Extracting the features

Once an appropriate image is captured, the algorithm must complete the following steps:

1. Detect the face, and then crop and normalize the image.
2. Discover the facial landmarks.
3. Convert the metrics of the facial landmarks into a template to be either stored or compared to an existing database.

The first step in the process is to localize the face, as discussed next.

Localizing the face

After acquiring an image, the next step is to detect a human face in an image or video. Once the face is located, the image is prepared by cropping and normalizing. As shown in the following image, this phase seeks to prepare the image.

At a high level, you might see the following steps:

1. Remove any non-face artifacts.
2. Detect the eyes.
3. Rotate and scale the image appropriately.

The following diagram shows the high-level steps taken to prepare an image:

Figure 6.3 – Preparing the image

As shown...