One critical aspect neural network designers face today is choosing an appropriate size for a given application. Network involves in the case of layered architectures, number layers network, nodes per layer, and connections. Roughly speaking, implements nonlinear mapping u=G(x). The function G established during training phase where learns to correctly associate input patterns x output u. Given set examples (x, u), there probably infinite different networks that can learn map into question...

Robust and reliable vehicle detection from images acquired by a moving (i.e., on-road detection) is an important problem with applications to driver assistance systems autonomous, self-guided vehicles. The focus of this work on the issues feature extraction classification for rear-view detection. Specifically, treating as two-class problem, we have investigated several different methods such principal component analysis, wavelets, Gabor filters. To evaluate extracted features, experimented...

Robust and reliable vehicle detection from images acquired by a moving is an important problem with numerous applications including driver assistance systems self-guided vehicles. Our focus in this paper on improving the performance of on-road employing set Gabor filters specifically optimized for task detection. This essentially kind feature selection, critical issue when designing any pattern classification system. Specifically, we propose systematic general evolutionary filter...

On-road vehicle detection is an important problem with application to driver assistance systems and autonomous, self-guided vehicles. The focus of this paper on the feature extraction classification for rear-view detection. Specifically, we propose using Gabor filters support vector machines (SVM) provide a mechanism obtaining some degree invariance intensity due global illumination, selectivity in scale, orientation. Basically, they are orientation scale tunable edge line detectors....

We consider the problem of gender classification from frontal facial images using genetic feature subset selection. argue that selection is an important issue in and demonstrate Genetic Algorithms (GA) can select good subsets features (i.e., encode mostly information), reducing error. First, Principal Component Analysis (PCA) used to represent each image as a vector eigen-features) low-dimensional space. (GAs) are then employed representation by disregarding certain eigenvectors do not seem...

There has been a strong trend lately in face processing research away from geometric models towards appearance models. Appearance-based methods employ dimensionality reduction to represent faces more compactly low-dimensional subspace which is found by optimizing certain criteria. The most popular appearance-based method the of eigenfaces that uses Principal Component Analysis (PCA) spanned eigenvectors covariance matrix data corresponding largest eigenvalues (i.e., directions maximum...

Understanding intent is an important aspect of communication among people and essential component the human cognitive system. This capability particularly relevant for situations that involve collaboration agents or detection can pose a threat. In this paper, we propose approach allows robot to detect intentions others based on experience acquired through its own sensory-motor capabilities, then using while taking perspective agent whose should be recognized. Our method uses novel...

Vehicle detection from aerial images is becoming an increasingly important research topic in surveillance, traffic monitoring and military applications. The system described this paper focuses on vehicle rural environments its applications to oil gas pipeline threat detection. Automatic by unmanned vehicles (UAV) will replace current patrol services that rely pilot visual inspection of the low altitude high risk flights are often restricted weather conditions. Our compares a set feature...

Presents an indexing-based approach to fingerprint identification. Central the proposed is idea of associating a unique topological structure with minutiae using Delaunay triangulation. This allows for choosing more "meaningful" groups (i.e., triangles) during indexing, preserves index selectivity, reduces memory requirements without sacrificing recognition accuracy, and improves time. Specifically, assuming N per on average, considers only O(N) triangles indexing or recognition. compares...

Considerable progress has been made in face recognition research over the last decade especially with development of powerful models appearance (i.e., eigenfaces). Despite variety approaches and tools studied, however, is not accurate or robust enough to be deployed uncontrolled environments. Recently, a number studies have shown that infrared (IR) imagery offers promising alternative visible due its relative insensitive illumination changes. However, IR other limitations including it opaque...

The authenticity of a digital image suffers from severe threats due to the rise powerful editing tools that easily alter contents without leaving any visible traces such changes. In this paper, novel passive splicing forgery detection scheme based on Local Binary Pattern (LBP) and Discrete Cosine Transform (DCT) is proposed. First, chrominance component input divided into overlapping blocks. Then, for each block, LBP calculated transformed frequency domain using 2D DCT. Finally, standard...

The rising severity and frequency of wildfires in recent years the United States have raised numerous concerns regarding improvement wildfire emergency response management decision-making systems, which require operational high temporal spatial resolution monitoring capabilities. Satellites are one tools that can be used for monitoring. However, none currently available satellite systems provide both resolution. For example, GOES-17 geostationary fire products (1–5 min) but low (≥2 km),...

This paper presents an in-vehicle real-time monocular precrash vehicle detection system. The system acquires grey level images through a forward facing low light camera and achieves average rate of 10Hz. algorithm consists two main steps: multi-scale driven hypothesis generation appearance-based verification. In the step, possible image locations where vehicles might be present are hypothesized. step uses techniques to speed up but also improve robustness by making performance less sensitive...