Fundus photography is employed by medical practitioners to track the advancement of specific eye conditions. Retinal blood vessels (BV) serve as a pivotal indicator for identifying and addressing retinal ailments; hence, accurately segmenting them is imperative for a thorough retinal examination. This paper focuses on the study of the performance of different color channels on BV segmentation using the U-Net Architecture. Statistical analysis was carried out on the DRIVE dataset, and it reveals significant variations in segmentation performance across channels. The green channel was most effective and showed the best performance for BV segmentation. This can be ascribed to the enhanced contrast properties, which effectively delineate BVs from the retinal background. The red and blue channels exhibit relatively poorer performance due to over-saturation and under-saturation issues, respectively. Evaluation metrics like accuracy, sensitivity, specificity, precision, and intersection over union (IoU) define the efficacy of results. Green channel achieved accuracy, sensitivity, and specificity of 0.9714, 0.7981, and 0.9864, respectively.