The measure J in J value segmentation (JSEG) fails to represent the discontinuity of color, which degrades the robustness and discrimination of JSEG. An improved approach for JSEG algorithm was proposed for unsupervis...
详细信息
The measure J in J value segmentation (JSEG) fails to represent the discontinuity of color, which degrades the robustness and discrimination of JSEG. An improved approach for JSEG algorithm was proposed for unsupervised color-texture image segmentation. The texture and photometric invariant edge information were combined, which results in a discriminative measure for color-texture homogeneity. Based on the image whose pixel values are values of the new measure, region growing-merging algorithm used in JSEG was then employed to segment the image. Finally, experiments on a variety of real color images demonstrate performance improvement due to the proposed method.
In order to implement vehicle license plate (VLP) location at complex environment, a VLP fusion location method with object authenticity confirmation is presented. In this way, there are four steps. First, HSV color s...
详细信息
In order to implement vehicle license plate (VLP) location at complex environment, a VLP fusion location method with object authenticity confirmation is presented. In this way, there are four steps. First, HSV color space feature is studied and every region connected with object color is marked in one gray-level image according to multi-gray-level for multi-style. Second, gray-level vertical edge of absolute accumulation differential image (AADI) is extracted. Third, each kind of possible object region is detected by fusion feature respectively. Last, object authenticity confirmation method based on companion and compensation is designed to throw off the false-alarm and improve the location accuracy. Various actual RGB color VLP images are used to test the proposed method. The method is suitable for multi-object and multi-style location at complex environment. The experimental results proved its effectiveness.
This paper introduces a new approach, nearest convex hull (NCH), for remote sensing classification. NCH is an intuitive classification method which labels the test point as the training class whose convex hull is clos...
详细信息
This paper introduces a new approach, nearest convex hull (NCH), for remote sensing classification. NCH is an intuitive classification method which labels the test point as the training class whose convex hull is closest to it. Some attractive advantages of this learning algorithm are the robustness to noises and the scale of training samples, the straightforward way to handle multi-class tasks, and most of all the capability of processing high dimensional and nonlinear data. In our work, we deduce the NCH algorithm again basing on theories of the computational geometry, from which a simpler implementation of it is presented. Then we apply it to real-world remote problems and compare it with two other state-of-arts classifiers: K-NN and SVM. Experiments in this paper confirm the promising performance of NCH for remote sensing classification.
image quality evaluation is becoming essential in many imageprocessing problems. This paper proposes a new image quality evaluation approach based on decision fusion method of canonical correlation analysis (CCA). By...
详细信息
How to accurately predict traffic data with weak regularity is difficult for various forecasting models. In this paper, least squares support vector machines (LS-SVMs) are proposed to deal with such a problem. It is t...
详细信息
Medical imaging techniques like computed/digital radiography (CR/DR) have introduced a formidably powerful tool in medicine. image enhancement takes an important roll in the CR/DR computerized analysis process. Much e...
详细信息
Locality Preserving Projection (LPP), as a linear manifold learning algorithm, has attracted much interests in recent years. LPP considers an n1× n2image as a vector in €n1×n2space, and thus is limited by th...
详细信息
In this article, we present the projective equation of a circle in a perspective view, which naturally encodes such important geometric entities as the projected circle center, the vanishing point of the normal direct...
详细信息
In this paper, a novel subspace learning method, semi-supervised marginal discriminant analysis (SMDA), is proposed for classification. SMDA aims at maintaining the intrinsic neighborhood relations between the data po...
详细信息
We present a method for tracking deformable surfaces in 3D using a stereo rig. Different from traditional recursive tracking approaches that provide a strong prior on the pose for each new frame, the proposed method t...
详细信息
暂无评论