Optical imaging based on intrinsic signals is widely used in visualizing the functional organization of the cortex. However, the evoked optical changes are small compared to the background optical intensity. It is critical for the application of this technique to separate the stimulus-driven optical change from the background optical changes. The background optical changes or noises are generally independent of the stimulus presentation. Therefore, in this study we propose a signal extraction method based on the temporal relation between the stimulus presentation and optical changes recorded. We have applied this method to the primary visual cortex of cats. It was demonstrated that the noises caused by the movement of blood vessels or the cortical surface could be removed by the proposed method.