Unstructured Road Based On Independent Component Analysis, Feature Extraction And Segmentation

The development of Mobile Robot has imposing on the defense,society,economy and academy,and becomes the tactic research object of high technology of all countries. Visual aided navigation is one of the hot spot on mobile robot navigation. And visual positioning navigation system is an important component in visual aided navigation system. The research of road detection includes two field, structured road and unstructured road. In this paper, Independent Component Analysis(ICA) was introduced and applied to do character extraction and road segmentaion of unstructured road images. The main content of this paper is:For unstructured road images, the ICA texture basis images calculated with FastICA algorithm have multi-frequency and multi-orientation features and powerful to describe texture images,which is useful for texture feature extraction in image segmentation. We show that the ICA filters are able to capture the inherent properties of road images. The new filters are similar to Gabor filters, but seem to be richer in the sense that their frequency responses may be more complex. These properties enable us to use the ICA filter bank to create energy features for effective image segmentation.Integrating Gabor filters and independent component analysis is used for extracting unstructured road images' feature. That is,the road image is firstly filtered by a given bank of Gabor filters,and then higher-dimensional feature vectors are constructed from the filtered images.Next,the dimensionality of these vectors is reduced by means of principle component analysis(PCA).Finally,the independent components in the resulting vectors with dimensionality reduced are analyzed and extracted by using ICA for road classification. Experiments show which have better performance than ICA and Gabor filters in image segmentation.In order to extend the spatial describing ability of ICA basis images,this paper proposes a method for feature extraction by integrating wavelet and ICA,which make it be able to segment road images with large textons. Using multi-scale independent component analysis in wavelet transformed images(MWICA) were created according to the frequency of road image which extend the frequency describing ability of ICA basis images.