Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images
نویسندگان
چکیده
Traditional feature-based image stitching technologies rely heavily on feature detection quality, often failing to stitch images with few features or low resolution. The learning-based solutions are rarely studied due the lack of labeled data, making supervised methods unreliable. To address above limitations, we propose an unsupervised deep framework consisting two stages: coarse alignment and reconstruction. In first stage, design ablation-based loss constrain homography network, which is more suitable for large-baseline scenes. Moreover, a transformer layer introduced warp input in stitching-domain space. second motivated by insight that misalignments pixel-level can be eliminated certain extent feature-level, reconstruction network eliminate artifacts from pixels. Specifically, implemented low-resolution deformation branch high-resolution refined branch, learning rules enhancing resolution simultaneously. establish evaluation benchmark train framework, comprehensive real-world dataset presented released. Extensive experiments well demonstrate superiority our method over other state-of-the-art solutions. Even compared solutions, quality still preferred users.
منابع مشابه
Image Stitching of the Computed Radiology images Using a Pixel-Based Approach
In this paper, a method for automatic stitching of radiology images based on pixel features has been presented. In this method, according to the smooth texture of radiological images and in order to increase the number of the extracted features after quality enhancement of initial radiology images, 45 degree isotropic mask is applied to each radiology image to observe the image details. After t...
متن کاملDeep Unsupervised Domain Adaptation for Image Classification via Low Rank Representation Learning
Domain adaptation is a powerful technique given a wide amount of labeled data from similar attributes in different domains. In real-world applications, there is a huge number of data but almost more of them are unlabeled. It is effective in image classification where it is expensive and time-consuming to obtain adequate label data. We propose a novel method named DALRRL, which consists of deep ...
متن کاملUnsupervised Deep Belief Features for Speech Translation
We present a novel formalism for introducing deep belief features to Hierarchical Machine Translation Model. The deep features are generated by unsupervised training of a deep belief network built with stacked sets of Restricted Boltzmann Machines. We show that our new deep feature based hierarchical model is better than the baseline hierarchical model with gains for two different languages pai...
متن کاملSeamless Stitching of Images
A method for seamless stitching of images with photometric inconsistencies in the overlapping region is presented. The method is based on generating a set of stitched gradients from the gradients of the input images and then reconstructing the mosaic image using the Haar wavelet integration technique of [11], [12]. This O(N) reconstruction technique is based on obtaining the Haar wavelet decomp...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE transactions on image processing
سال: 2021
ISSN: ['1057-7149', '1941-0042']
DOI: https://doi.org/10.1109/tip.2021.3092828