Corrections to "Optimal detection using bilinear time-frequency and time-scale representations"

Sayeed and Jones (IEEE Trans. Signal Processing, vol.43, p.2872-83, 1995) develop an optimal detection framework based on quadratic time-frequency and time-scale representations. The expressions for certain test statistics in that paper contain a factor of 1/2 that is not needed. Moreover, certain "MAP GLRT" detectors are proposed in Sayeed and Jones whose forms should be slightly different than those described there. The following section describes the appropriate changes that rectify the situation.