Convolution calculation is executed on the basis of a calculation result from an impulse response calculation unit (10), (11), ... (1n) using a peak value of each pulse as a multiplier and an estimated value of an unknown impulse response as a multiplicand and using a transmission waveform as a pulse train, and the difference from a reception waveform is calculated by a difference calculation unit (3). Correction of the estimated impulse response is made at a correction unit (10a), (11a), ... (1na) of the impulse response calculation unit (10), (11), ... (1n) on the basis of the calculated difference, and the corrected impulse response is supplied to an input response calculation unit of a next stage. When these processings are executed, high resolution of inspection as well as real time processing can be accomplished, and moreover, the range of inspection can be expanded while keeping resolution at a high level.