Application of PCA as an acoustical signal processing tool for object classification purposes

The authors presented a technique for an optimal representation of acoustical signals for further object classification purposes using different statistical and neural methods. It is based on principal component analysis (PCA) which is a transformation of vectors localized in k-dimensional observation (feature) space into lower n-dimensional component space retaining majority of included information. The resulting improvement in classification efficiency by a chosen statistical classifier was verified by a numerical experiment.