The Study of Image Feature Extraction Based on Independent Component Analysis

摘要

This paper mainly focuses on the independent component analysis (ICA) and image recognition algorithm research, presents the supersaturated algorithm which mixed signal dimension can be smaller than the separation independent component dimension, ensure the reasonable precision. In the actual application it means that using a small number of signal acquisition device can get the character of the object, which largely reduces the processing cost, extends the application domains of ICA.