Image segmentation in light field imaging is a fundamental problem in digital image processing and analysis, with broad applications in areas such as augmented reality, healthcare, and biomedical imaging. While numerous algorithms have been proposed primarily leveraging convolutional neural networks and deep learning approaches to estimate depth and define pixel clusters this work introduces a novel pipeline that integrates light field depth estimation with the directed random walk algorithm, a method frequently used for refining depth maps. We evaluated our approach using both a controlled 4D Light Field Benchmark dataset and a real-world image database. Although the algorithm showed difficulty improving depth quality in real-world conditions due to increased noise and uncontrolled environments it maintained computational efficiency and demonstrated promising results when compared to existing techniques. These outcomes suggest that our method offers a viable alternative to heavier deep learning models and opens new avenues for future research in light field-based image segmentation.