The main motivation behind SIFT is to extract local features from an image that is robust. To achieve this, the algorithm is divided into the following four main stages:
- Scale-space extrema detection
- Keypoint localization
- Orientation assignment
- Keypoint descriptor
If you carefully read Chapter 3, Drilling Deeper into Features-Object Detection, you will realize that this sounds very similar to the ORB algorithm (SIFT was proposed before ORB). The takeaway from this point is that most of the feature-detection algorithms have the same motivation to extract robust features and hence they have a very similar approach. All the algorithms do something unique in each of these stages, which differentiates them from each other ...