We present temporal segmentation and retrieval of stored memories or patterns with neural networks of a widely used model, the Fitzhugh-Nagumo neurons. For a superposition of several stored input patterns, it is shown that the proposed neuronal network is capable of segmenting out each pattern one after another in the time domain as synchronous firings of a subgroup of neurons. And as a corrupted input pattern is presented, the network is shown to be able to retrieve the perfect one. That is to say it has the function of associative memory. By adjusting parameters as coupling strength and intensity of the noise it is shown that the temporal segmentation attains its optimal performance at intermediate noise intensity, which is reminiscent of the stochastic resonance observed in the coupled spiking neuronal networks.