A radar wherein a first threshold value is determined in accordance with the intensity of background noise; second threshold values are set in a fixed frequency area around base portions of peaks in the frequency spectrum; the higher one of these second threshold values is adopted as a decisive threshold value in each FFT bin; and a peak exceeding this threshold value is considered a target peak and detected. Thus, the target peak caused by the reflected wave from a target included in the frequency spectrum of a beat signal is surely detected and detection accuracy of the target is increased.
An on-vehicle radar device has a transmission section for transmitting a radio wave to an object, a receive section for receiving the radio wave reflected by the object, and a processing section for dividing an object detection range into three or more of a plurality of areas, setting a threshold of an intensity of the radio wave received for each of the plurality of areas, and judging the existence of an object by comparing the intensity of the radio wave and the threshold. This processing section sets, based on an auto-cruise control mode or pre-crash mode of the vehicle, a threshold of a part of an area in the object detection range to be lower than the threshold of the other areas, or changes the threshold of the detection area according to the object detection status.
A frequency modulation radar apparatus for vehicle use can suppress the influence of noise to avoid incorrect estimation due to noise thereby to provide a beat frequency with high accuracy and at high speed without increasing the frequency resolution of the beat frequency that causes an increase in an observation time. The apparatus includes a frequency correction section that calculates a corrected frequency (f.sub.n+.delta.) by adding an amount of frequency correction (.delta.) to the frequency (f.sub.n) of a peak signal, and a CPU that calculates a distance or a relative speed to a target object based on the corrected frequency (f.sub.n+.delta.). In an FFT calculation section, the frequency (f.sub.t) of the true peak signal is calculated based on the characteristic of a window function, and if the frequency (f.sub.t) thus calculated is determined as an incorrect estimation, the frequency of the true peak signal is further corrected.