Progressive edge-sensing dynamic scene deblurring

Deblurring images of dynamic scenes is a challenging task because blurring occurs due to a combination of many factors. In recent years, the use of multi-scale pyramid methods to recover high-resolution sharp images has been extensively studied. We have made improvements to the lack of detail recovery in the cascade structure through a network using progressive integration of data streams. Our new multi-scale structure and edge feature perception design deals with changes in blurring at different spatial scales and enhances the sensitivity of the network to blurred edges. The coarse-to-fine architecture restores the image structure, first performing global adjustments, and then performing local refinement. In this way, not only is global correlation considered, but also residual information is used to significantly improve image restoration and enhance texture details. Experimental results show quantitative and qualitative improvements over existing methods.

If You Are Old, Videos Look Slow. The Paradoxical Effect of Age-Related Motor Decline on the Kinematic Interpretation of Visual Scenes

Perception and action are tightly coupled. However, there is still little recognition of how individual motor constraints impact perception in everyday life. Here we asked whether and how the motor slowing that accompanies aging influences the sense of visual speed. Ninety-four participants aged between 18 and 90 judged the natural speed of video clips reproducing real human or physical motion (SoS, Sense-of-Speed adjustment task). They also performed a finger tapping task and a visual search task, which estimated their motor speed and visuospatial attention speed, respectively. Remarkably, aged people judged videos to be too slow (speed underestimation), as compared to younger people: the Point of Subjective Equality (PSE), which estimated the speed bias in the SoS task, was +4% in young adults (<40), +12% in old adults (40–70) and +16% in elders. On average, PSE increased with age at a rate of 0.2% per year, with perceptual precision, adjustment rate, and completion time progressively worsening. Crucially, low motor speed, but not low attentional speed, turned out to be the key predictor of video speed underestimation. These findings suggest the existence of a counterintuitive compensatory coupling between action and perception in judging dynamic scenes, an effect that becomes particularly germane during aging.

Fantastic Fauna in a Global Perspective: understanding composites in early Eurasian Antiquity

Ancient tombs and hoards across the Eurasian steppe call for a thorough revision of art-historical categories associated with pastoral societies from Mongolia to Crimea. This study focuses on one such category. “Animal style” is an umbrella term traditionally used to categorise portable precious metalwork ornamented with dynamic scenes of vigorous animal fights and entwined zoomorphic designs. With its emphasis on irregular animal anatomies and deeply rooted in a “pars-pro-toto” mode of expression, steppe imagery of fantastic fauna presents a useful case study in broader investigations of composites in the ancient world and their diffusion across cultural spheres. This study views beasts through a binary lens, the structured monsters of Greco-Roman thinkers and the organic composites of nomadic steppe artisans. In the Western canon, “composites” existed within a politically-manufactured framework of governable “otherness”, in which fantastic fauna conveys a certain tension with the exotic, unknown and uncontrollable East. Meanwhile, in the visual rhetoric of steppe artisans, monsters represented a tension with the (cyclical) shifts occurring in one's biota rather than the tumultuous events in one's constructed environment. This paper explores how the contrasting steppe pastoralist and sedentary imperial world-views came to define the various functions and meanings of “composites” in Eurasian Antiquity.

OC-SLAM: Steadily Tracking and Mapping in Dynamic Environments

Visual Simultaneous Localization and Mapping (SLAM) system is mainly used in real-time localization and mapping tasks of robots in various complex environments, while traditional monocular vision algorithms are struggling to cope with weak texture and dynamic scenes. To solve these problems, this work presents an object detection and clustering assisted SLAM algorithm (OC-SLAM), which adopts a faster object detection algorithm to add semantic information to the image and conducts geometrical constraint on the dynamic keypoints in the prediction box to optimize the camera pose. It also uses RGB-D camera to perform dense point cloud reconstruction with the dynamic objects rejected, and facilitates European clustering of dense point clouds to jointly eliminate dynamic features combining with object detection algorithm. Experiments in the TUM dataset indicate that OC-SLAM enhances the localization accuracy of the SLAM system in the dynamic environments compared with original algorithm and it has shown impressive performance in the localizition and can build a more precise dense point cloud map in dynamic scenes.

Point-based Acoustic Scattering for Interactive Sound Propagation via Surface Encoding

We present a novel geometric deep learning method to compute the acoustic scattering properties of geometric objects. Our learning algorithm uses a point cloud representation of objects to compute the scattering properties and integrates them with ray tracing for interactive sound propagation in dynamic scenes. We use discrete Laplacian-based surface encoders and approximate the neighborhood of each point using a shared multi-layer perceptron. We show that our formulation is permutation invariant and present a neural network that computes the scattering function using spherical harmonics. Our approach can handle objects with arbitrary topologies and deforming models, and takes less than 1ms per object on a commodity GPU. We have analyzed the accuracy and perform validation on thousands of unseen 3D objects and highlight the benefits over other point-based geometric deep learning methods. To the best of our knowledge, this is the first real-time learning algorithm that can approximate the acoustic scattering properties of arbitrary objects with high accuracy.