To reduce the influence of noise in infrared spectral signal measurement, a topological derivative improved partial differential equation method for infrared spectral data denoising is proposed. As an indicator function, topological derivative through a minimization process to find the best position to introduce disturbance, where are spectral edge points, then select the most excellent diffusion coefficient, so the cost of the minimum functional value. Based on the idea of topological optimization, it makes the lowest topological derivative to be optimum one. Then the diffusion is applied by using partial differential equation. Several simulated infrared spectral sequences are utilized to verify the performance of the proposed method. The experiment results show that our method is better in denoising.