has good denoising effect.The basic principle of image gradient edge detection is analyzed,and the edge detection effects of different edge detection operators are compared,and the influence of image brightness on edge detection results is compared.The shortcomings of traditional Canny operator are analyzed.To overcome them,the OTSU method is used to set the threshold,and the combination of improved wavelet threshold denoising and adaptive median filter denoising is used to improve the adaptive ability and anti-noise ability of Canny operator.By analyzing the relationship between the relative position of boundary points and the direction of boundary,a boundary tracking algorithm based on the priority search direction is proposed to narrow the search scope,improve the search efficiency.At the same time,the accuracy of tracking results is improved considering the existence of boundary breaks.Hu moments and Zernike moments of boundary images are used as image features for pattern recognition,and their pattern recognition effects are compared and analyzed.The better Zenike moment is used as image feature to classify the toughened film image of mobile phone,and good classification results are obtained.