Abstract: The invention relates to a method for improving the angular resolution of a monopulse radar.Starting from the sum signal S, a direction signal .theta..sub.b, and a signal representing the quadrature angle-error measurement .vertline..epsilon..sub.q .epsilon., the method consists of calculating in a first device 201 a signal .DELTA..theta. representing the wingspan of the target, then of calculating in a second device 202 two measuring signals .theta..sub.A and .theta..sub.B representing respectively the direction of the external limits of the target, as well as a signal Q representing the quality of these measurements and finally of carrying out an adaptive filtering operation on the measurements in a third device 203, in order to obtain filtered estimates for .DELTA..theta., .theta..sub.A and .theta..sub.B.

Abstract: To improve the accuracy of angular measurements in a radar or sonar system by suppressing noise due to target glint, an angular-deviation signal .epsilon. derived from the output signals .SIGMA. and .DELTA. of a sum channel and a difference channel is multiplied with a gain-control signal K(t) before being subjected to recursive filtering. Signal K(t) is obtained, also by recursive filtering, from a weighting function P(t) which goes to zero during periods when the measurement of the angular deviation is unreliable as determined from a low value of sum signal .SIGMA. and from a peaking of a quadrature deviation signal .epsilon.q, function P(t) being proportional to the product .SIGMA..sup.n .multidot.(S-.vertline..epsilon.q.vertline.) where the exponent n is at least equal to 1 and S is a predetermined threshold. The recursive filter processing the deviation signal .epsilon. has two cross-connected stages, both controlled by signal K(t), respectively emitting a corrected deviation measurement .epsilon.