To improve the accuracy and efficiency of the dynamic speckle metric for non-destructively detecting far-field target hit-spot intensity in a Target-In-the-Loop (TIL) system, a multi-channel cooperative detection system for acquiring speckle signals is established. The theory of dynamic speckles, the simulation model of this system and the spatial-temporal spectral fusion characteristics are also investigated. As a first step, the power spectrum is obtained by filtering, auto-correlating and Fourier transforming the intensity fluctuations of dynamic speckle detected by the point detector. Then, the feasibility of speckle-metric, obtained by multiplying the spectrum with weights, is explored to monitor the target-focused spot. As a second step, the approach of splicing the temporal signals obtained from different spatial locations on the receiving plane is proposed. Moreover, the prerequisites of this approach are listed. Finally, the effectiveness of the proposed speckle metric obtained by fusing the spectrum is verified through simulations and experiments. The results show that the speckle metric decreases with an increase in the hit-spot size, and the four-channel space-averaging metric can improve the accuracy by a factor of 2 when each group of signals is uncorrelated. Moreover, the metric obtained by spatial-temporal fusion spectrum not only guarantees accuracy but also quadruples the system’s bandwidth. Therefore, the multi-channel cooperative acquisition of the speckle metric can monitor the hit-spot change of far-field moving targets more rapidly than current solutions.