Mastering Computer Vision with TensorFlow 2.x

Summary

In this chapter, we learned how to take an image pixel and threshold it with its neighboring pixels within a given radius and then perform a binary and integral operation to create an LBP pattern. The LBP pattern is a good example of unsupervised machine learning as we did not train the classifier with the output; instead, we learned how to adjust the parameters of LBP (radius and number of points) to arrive at the correct output. LBP was found to be a very powerful and simple tool for texture classification. However, when the image was non-textured, LBP did not return good results and we learned how to develop an RGB color matching model to match colored non-textured images such as face and foundation color. To create an LBP representation, ...