Recently, high-frequency oscillations (HFOs), which consists of ripple and fast ripple frequency components, are the promising indicator for epilepsy surgery. The automatic detection of epileptic focus from long-term interictal intracranial electroencephalogram (iEEG) is highly demanding to epileptologists for accurate treatment. In this study, filter-bank features are used to detect epileptic focus in high-frequency components of interictal iEEG signals. The interictal iEEG signal is segmented into 20-s duration and decomposed into a number of subbands using filter-bank technique. Five feature-extraction methods including coefficient of variation (CV), fluctuation index (FI), variance (Var), root mean square (RMS), and difference absolute standard deviation value (DASDV) are computed from each subbands. The features obtained from all subbands are concatenated to construct the features, which are discriminated into focal or nonfocal iEEG using the support vector machine (SVM) with radial basis function kernel. The experiment is performed with eight patients with temporal lobe epilepsy caused by focal cortical dysplasia (FCD) to evaluate the performance of the proposed method. The experimental results show that proposed method can effectively identify the focal and non-focal segments, which has a neurophysiological perception and much appropriate by clinical experts.