Abstract: Adaptive Compressive Sensing and Processing (ACSP) can reduce the computational load, but existing radar target tracking methods based on adaptive compressive sensing and processing are limited to single-target tracking. ACSP achieves multi-target tracking. Through the sparse representation of echoes, the improved dictionary (sparse transformation matrix) is designed. In the measurement process, adaptive weights are used instead of random Gaussian matrices to construct and configure the perceptual matrix. The measurement model is established based on compressed sensing sampled data. This overcomes the data association problem in multi-target tracking. Due to the nonlinear relationship between the measurement and the target state, the likelihood particle filter combined with the joint probability data association method is used to estimate the target state in real time. Theoretical simulation experiments show that the improved adaptive sensing and processing method achieves multi-target tracking.