Image Stitch Algorithm Based on SIFT and Wavelet Transform

An algorithm of image stitch based on scale invariance feature transform(SIFT) and wavelet transform is proposed to improve the stitch effect of significantly different luminance images. Features are first detected between two images using SIFT algorithm. After identified locations and scales, features are described by a 128-dimensional feature vector. Feature is matched based on the nearest neighbor method in order to identify overlap regions. Images are blended by wavelet transform. Experimental results showed that the presented method is effective for significant different luminance images adopted from outdoor environments.