Application of Matrices for Facial Recognition
Abstract
Facial recognition has grown exponentially, driven by complex techniques such as artificial intelligence. There remains a need for simple, low-cost systems that can compare faces to pre-constructed data sets. Here we utilize the Principal Component Analysis (PCA) to process photos with numerous identifying variables and linearly reduce them into a matrix so that we can efficiently compare faces stored in a database. Two methods were used side by side to demonstrate effectiveness and results. Both methods took the facial scan and processed it into large vectors. Then placed those vectors into a large matrix where each column represents a face, and each row represents a number pixel. This matrix helps find the average face in the form of a column vector so that it can be deducted from the original face. Comparing each pixel when finding a face in a small data set works but proves exhausting in large data sets or large images. The PCA reduces the matrix so that the eigen values and eigen vectors can be found and used as identifying values. This method is simple, fast, and low cost. It's intended for police stations, hospitals, or even surveillance cameras for the average homeowner.
Application of Matrices for Facial Recognition
Facial recognition has grown exponentially, driven by complex techniques such as artificial intelligence. There remains a need for simple, low-cost systems that can compare faces to pre-constructed data sets. Here we utilize the Principal Component Analysis (PCA) to process photos with numerous identifying variables and linearly reduce them into a matrix so that we can efficiently compare faces stored in a database. Two methods were used side by side to demonstrate effectiveness and results. Both methods took the facial scan and processed it into large vectors. Then placed those vectors into a large matrix where each column represents a face, and each row represents a number pixel. This matrix helps find the average face in the form of a column vector so that it can be deducted from the original face. Comparing each pixel when finding a face in a small data set works but proves exhausting in large data sets or large images. The PCA reduces the matrix so that the eigen values and eigen vectors can be found and used as identifying values. This method is simple, fast, and low cost. It's intended for police stations, hospitals, or even surveillance cameras for the average homeowner.