<p>Ensuring geometric consistency and alignment accuracy is critical in image stitching, especially when dealing with complex scenes in sequences of images. However, most existing methods struggle with perspective distortion (PD) and changes in viewpoint across image sequences, resulting in unreliable correspondences, misalignments, and artifacts in the final image mosaic. To address these limitations, we propose an Adaptive Feature Weighting (AFW) approach that improves image stitching quality by assigning greater importance to key matches based on their reliability. Our method integrates multiple feature detectors and descriptors, applying an adaptive weighting scheme guided by quality metrics to enhance homography estimation. Furthermore, we incorporate RANSAC with dynamic distance thresholding to iteratively refine correspondence selection, reducing the influence of outliers and improving overall alignment. Additionally, we introduce a hybrid feature detector and descriptor selection mechanism based on inlier ratio analysis and matching averages to identify the optimal combination of feature detectors and descriptors for stitching in both pairwise and multi-image scenarios. The proposed method is rigorously evaluated through extensive quantitative and qualitative experiments, demonstrating superior performance over state-of-the-art techniques in maintaining geometric consistency and preserving scene integrity in both aerial and ground-based images. The quantitative evaluation highlights improvements in matching accuracy and reliable correspondences, while the qualitative assessment showcases significant reductions in visual artifacts and distortions.</p>