Two-Dimensional Panoramic Image Stitching using PCA-SIFT Approach

Image stitching is a prominent field within computer vision and graphics, widely integrated into tools like cell phones and image processing APIs. This paper delves into image stitching using Python and OpenCV, aiming to merge images with shared regions into panoramic scenes. The technique merges photographs from different cameras to create seamless panoramas. The approach relies on invariant features, specifically the PCA-SIFT algorithm, which uses Scale Invariant Feature Transform (SIFT) for detecting and describing local image features. SIFT identifies significant points, extracting position, scale, and rotation invariants for feature descriptors. This method benefits both general image stitching and stitching based on local features. The Python-implemented PCA-SIFT outperforms standard SIFT in key point matching accuracy, leading to better retrieval results. Additionally, PCA-SIFT demonstrates faster matching performance. Here, the research objective is to enhance image stitching quality by using the PCA-SIFT algorithm, addressing the limitations of standard SIFT. The projected achievements include extracting key points using the SIFT algorithm, matching them with Euclidean distance, and comparing stitching quality between SIFT and PCA-SIFT.