The feasibility of temperature field distribution reconstructing and its changes reconstructed by light-field probe technology is discussed. The light-field probe and CCD camera are used to obtain multiple groups of incident and outgoing light which is correspondence to each other through the target area. Fermat principle is used to establish optical path function to reconstruct the temperature field of the area. Matlab software is used for simulation. PCG iterative algorithm is chosen to solve the reconstruction problem of incomplete projection data. Reconstruction results show that the method is effective that can be used to reconstruct entire temperature field distribution. And it has some advantages comparing with traditional temperature measurement methods.