The closely spaced objects (CSO) create blur pixel cluster on the infrared focal plane; therefore, in order to effectively track and identify each object of CSO, it is necessary for the sensor signal processing to resolve them. A novel method of CSO infrared resolution based on reversible jump Markov chain Monte-Carlo (RJMCMC) method is presented. The method firstly creates an infrared focal plane image model, then constructs a framework of Bayesian inference for CSO resolution, and subsequently uses the RJMCMC to perform computation of the parameters′ posterior distribution, ultimately the joint estimation of objects number is obtained, positions and intensities. Simulation with infrared sensor viewing midcourse CSO are carried out to test the performance of the method, and the results confirm the effectiveness of the method.