Features use cases
Following are some of the generic applications that are popular in computer vision:
- We have two images and we would like to quantify whether these images match each other. Assuming a comparison metric, we say that the image matches when our comparison metric value is greater than a threshold.
- In another example, we have a large database of images, and for a new image, we want to perform an operation similar to matching. Instead of recomputing everything for every image, we can store a smaller, easier to search and robust enough to match, representation of images. This is often referred to as a feature vector of the image. Once a new image comes, we extract similar representation for the new image and search for the nearest match among the previously generated database. This representation is usually formulated in terms of features.
- Also, in the case of finding an object, we have a small image of an object or a region called a template. The goal is to check whether an image...
