Zhongyang Ren
I am a MPhil student at NAIL@HNU (Neuromorphic Automation and Intelligence Lab at Hunan University), where I work on Event-based Robotic Vision. Before that, I did my Bachelors at School of Information Science and Technology, Shandong Normal University.
CV / Google Scholar / Email / LinkedIn
News
Research
Recovering the camera motion and scene geometry from visual data is a fundamental problem in the field of computer vision. Its success in standard vision is attributed to the maturity of feature extraction, data association and multi-view geometry. The recent emergence of neuromorphic event-based cameras places great demands on approaches that use raw event data as input to solve this fundamental problem. Existing state-of-the-art solutions typically infer implicitly data association by iteratively reversing the event data generation process. However, the nonlinear nature of these methods limits their applicability in real-time tasks, and the constant-motion assumption leads to unstable results under agile motion. To this end, we rethink the problem formulation in a way that aligns better with the differential working principle of event cameras. We show that the event-based normal flow can be used, via the proposed geometric error term, as an alternative to the full flow in solving a family of geometric problems that involve instantaneous first-order kinematics and scene geometry. Furthermore, we develop a fast linear solver and a continuous-time nonlinear solver on top of the proposed geometric error term. Experiments on both synthetic and real data show the superiority of our linear solver in terms of accuracy and efficiency, and indicate its complementary feature as an initialization method for existing nonlinear solvers. Besides, our continuous-time non-linear solver exhibits exceptional capability in accommodating sudden variations in motion since it does not rely on the constant-motion assumption.
Hyperspectral image is an image acquired by imaging a specified target simultaneously on many spectral bands, and has been widely used in atmospheric and environmental remote sensing, medical imaging, geological exploration and other fields. In order to improve the spatial resolution of hyperspectral images, it is often necessary to fuse hyperspectral images with multispectral images with higher spatial resolution. Combining traditional algorithms and convolutional neural networks, this paper makes full use of the prior knowledge inside and outside the image to propose a new fusion method. Specifically, this method takes advantage of the local low rank characteristics inside the image, first using superpixels to segment the image into blocks, and then using the kernel norm as a low-rank constraint to learn the internal a priori; and then training the convolutional neural network to learn the external a priori of the image on the more easily obtained grayscale image, which is applied to the fusion process of hyperspectral images. By experimenting and comparing on actual hyperspectral data, the proposed method has achieved excellent results.