Resaerch On Visual Feature Extraction Technology Of Industrial Robot Based On BOW

With the increasing degree of digitalization and informatization in today's society,more and more visual information exists in the form of digital images in people's daily life and production.As an important part of image processing and robot vision,feature extraction technology can deal with visual information efficiently,get the information people need,and bring convenience to life and industrial production.It is widely used in industrial robot vision technology,satellite remote sensing technology,information retrieval technology and image processing.As the eyes of industrial robots,robot vision technology is essential for industrial robots,and feature extraction technology is one of the most important technologies in the field of industrial robot vision.In this paper,the identification and classification of objects by industrial robots as the application background,combined with the text processing from the field of BOW(Bag of Words)model,the industrial robot visual feature extraction technology research.The feature extraction algorithm of SIFT(Scale Invariant Feature Transformation)is the main point of entry,and the improvement of SIFT algorithm itself is made.At the same time,combined with the SIFT feature,the BOW model in the text field is redesigned to form a new feature expression model,which makes the SIFT feature extraction algorithm better applied to industrial robots.The main contents and innovations of this paper are as follows:(1)In this paper,the feature extraction algorithm is researched for the problem that the industrial robot is not enough to identify and judge the object in complex background environment.Using the SIFT algorithm as the main point to start the research which is a better algorithm in complex background image feature extraction,through a large number of reading literature and experiment.(2)In view of the shortcomings of SIFT feature extraction algorithm for the lack of illumination ability of edge and non-linear change,the Laplacian edge operator is used to improve the problem of insufficient edge feature extraction.At the same time,an arctangent normalization method is proposed to replace the original normalization method to improve the non-linear variation of the illumination condition.(3)After a lot of research found that SIFT features are used more in the object matching at the field of robot object recognition,because the SIFT feature can not be directly classified by the existing classifier.In view of this,this paper proposes a new feature expression model based on the BOW model of text categorization and PCA(Principal Component Analysis).The SIFT extracted visual features are re-expressed,and then enter the classifier for classification,in order to achieve the object recognition and judgment.(4)Aiming at the shortcomings of existing industrial robot safety protection technology,this paper presents a new scheme of industrial robot security technology based on robot vision technology,and proves the feasibility of the scheme by experiment.All above research and innovation content is finally verified experimentally.The experimental results show that the improvement of the algorithm is successful and the design of the feature expression model is reliable.The research results of this paper have practical application value.