Fourier Descriptor for Pedestrian Shape Recognition using Support Vector Machine

The main objective of this study is to analyse Fourier Descriptor (FD) as feature vectors for pedestrian shape representation and recognition. FD is chosen since it is the best known boundary based shape descriptor and has proven to outperform most other boundary based methods in terms of accuracy. FD is also invariant to geometric transformations and has good noise tolerance. Initial results showed that using 10 descriptors of both low and high frequency components of pedestrian and vehicle shapes are sufficient for recognition based on high classification rate achieved. Moreover, the tremendous performance of Support Vector Machine (SVM) as classifier is confirmed based on the Kappa Score calculated. These findings have proven that our method is an effective approach for pedestrian recognition.