A popular neuroimaging method for tracking brain activity is fNIRS. However, motion artifacts might skew the recorded data, making fNIRS less useful. In the context of designing and analyzing brain imaging data for both adults and children, it is the most significant source of noise and can seriously impair the quality of the collected data. The relative effectiveness of the several methods established to clean the fNIRS data from motion artifacts can sometimes be ambiguous due to a lack of knowledge about the ground truth signal. In this work, we employed a signal processing method that includes the ground truth signal in order to distinguish between motion artifacts and actual brain activity. A custom hybrid filter is proposed including an Adaptive filter and a Moving average filter for removal of artifacts from fNIRS signal and compares its performance with existing techniques like Principal Component Analysis (PCA). For both the motion artifact correcting methods, three most crucial validating parameters—Signal to noise ratio (SNR), Pearson product-moment correlation coefficient (CC) and Root mean square error (RMSE) were examined. According to the experimental results, the hybrid filter outperforms PCA and offers a reliable method for motion artifact reduction.