What is a Feature?

In the previous chapter, our main focus was filtering an image and applying different transformations on it. These are good techniques to analyze images but are not sufficient for the majority of computer vision tasks. For example, if we were to make a product detector for a shopping store, computing only edges may not be enough to say whether the image is of an orange or an apple. On the other hand, if a person is given the same task, it is very intuitive to differentiate between an orange and an apple. This is because of the fact that human perception combines several features, such as texture, color, surface, shape, reflections, and so on, to distinguish between one object with another. This motivates to look for more details that relates to complex features of objects. These complex features can then be used in high level image vision tasks like image recognition...