Shape descriptor is a concise yet informative representation that provides 3D object with an identification as member of some category. We have developed deep shape to address challenging issues from ever-growing datasets in areas diverse engineering, medicine, and biology. Specifically, this paper, we novel techniques extract but geometrically new methods defining Eigen-shape Fisher-shape guide the training neural network. Our tends maximize inter-class margin while minimize intra-class...

In recent years, ship detection in satellite remote sensing images has become an important research topic. Most existing methods detect ships by using a rectangular bounding box but do not perform segmentation down to the pixel level. This paper proposes and method based on improved Mask R-CNN model. Our proposed can accurately segment at By adding bottom-up structure FPN of R-CNN, path between lower layers topmost layer is shortened, allowing features be more effectively utilized top layer....

Complex geometric variations of 3D models usually pose great challenges in shape matching and retrieval. In this paper, we propose a novel feature learning method to extract high-level features that are insensitive deformations shapes. Our uses discriminative deep auto-encoder learn deformation-invariant features. First, multiscale distribution is computed used as input the auto-encoder. We then impose Fisher discrimination criterion on neurons hidden layer develop Finally, outputs from...

With the recent development of deep learning on steganalysis, embedding secret information into digital images faces great challenges. In this paper, a secure steganography algorithm by using adversarial training is proposed. The architecture contain three component modules: generator, an simulator and discriminator. A generator based U-NET to translate cover image change probability To fit optimal propagate gradient, function called Tanh-simulator As for discriminator, selection-channel...

With the proliferation of digital media such as images, audio, and video, robust watermarking data hiding techniques are needed for copyright protection, copy control, annotation, authentication. While many have been proposed color grayscale not all them can be directly applied to binary text images. The difficulty lies in fact that changing pixel values a document could introduce irregularities very visually noticeable. We propose new method documents by embedding 8-connected boundary...

One of the most challenging task in face recognition is to identify people with varied poses. Namely, test faces have significantly different poses compared registered faces. In this paper, we propose a high-level feature learning scheme extract pose-invariant identity for recognition. First, build single-hidden-layer neural network sparse constraint, supervised fashion. Second, further enhance discriminative capability proposed by using multiple random as target values encoders. By...

Abstract: We demonstrate the feasibility of applying image processing techniques to analysis pavement distress due cracking. Pavement samples were obtained using a custom‐designed data acquisition system called Automatic Crack Monitor (ACM). The conteining cracks are parameters, extracted techniquesw described in this paper. crack parameters necessary measures used calculations Serviceability Index (PSI), which is by highway maintenance engineers decide whether certain section needs be...

Sensor fusion represents a robust approach to ecologically valid obstacle identification in building comprehensive electronic travel aid (ETA) for the blind and visually impaired. A stereoscopic camera system an infrared sensor with 16 independent elements is proposed be combined multi-scale convolutional neural network this framework. While object detection can depth information from stereo system, our experiments demonstrate that may inconsistent given material surfaces of specific...

Creating, manipulating, and detecting coherent electrons is at the heart of future quantum microscopy spectroscopy technologies. Leveraging specifically altering features an electron beam source low temperatures can enhance its emission properties. Here, we describe field from a monocrystalline, superconducting niobium nanotip temperature 5.9 K. The emitted energy spectrum reveals ultranarrow distribution down to 16 meV due tunable resonant tunneling via localized band states...

With the proliferation of digital media such as images, audio, and video, robust watermarking data hiding techniques are needed for copyright protection, copy control, annotation, authentication. While many have been proposed color grayscale not all them can be directly applied to binary document images. The difficulty lies in fact that changing pixel values a could introduce irregularities very visually noticeable. Over last few years, we seen growing but limited number papers proposing new...

Different from the conventional deep learning work based on an images content in computer vision, steganalysis is art to detect secret information embedded image via learning, pose challenge of detection weak invisible hidden a host thus very low signal-to-noise (SNR) case. In this paper, we propose 32- layer convolutional neural Networks (CNNs) improve efficiency preprocess and reuse features by concatenating all previous layers with same feature- map size, flow gradient. The shared...

The visually impaired are consistently faced with mobility restrictions due to the lack of truly accessible environments. Even in structured settings, people low vision may still have trouble navigating efficiently and safely hallway threshold ambiguity. Assistive technologies that currently available do not provide door door-handle object detections nor they concretely help reaching towards object. In this paper, we propose an AI-driven wearable assistive technology integrates handle...

We investigate the use of frequency domain techniques to watermark text documents. A image is essentially binary and hence contains large high-frequency components. This has several implications on obtrusiveness detection performance marking images, as illustrated by our extensive experiments. Generally more obtrusive a than pictorial image. It almost always creates 'dirty' background. 'Cleaning' background thresholding light grays white renders less but also sharply reduces detector...

The <i>Journal of Electronic Imaging</i> (JEI), copublished bimonthly with the Society for Imaging Science and Technology, publishes peer-reviewed papers that cover research applications in all areas electronic imaging science technology.

Purpose: NYU 05‐181 protocol compared the CT simulation in both supine and prone positions for 400 patients with breast cancer (200 left‐breast 200 right‐breast) to identify which setup is better at sparing heart lung involvement treatment process. The results demonstrated that all right‐breast benefited from position, while patients, 85% were treated 15% supine. Using clinical data collected this protocol, authors aimed developing an automated tool capable of identifying are without...

We report on the development and implementation of a new algorithm for extracting text in digitized color video. The operates by locating potential line segments from horizontal scan lines. Detected are expanded or combined with adjacent lines to form larger blocks, which then subject filtering refinement. Text pixels within blocks found using bi-color clustering connected-component analysis. Morphological contour smoothing morphological resolution enhancement algorithms applied detected...

With the emergence of RGB-D cameras (e.g., Kinect), sensing capability artificial intelligence systems has been dramatically increased, and as a consequence, wide range depth image-based human-machine interaction applications are proposed. In design industry, 3D model always contains abundant information, which required for manufacture. Since images can be conveniently acquired, retrieval system that return models based on image inputs assist or improve traditional product process. this...