We propose a text scanner which detects wide text strings in a sequence of scene images. For scene text detection, we use a multiple-CAMShift algorithm on a text probability image produced by a multi-layer perceptron....
详细信息
ISBN:
(纸本)076951695X
We propose a text scanner which detects wide text strings in a sequence of scene images. For scene text detection, we use a multiple-CAMShift algorithm on a text probability image produced by a multi-layer perceptron. To provide enhanced resolution of the extracted text images, we perform the text detection process after generating a mosaic image in a fast and robust image registration method.
Four parameters, φ (electronegativity), nws1/3 (valence electron density in Wagner-Seitz cell),R (Pauling's metallic radius) and Z (number of valence electrons in atom), and the patternrecognition methods were u...
详细信息
Four parameters, φ (electronegativity), nws1/3 (valence electron density in Wagner-Seitz cell),R (Pauling's metallic radius) and Z (number of valence electrons in atom), and the patternrecognition methods were used to investigate the regularities of formation of ternary intermetallic compounds between three transition elements. The obtained mathematical model expressed by some inequalities can be used as a criterion of ternary compound formation in "unknown" phase diagrams of alloy systems.
The patternrecognition methods and a four-parameter model, based on extended Miede-ma’s cellular model of alloy phases, are used to study the regularities of formation of ternary compounds between two transition ele...
详细信息
The patternrecognition methods and a four-parameter model, based on extended Miede-ma’s cellular model of alloy phases, are used to study the regularities of formation of ternary compounds between two transition elements (T,T ’) and one non-transition element (N) (T-T ’-N system) . The criterion of formation can be expressed as some functions of φ (electronegativity), n1/3ws (valence electron density in Wagner-Seitz cell), R (Pauling’s metallic radii) and Z (number of valence electrons in atom).
In this paper we propose a novel method for the construction of invariant textural features for grey scale images. The textural features are based on an averaging over the 2D Euclidean transformation group with relati...
详细信息
In many branches of industry, piled box-like objects have to be recognized, grasped and transferred. Unfortunately, existing systems only deal with the most simple configurations (i.e. neatly placed boxes) effectively...
详细信息
In this paper we present SIMBA, a content based image retrieval system performing queries based on image appearance. We consider absolute object positions irrelevant for image similarity here and therefore propose to ...
详细信息
Graph pyramids allow to combine pruning of skeletons with a concept known from the representation of line images, i.e. generalization of paths without branchings by single edges. Pruning will enable further generaliza...
详细信息
Discusses mosaicking and warping for forward-motion (or forward-moving) images (FMI). In FMI, the field of view of one frame is wider than that of a later one; however, the resolution of the later frame is higher than...
详细信息
ISBN:
(纸本)0769510078
Discusses mosaicking and warping for forward-motion (or forward-moving) images (FMI). In FMI, the field of view of one frame is wider than that of a later one; however, the resolution of the later frame is higher than that of previous one. Warp-function algorithms based respectively on a mesh warp and a triangle grid are presented, and a multi-resolution mosaic based on the wavelet transform is given. In-between images could be warped from the single mosaic image. An experiment with a real scene shows that the mesh warp algorithm is more time-consuming than that of triangle grid, but the image quality of the latter is better than that of the former. It also shows that the multi-resolution mosaic image is smooth and natural, eliminating the impact of the seams.
In this paper, we discuss two kinds of VQ algorithms. One is based on minimizing the total variance and the other is based on minimizing the maximum deviation. The algorithms of the first kind better reflect the overa...
详细信息
In this paper, we discuss two kinds of VQ algorithms. One is based on minimizing the total variance and the other is based on minimizing the maximum deviation. The algorithms of the first kind better reflect the overall fit, but may discount large, but highly localized, deviations. Those of the second kind provide absolute distance bounds that are a useful error guarantee, but may be overly sensitive to any noise that might be present in the original models. A new algorithm, combining the two criteria, is presented in this paper. It not only improves the total variance, but also provides a useful maximum error guarantee. The experiments indicate the new quantizer is a better choice in some practical operations.
In the paper, a new algorithm using a competitive Hopfield neural network for color quantization is proposed, and it completes two tasks of color palette design and pixels color mapping based on the pixel color value ...
详细信息
ISBN:
(纸本)0780370449
In the paper, a new algorithm using a competitive Hopfield neural network for color quantization is proposed, and it completes two tasks of color palette design and pixels color mapping based on the pixel color value distribution. Results of the experiment show that the approach has the advantages of fast convergent speed and high quantization.
暂无评论