In brief, feature is image pattern which neighborhood is different from neighborhoods of other features (for example by texture, or intensity). It is used for description of the image content for image retrieval and recognition purposes.
Feature descriptor is vector that robust for various changes for example for shifts, rotations, scales, illuminating variations. Descriptor is computed from pixels around feature point and is used for matching purposes. Feature matching is performed by comparing the descriptors using suitable similarity measure.
There are another variants of the term "feature" such as interest point, key point, local feature, feature point. I did not see the term "feature pattern" in the appropriate literature, but I think it is the same as the term "feature point" in your case.
There are a lot of algorithms for feature point detection and descriptor computation depending on your aims. You can see the excellent article about feature point detection Local Invariant Feature Detectors: A Survey and also the article in Wikipedia Feature (computer vision).
Here are the some links to the articles explaining feature detection and descriptor computation: Wikipedia Feature detection (computer vision), OpenCV Feature Detection and Description, Matlab Local Feature Detection and Extraction.