Method and equipment for adjusting a radar detection threshold

Claim 2

Method for adjusting a radar detection threshold, which consists in: breaking down the area monitored by the radar in one sweep period into sectors, for each sector, determining the histogram y(x) relating the possible values x of signal moduli received by the radar in the sector considered during one sweep period, with their frequencies of appearance y, approximating the histogram y(x) measured for each sector by means of at least one model m(x) defined analytically by an equation of variable x containing an exponential function on a polynomial of log x and/or a polynomial of x, in which the coefficients are adjusted so as to minimize the differences between the model m(x) and the measured histogram, and for each sector, deducing the value S of the radar detection threshold, starting from a set value of the false alarm probability value Pfa by using the relation with the model: ##EQU30## where N is the total number of values of received signal moduli contained in the model.

Claim 3

Method according to claim 1, wherein the histogram y(x)for a sector is created starting from a sample of values x of signal moduli received by the radar in the sector considered during one sweep period.

Claim 4

Method according to claim 1, wherein after creating a histogram and before its approximation by at least one model, it also consists in keeping in the histogram only those points located within a variation interval around the maximum appearance frequency, referred to as the validity interval.

Claim 5

Method according to claim 3, wherein said variation interval limiting the histogram starts with the smallest value of the received signal modulus appearing with a frequency equal to half the maximum frequency of appearance of a value of the received signal modulus, and ends at the smallest value of the received signal modulus that is both higher than that with the maximum frequency of appearance and for which the frequency of appearance becomes zero.

Claim 6

Method according to claim 3, wherein slope changes are detected on the side of the histogram closest to the highest values of received signal moduli, and the variation interval is stopped at the first detected slope change.

Claim 7

Method according to claim 5, wherein histogram slope changes are detected by smoothing by averaging on a sliding window several points wide followed by a calculation of the first derivative of the smoothed histogram and detection of the discontinuity when the derivative of the current point differs from the average of several previously calculated derivatives by a certain percentage.

Claim 8

Method according to claim 1, wherein model m 5 (x) used is defined analytically by the following equation: m.sub.5 (x)=exp[a.sub.2 l.sub.n.sup.2 (x)+a.sub.1 ln (x)+c+b.sub.2 x.sup.2 +b.sub.1 x] where a 1 , a 2 , b 1 , b 2 and c are coefficients adjusted so as to minimize the differences between the model and the measured histogram.

Claim 9

Method according to claim 1, wherein the model used m 3 (x) is defined analytically by the equation: m.sub.3 (x)=exp[a.sub.2 l.sub.n.sup.2 (x)+a.sub.1 ln (x)+c+b.sub.2 x.sup.2 ] where a 1 , a 2 , b 2 and c are adjusted so as to minimize the differences between the model and the measured histogram.

Claim 10

Method according to claim 7, wherein, before using the model to determine the value of the detection threshold, its coefficients a 2 and b 2 are compared so as to keep only the most influential and correct the model over several consecutive sweep periods either to the form m.sub.1 (x)=exp[a.sub.2 l.sub.n.sup.2 (x)+a.sub.1 ln (x)+a.sub.0 ] or to the form m.sub.2 (x)=exp[a.sub.1 ln (x)+c+b.sub.2 x.sup.2 ].

Claim 11

Method according to claim 1, wherein two models are used, one defined analytically by the relation: m.sub.1 (x)=exp[a.sub.2 l.sub.n.sup.2 (x)+a.sub.1 ln (x)+a.sub.0 ] where a 0 , a 1 , a 2 are coefficients adjusted so as to minimize differences between the model and the measured histogram and the other is defined analytically by the equation: m.sub.6 (x)=x exp(d.sub.2 x.sup.2 +d.sub.1 x+d.sub.0) where d 0 , d 1 , d 2 are coefficients adjusted so as to minimize differences between the model, and wherein the measured histogram and the model with the smallest differences from the measured histogram is used to determine the detection threshold.

Claim 12

Method according to claim 1, in which the model used is defined analytically by the following equation: m.sub.1 (x)=exp[a.sub.2 l.sub.n.sup.2 (x)+a.sub.1 ln (x)+a.sub.0 ] wherein the following variable changes are made in order to determine the values of coefficients a 2 , a 1 and a 0 minimizing differences between the model and the measured histogram y(x): u=ln (x) et v=ln (y) followed by a polynomial regression using the least squares method of order 2.

Claim 13

Method according to claim 1 using a model defined analytically by the following equation: m.sub.4 (x)=x exp(b.sub.2 x.sup.2 +b.sub.1 x+b.sub.0) wherein, in order to determine values of the coefficients b 0 , b 1 , b 2 minimizing the differences between the model and the measured histogram, the following variable change is applied: w=ln (y/x) followed by a polynomial regression using the least squares method of order 2.

Claim 14

Pulse radar for implementation of the process according to one of the previous claims, comprising a transmitter transmitting interrogation pulses through an antenna in order to scan the area monitored by the radar, a receiver, a distance gate unit connected to the receiver output and that switches the received signal to one of several parallel channels as a function of its lag behind the most recently transmitted pulse and therefore the distance of the obstacle at the origin of the returned echo, and a computer connected to the outputs of the distance gate unit and the direction controller capable of controlling the distribution of complex samples comprising a phase component and a quadrature component of the received, demodulated and sampled signal available at the output of the distance gate unit, into distance-direction cells covering the area monitored by the radar, the said radar being characterized by the fact that is also includes means of: grouping distance-direction cells into sectors partitioning the area monitored by the radar, defining a histogram y(x) for each sector relating possible values x of signal moduli received by the radar in the sector considered in a sweep period, with their frequencies of appearance in distance-direction cells in the sector considered, approximating the measured histogram y(x) for each sector using at least one model m(x) defined analytically by an equation of the variable x containing an exponential function on a polynomial of log x and/or a polynomial of x, for each sector, deducing the value S of the radar detection threshold, a set value of the false alarm probability Pfa using the relation with the following analytical model definition: ##EQU31## where N is the total number of values of received signal moduli contained in the model.