Motion estimation and correction implementations are important components of medical image analysis. We present a new approach for motion estimation and correction of thermographic images in brain surgery. In a pre-processing step, the Phase Correlation (PC) method is performed in order to detect large displacements of objects in two successive images. Additionally, due to noise in thermographic images, a Cellular Nonlinear Network (CNN) based image enhancement method is applied. Then, in the following processing step, the Optical Flow (OF) method is employed to compensate local motion artifacts. The proposed algorithm is evaluated during an offline analysis of the recorded dataset of brain surgeries and the performance evaluation between different algorithms is made based on the determination of the Normalized Cross-Correlation (NCC). The results clearly indicate that the proposed algorithm is able to reduce breathing motion artifacts effectively as well as the NCC evaluation show better results in comparison to other mentioned algorithms.