Image stitching algorithm based on feature block and wavelet transform
Ling Wang · Jisuanji gongcheng yu sheji · 2010
To overcome shortcomings of existing intensity-based image stitching methods and consider the two factors of speed and accuracy, an image stitching algorithm based on wavelet transform and feature block is proposed. Firstly, wavelet transform is carried out on images and feature block is picked up using a simple method of edge information threshold on high-frequency signals. Then images are matched on low-frequency signals, and original images registrations are finished based on displacement of low-frequency image registration. The result of experiments, which is compared to traditional intensity-based image stitching methods, indicated that the speed of the image stitching is greatly enhanced in ensuring the accuracy of image mosaic at the same time.