In this thesis a contactless-type algorithm by Leng et al. (2014) for region of interest (ROI) extraction from palmprint images is presented. The algorithm identifies the palm area, detects the finger valleys, and extracts the ROI used for biometric identification. The implementation is done in Python using OpenCV and NumPy, following the theoretical model of Leng et al. The system is tested on the CASIA and 11k-Hands databases as well as on self-recorded images. Results show that the algorithm performs well under good lighting conditions and with spread fingers, achieving reliable ROI extraction for most images. Limitations occur when the thumb valley is not visible, when the background is similar to skin color, or when the hand posture deviates from the expected orientation. The work also proposes an alternative hand-side detection method to improve recognition accuracy. Possible future work includes using convolutional neural networks and improved skin models for more robust ROI extraction.