With the continuous growth of the number of satellites in orbit, the requirements for observation quality in remote sensing applications are constantly increasing. For compound tasks that require multiple observations and where the observation intervals affect the imaging quality, existing studies have difficulty effectively characterizing the impact of observation time intervals on the consistency of imaging information. To address this issue, a collaborative scheduling method for hybrid remote sensing constellations oriented to compound observation requirements is proposed. Firstly, a time window pruning preprocessing method based on conflict degree is adopted to reduce the huge search space generated by the collaboration of heterogeneous payloads. In the mathematical model, a time-varying weight function with the time interval as the independent variable is introduced, and a piecewise approximation method is used to represent the nonlinear relationship between the interval length and the information quality. At the algorithm level, a hybrid variable neighborhood search algorithm based on divide-and-conquer framework (HVNSA-DCF) is proposed. Through a two-stage architecture with a feedback mechanism, multiple neighborhood operators, and an adaptive neighborhood scoring selection system, efficient search of large-scale scenarios is achieved. Experimental results show that the HVNSA-DCF demonstrates good adaptability and convergence in scenarios of different scales.