Abstract

This paper discusses the deployment of a Machine Learning (ML) classifier for categorising the level of deformation of thin plates. This work involves acquisition of a sequence of reflected images, an image processing sequence as required and the deployment of a suitable ML classification model. A laser optical bench is used for acquiring sequence of reflected images from the surface of the specimen, which is under continuous micro-scale deformation. Then, two data sets are generated from the acquired images, from their image features, using statistical functions and geometrical functions. A combined data set, containing both the statistical features and geometrical features, is given to various machine learning classifier models and their performance is analysed. After a comparative analysis of various classifier models, it is found that a Naive Bayes classifier is highly suitable for classification of the images from their features. After ranking the image features, the Naïve Bayes Classifier is deployed for interpreting the image features towards categorising the level of deformation of the specimen, as per the defined labelling process. The Naïve Bayes Classifier generates a model after getting trained itself from the image data and the labelling process. On testing, it is found that the generated Naïve Bayes classifier model is capable of classifying the level of deformations accurately as safe deformation and excessive deformation.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call