Evaluation of the conspicuity of roadway environments for their environmental impact on driving performance is vital for roadway safety. Existing meters and tools for roadway measurements cannot record light and geometry data simultaneously in a high resolution. This study introduced a new method that adopted recently developed high dynamic range (HDR) photogrammetry to measure the luminance and XYZ coordinates of millions of points across a road scene with the same device-a camera, and a MatLab code for data treatment and visualization. To validate this method, the roadway environments of a straight and flat section of Jayhawk Boulevard (482.8 m long) at Lawrence, KS and a roundabout (15.3 m in diameter) at its end were measured under clear and cloudy sky in the daytime and at nighttime with dry and wet pavements. Eight HDR images of the roadway environments under different viewing conditions were generated using the HDR photogrammetric techniques and calibrated. From each HDR image, synchronous light and geometry data were extracted in Radiance and further analyzed to identify potential roadway environmental hazards using the MatLab code (http://people.ku.edu/~h717c996/research.html). The HDR photogrammetric measurement with current equipment had a margin of errors for geometry measurement that varied with the measuring distance, averagely 23.1%-27.5% for the Jayhawk Boulevard and 9.3%-16.2% for the roundabout. The accuracy of luminance measurement was proven in the literature as averagely 1.5%-10.1%. The camera-aided measurement is fast, non-contact, non-destructive, and off the road, thus, it is deemed more efficient and safer than conventional ways using meters and tools. The HDR photogrammetric techniques with current equipment still need improvements on accuracy and speed of the data treatment.

Molecular generation is an important but challenging task in drug design, as it requires optimization of chemical compound structures as well as many complex properties. Most of the existing methods use deep learning models to generate molecular representations. However, these methods are faced with the problems of generation validity and semantic information of labels. Considering these challenges, we propose a cross-adversarial learning method for molecular generation, CRAG for short, which integrates both the facticity of VAE-based methods and the diversity of GAN-based methods to further exploit the complex properties of Molecules. To be specific, an adversarially regularized encoder-decoder is used to transform molecules from simplified molecular input linear entry specification (SMILES) into discrete variables. Then, the discrete variables are trained to predict property and generate adversarial samples through projected gradient descent with corresponding labels. Our CRAG is trained using an adversarial pattern. Extensive experiments on two widely used benchmarks have demonstrated the effectiveness of our proposed method on a wide spectrum of metrics. We also utilize a novel metric named Novel/Sample to measure the overall generation effectiveness of models. Therefore, CRAG is promising for AI-based molecular design in various chemical applications.

The speckle effort resulting from the coherent imaging principle of Synthetic Aperture Radar brings serious speckle noise to SAR images, seriously reduced the image quality. This paper divides SAR image into small regions of the similar distribution characteristics, and reduce speckle noise in different regions. This algorithm has very good speckle reduction effort, meanwhile keeps edge-characteristics of the image. Experiments shows that this algorithm has better efforts than present speckle reduction methods.

No abstract

No abstract