Discriminative feature encoding for intrinsic image decomposition

Intrinsic image decomposition is an important and long-standing computer vision problem.Given an input image,recovering the physical scene properties is ill-posed.Several physically motivated priors have been used to restrict the solution space of the optimization problem for intrinsic image decomposition.This work takes advantage of deep learning,and shows that it can solve this challenging computer vision problem with high efficiency.The focus lies in the feature encoding phase to extract discriminative features for different intrinsic layers from an input image.To achieve this goal,we explore the distinctive characteristics of different intrinsic components in the high-dimensional feature embedding space.We define feature distribution divergence to efficiently separate the feature vectors of different intrinsic components.The feature distributions are also constrained to fit the real ones through a feature distribution consistency.In addition,a data refinement approach is provided to remove data inconsistency from the Sintel dataset,making it more suitable for intrinsic image decomposition.Our method is also extended to intrinsic video decomposition based on pixel-wise correspondences between adjacent frames.Experimental results indicate that our proposed network structure can outperform the existing state-of-the-art.

ACCURATELY DETERMINING PROPAGATION VELOCITY OF CORTICAL SPREADING DEPRESSION IN RATS BY OPTICAL INTRINSIC SIGNAL IMAGING

Cortical spreading depression(CSD)is a wave of neuronal and glial depolarization that propagates across the cortex at a rate of 2–5mm/min accompanied by reversible electroencephalogram(EEG)suppression,a negative shift of direct current(DC)potential,and change of optical intrinsic signals(OIS).Propagation velocity of CSD is an important parameter used to study this phenomenon.It is commonly determined in an electrophysiological way that measures the time required for a CSD wave to pass along two electrodes.Since the electrophysiology technique fails to reveal the spreading pattern of CSD,velocity calculated in this manner might be inaccurate.In this study,we combined the electrophysiological recording and OIS imaging(OISI)for detecting changes in DC potential and OIS during CSD simultaneously.An optical method based on OISI to determine the CSD velocity,which is measured by generating a series of regions of interest(ROI)perpendicular to the advancing wavefront along propagation direction of CSD at different time points and then dividing by the distance between ROIs over time,is presented.Comparison of the accuracy of the two approaches in determining the CSD velocity is made as well.The average rate of 33 CSDs is 3.52±0.87mm/min by use of the optical method and 4.36±1.65mm/min by use of the electrophysiological method.Because of the information about spreading pattern of CSD provided optically,the velocity determined by OISI is of smaller deviation and higher accuracy.

RADepthNet:Reflectance-aware monocular depth estimation

Background Monocular depth estimation aims to predict a dense depth map from a single RGB image,and has important applications in 3D reconstruction,automatic driving,and augmented reality.However,existing methods directly feed the original RGB image into the model to extract depth features without avoiding the interference of depth-irrelevant information on depth-estimation accuracy,which leads to inferior performance.Methods To remove the influence of depth-irrelevant information and improve the depth-prediction accuracy,we propose RADepthNet,a novel reflectance-guided network that fuses boundary features.Specifically,our method predicts depth maps using the following three steps:(1)Intrinsic Image Decomposition.We propose a reflectance extraction module consisting of an encoder-decoder structure to extract the depth-related reflectance.Through an ablation study,we demonstrate that the module can reduce the influence of illumination on depth estimation.(2)Boundary Detection.A boundary extraction module,consisting of an encoder,refinement block,and upsample block,was proposed to better predict the depth at object boundaries utilizing gradient constraints.(3)Depth Prediction Module.We use an encoder different from(2)to obtain depth features from the reflectance map and fuse boundary features to predict depth.In addition,we proposed FIFADataset,a depth-estimation dataset applied in soccer scenarios.Results Extensive experiments on a public dataset and our proposed FIFADataset show that our method achieves state-of-the-art performance.

Separation of cortical arteries and veins in optical neurovascular imaging

Separation of arteries and veins in the cerebral cortex is of significant importance in the studies of cortical hemodynamics,such as the changes of cerebral blood flow,perfusion or oxygen con-centration in arteries and veins under different pathological and physiological conditions.Yet the cerebral vessel segmentation and vessel-type separation are challenging due to the complexity of cortical vessel characteristics and low spatial signal-to-noise ratio.In this work,we presented an effective full-field method to differentiate arteries and veins in cerebral cortex using dual-modal optical imaging technology including laser speckle imaging(LSI)and optical intrinsic signals(OIS)imaging.The raw contrast images were acquired by LSI and processed with enhanced laser speckle contrast analysis(eLASCA),algorithm.The vascular pattern was extracted and seg-mented using region growing algorithm from the eLASCA-based LSI.Meanwhile,OIS imageswere acquired altermatively with 630 and 870 nm to obtain an oxy hemoglobin concentration mapover cerebral cortex.Then the separation of arteries and veins was accomplished by Otsuthreshold segmentation algorithm based on the OIS information and segmentation of LSI.Finally,the segmentation and separation performances were assessed using area overlap measure(AOM).The segmentation and separation of cerebral vessels in cortical optical imaging have great potential applications in full-field cerebral hemodynamics monitoring and pathological study of cerebral vascular diseases,as well as in clinical intraoperative monitoring.

EFFECTS OF NAAG AND MPEP ON RAT CORTICAL SPREADING DEPRESSION

Cortical spreading depression(CSD)is a pathophysiological phenomenon.There are sufficient evidences to prove that CSD plays an important role in some neurological disorders.However,exact mechanisms of its initiation and propagation are still unclear.Previous studies showed that glutamate receptors could be concerned with CSD,but those studies were mostly performed oriented to ionotropic glutamate receptors(iGluRs).There is relatively little report about effects of metabotropic glutamate receptors(mGluRs)on CSD.Here,we applied optical intrinsic signal imaging(OISI)combined with direct current(DC)potential recording to examine influences of some mGluRs antagonist(or agonist)on CSD propagation in rat’s brain,to indirectly validate actions of some mGluRs on CSD.We found that N-acetyl-l-aspartyl-l-glutamate(NAAG,an agonist at mGluR3)inhibited the propagation of CSD,and the inhibition was gradually developed with time.However,6-methyl-2-phenylethynyl-pyridine(MPEP,an antagonist of mGluR5)did not produce any significant alterations with the CSD propagation.Our findings suggest that mGluR3 could play an important role in the CSD propagation,but the activity of mGluR5 was comparatively weak.These findings can help to understand the propagation mechanism of CSD,and consider the therapy of some neurological diseases involved with CSD.

Interactive lighting editing system for single indoor low-light scene images with corresponding depth maps

We propose a novel interactive lighting editing system for lighting a single indoor RGB image based on spherical harmonic lighting.It allows users to intuitively edit illumination and relight the complicated low-light indoor scene.Our method not only achieves plausible global relighting but also enhances the local details of the complicated scene according to the spatially-varying spherical harmonic lighting,which only requires a single RGB image along with a corresponding depth map.To this end,we first present a joint optimization algorithm,which is based on the geometric optimization of the depth map and intrinsic image decomposition avoiding texture-copy,for refining the depth map and obtaining the shading map.Then we propose a lighting estimation method based on spherical harmonic lighting,which not only achieves the global illumination estimation of the scene,but also further enhances local details of the complicated scene.Finally,we use a simple and intuitive interactive method to edit the environment lighting map to adjust lighting and relight the scene.Through extensive experimental results,we demonstrate that our proposed approach is simple and intuitive for relighting the low-light indoor scene,and achieve state-of-the-art results.