The growth of remote sensing (RS) image data fuels the interest in embedded system energy-efficient processing. To reduce the storage and communication requirements, image compression is widely popular with the RS domain. However, RS image compression is a computationally intensive task. How to balance compression performance and efficiency is a challenging problem in embedded systems with limited resources. In this paper, an efficient heterogeneous programmable multi-processor system on chip (MPSoC) embedded processing system for RS image compression is proposed. Firstly, an RS image compression method is proposed based on the region of interest extraction and implemented method on programmable logic. Then, a processing structure is designed based on MPSoC and develops an embedded processing module. Finally, the experimental results show that our method can improve the compression ratio while ensuring the quality of the target information. The efficient RS image compression task could be achieved by using the embedded processing system.