A Smart Deep Convolutional Neural Network for Real-Time Surface Inspection

A proper detection and classification of defects in steel sheets in real time have become a requirement for manufacturing these products, largely used in many industrial sectors. However, computers used in the production line of small to medium size companies, in general, lack performance to attend real-time inspection with high processing demands. In this paper, a smart deep convolutional neural network for using in real-time surface inspection of steel rolling sheets is proposed. The architecture is based on the state-of-the-art SqueezeNet approach, which was originally developed for usage with autonomous vehicles. The main features of the proposed model are: small size and low computational burden. The model is 10 to 20 times smaller when compared to other networks designed for the same task, and more than 700 times smaller than general networks. Also, the number of floating-point operations for a prediction is about 50 times lower than the ones used for similar tasks. Despite its small size, the proposed model achieved near-perfect accuracy on a public dataset of 1800 images of six types of steel rolling defects.     

Er3+-Yb3+ ions doped fluoro-aluminosilicate glass-ceramics as a temperature-sensing material

The transparent Er³⁺-Yb³⁺-doped fluoro-aluminosilicate glass-ceramic (GC) was prepared by melt-quenching. The crystal phase, morphology, and up-conversion (UC) luminescence of as-produced GC were characterized by X-ray diffraction, scanning electron microscopy, and fluorescence spectrophotometry, respectively. The results show that BaYF₅ nanocrystals were uniformly distributed in the glass matrix of the as-produced GC. When the as-produced GC was subjected to heat treatment, the crystallinity was increased, but the crystal identity remains unchanged. Such heat-treatment doubled the intensity of the UC luminescence, and this enhancement was ascribed to the increased incorporation of both Er³⁺ and Yb³⁺ ions into the lower phonon energy environment of BaYF₅ nanocrystals. Furthermore, the heat-treated GC was stable against further crystallization, and consequently its UC luminescence was stable at the application temperature. The heat-treated GC was found to possess an outstanding temperature-sensing capability.     

Negative thermal quenching CsPbBr3 glass-ceramic based on intrinsic radiation and vacancy defect co-induced dual-emission

Halide perovskite glass-ceramic has recently moved into the center of the attention of perovskite research due to their potential for temperature sensing. However, quantum dots glass-ceramic with excellent luminescence performance still needs to be combined with rare-earth (RE) ions to accurately measure temperature. In this work, a novel non-RE doped dual-emission (460 nm and 512 nm) CsPbBr₃ quantum dots was obtained in telluride glass via the friction crystallization method, where 512 nm was derived from intrinsic luminescence of quantum dots, and 460 nm was originated from thermally induced bromine vacancy, which can be used for temperature sensing. Fluorescence intensity ratio results indicate that the relative sensitivity of dual-emission could reach 5.6 % K^?1 at 323 K. The discovery of non-RE doped CsPbBr₃ QDs glass-ceramic with negative thermal quenching uncovers a new optional sensing glass material that surpass traditional RE-doped QDs glass by their tunability and sensitivity.     

Quality assessment for color correction-based stitched images via bi-directional matching

With the deepening of social information, the panoramic image has drawn a significant interest of viewers and researchers as it can provide a very wide field of view (FoV). Since panoramic images are usually obtained by capturing images with the overlapping regions and then stitching them together, image stitching plays an important role in generating panoramic images. In order to effectively evaluate the quality of stitched images, a novel quality assessment method based on bi-directional matching is proposed for stitched images. Specifically, dense correspondences between the testing and benchmark stitched images are first established by bi-directional SIFT-flow matching. Then, color-aware, geometric-aware and structure-aware features are respectively extracted and fused via support vector regression (SVR) to obtain the final quality score. Experiments on our newly constructed database and ISIQA database demonstrate that the proposed method can achieve comparable performance compared with the conventional blind quality metrics and the quality metrics specially designed for stitched images.     

Various density light field image coding based on distortion minimization interpolation

In recent years, the light field (LF) as a new imaging modality has attracted wide interest. The large data volume of LF images poses great challenge to LF image coding, and the LF images captured by different devices show significant differences in angular domain. In this paper we propose a view prediction framework to handle LF image coding with various sampling density. All LF images are represented as view arrays. We first partition the views into reference view (RV) set and intermediate view (IV) set. The RVs are rearranged into a pseudo sequence and directly compressed by a video encoder. Other views are then predicted by the RVs. To exploit the four dimensional signal structure, we propose the linear approximation prior (LAP) to reveal the correlation among LF views and efficiently remove the LF data redundancy. Based on the LAP, a distortion minimization interpolation (DMI) method is used to predict IVs. To robustly handle the LF images with different sampling density, we propose an Iteratively Updating depth image based rendering (IU-DIBR) method to extend our DMI. Some auxiliary views are generated to cover the target region and then the DMI calculates reconstruction coefficients for the IVs. Different view partition patterns are also explored. Extensive experiments on different types LF images also valid the efficiency of the proposed method.     

Sequential alignment attention model for scene text recognition

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state-of-the-art solutions usually depend on the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. Unfortunately, there often exists severe misalignment between feature areas and text labels in real-world scenarios. To address this problem, this paper proposes a sequential alignment attention model to enhance the alignment between input images and output character sequences. In this model, an attention gated recurrent unit (AGRU) is first devised to distinguish the text and background regions, and further extract the localized features focusing on sequential text regions. Furthermore, CTC guided decoding strategy is integrated into the popular attention-based decoder, which not only helps to boost the convergence of the training but also enhances the well-aligned sequence recognition. Extensive experiments on various benchmarks, including the IIIT5k, SVT, and ICDAR datasets, show that our method substantially outperforms the state-of-the-art methods.     

Enhancing upconversion of Nd3+ through Yb3+-mediated energy cycling towards temperature sensing

Photon upconversion of lanthanides has been a powerful means to convert low-energy photons into high-energy ones. However, in contrast to the mostly investigated lanthanide ions, it has remained a challenge for the efficient upconversion of Nd³⁺ due to the deleterious concentration quenching effect. Here we report an efficient strategy to enhance the upconversion of Nd³⁺ through the Yb³⁺-mediated energy cycling in a core-shell-shell nanostructure. Both Nd³⁺ and Yb³⁺ are confined in the interlayer, and the presence of Yb³⁺ in the Nd-sublattice provides a more matched energy for the upconversion transitions occurring at the intermediate state of Nd³⁺ towards much better population at its emissive levels. Moreover, this design also minimizes the possible cross-relaxation processes at both intermediate level and the emissive levels of Nd³⁺ which are the primary factors limiting the upconversion performance for the Nd³⁺-doped materials. Such energy cycling-enhanced upconversion shows promise in temperature sensing.     

Molding,patterning and driving liquids with light

When a laser beam induces surface tension gradient at the free surface of a liquid, a weak surface depression is expected and has been observed. Here we report giant depression and rupture in “optothermocapillary fluids” under the illumination of laser and sunlight. Computational fluid dynamics models were developed to understand the surface deformation and provided desirable physical parameters of the fluid for maximum deformation. New optothermocapillary fluids were created by mixing transparent lamp oil with different candle dyes. They can be cut open by sunlight and be patterned to different shapes and sizes using an ordinary laser show projector or a common laser pointer. Laser driving and elevation of optothermocapillary fluids, in addition to the manipulation of different droplets on their surface, were demonstrated as an efficient controlling method and platform for optofluidic operations. The fundamental understanding of light-induced giant depression and creation of new optothermocapillary fluids encourage the fundamental research and applications of optofluidics.