Research On Color Feature Invariance Based On Lie Group Learning Model

The Lie group theory based on manifold not only provides a geometric representationto describe the data, but also gives a specific algebraic solution. Relative to traditionallearning algorithms, Lie group can efficiently handle matrix samples to avoid the problemto classification and identification because of the sharp increase of the number ofdimensions. The color feature invariance is one of the main research questions of cognitiveimage while Lie group has inherent advantages in dealing with transformation, so thisthesis study the color feature invariance with Lie group learning model. After nearly threeyears of research, the results achieved include:(1) Propose the Lie group color feature invariant face tracking algorithm.(2) Propose the Lie group color feature invariant descriptor which is also call Liefeature descriptor.(3) Propose the Lie feature descriptor template matching algorithm and apply it toimage matching and object tracking.In summary, the innovations of this paper are as follows.(1) Make full use of the color feature invariance of elliptical model of skin tone andpropose the Lie group color feature invariant face tracking algorithm.(2) Propose the fast computing method of Lie feature descriptor inspired by integralimage and improve the computing speed significantly.