This study presents a Gaussian noise removal method using the improved trace-based approach for color images. The presented method employs partial differential equations (PDEs)-based approach including both smoothing (regularization) term and fidelity (data) term in the energy functional. So, the structure of the input image is well preserved during noise removal processes. Also, we estimate the standard deviation of Gaussian noise in the wavelet domain. In addition, due to the fact that the type of Gaussian noise is easily perceived via a selected block in a flat (homogeneous) region of the noisy input image, the presented method is considered as a semi-automated noise removal approach. The validity of the presented method is depicted via experimental results.