Extraction of Depth Information for Micromanipulation Using Normalized Spectral Principal Component Analysis

This paper proposes a novel spectral analysis method named normalized spectral image,which is invariant to translation and scaling.The influence of large-scale movement of micro-manipulation tools in the testing window(X-Y plane) on the algorithm robustness can be eliminated by using normalized spectral images.Meanwhile,principal component analysis(PCA) technique is used to reduce the dimensions of data and to suppress noise,and the real-time performance and accuracy of the algorithm is improved.Experimental results demonstrate the effectiveness of the proposed method.