Weld bevel recognition based on two-dimensional wavelets transform and pattern recognition
Welding is important technology for the production of many products. It is necessary to accurately identify the welding groove shape in welding engineering. The welding groove shape has an important influence on the performance of the welding parts and the welding quality. And the profile of the welding joint is the groove shape. In recent years, the development of vision technology and automation have enabled automatic recognition of weld bevels. This paper discusses the application of two-dimensional wavelet transform and pattern recognition in weld bevel recognition technology.
Two-dimensional wavelet transformation (2DWT) as a mathematical tool for image processing was first proposed by Strang and King in 1992. Since then, wavelet transformation has been widely used in many fields such as image compression, image segmentation and pattern recognition. An image can be decomposed into low and high frequency sub-bands. The sub-bands are the endpoints and they are combined in different ways to represent an image in a better way. Compared with the traditional Fourier transform, the wavelet transform has the advantages of multi-resolution analysis, excellent directionality and spatial localization. It can effectively remove noise in images and improve the accuracy of image recognition.
Pattern recognition is an important branch of artificial intelligence. It is a technology for automatically recognizing or distinguishing objects through computer algorithms. There are many methods for pattern recognition, such as statistical method, structural method and information granulation method. It can be used to recognize the shape, size, structure and other characteristics of objects. In pattern recognition, feature evaluation is a very important way. Feature points or feature vectors are extracted from the position, size, and orientation shapes of objects, and then various pattern recognition methods are adopted to analyze them. This paper uses the two-dimensional wavelet transformation for image preprocessing, and then uses the feature points to identify the weld bevels.
The bevel recognition algorithm based on two-dimensional wavelet transform and pattern recognition is as follows. First, the image of the weld bevel is acquired by vision technology. Then, the two-dimensional wavelet transform is used to extract the features of the weld bevel. Finally, the pattern recognition method is adopted to identify the weld bevels according to the characteristics of the weld bevels.
In practical applications, the two-dimensional wavelet transformation can effectively reduce the noise interference in images and improve the accuracy of weld bevel recognition. After the two-dimensional wavelet transform, the feature points are extracted, and then different pattern recognition algorithms are used according to the characteristics of the weld bevel. Consider the feature points such as the position, size, orientation, etc., and then use appropriate methods to recognize the weld bevel shape.
In summary, this paper proposed the recognition of welding joint slope using two-dimensional wavelet transformation and pattern recognition methods. The application of two-dimensional wavelet transformation in image preprocessing improves the accuracy of weld bevel recognition. The extracted feature points are used for pattern recognition, which can recognize the weld bevel shape more accurately. Therefore, this method is effective and applicable for recognizing weld bevels.
