Local discriminant basis algorithm-a review of theory and applications in signal processing

Local discriminant basis (LDB) algorithm is a powerful algorithmic framework that was originally developed by Coifman and Saito as a technique for analyzing object classification problems. Prior to the development of LDB, an adapted waveform framework called best basis algorithm had been developed mainly for signal compression problems. The main advantage of LDB over other similar techniques such as Karhunen-Loeve transform (KLT), also known as principal component analysis (PCA), is its lower computational cost of O(n log n) order. This paper is the outcome of a literature review on theory and applications of LDB in signal processing.