This paper shows filtering effects on performances for multi scale and orientation of filter bank in the image analysis using subspace classifier. Based on three kinds of intensity distributions, the feature extraction was firstly conducted, and the feature vectors and subspace dimension for recognition were examined.
Afterwards, a series of the analysis concerning accuracies were conducted in the cases of single filter and filter bank. The effects on classification performances for filter bank were compared with the performances for single’s type filter. The deference of filtering effect with color spaces was clarified using the significance degree by SOM.
