The problem of estimating the error probability of a given classification system is considered. Statistical properties of the empirical error count (C) and the average conditional error (R) estimators are studied. It ...
详细信息
The problem of estimating the error probability of a given classification system is considered. Statistical properties of the empirical error count (C) and the average conditional error (R) estimators are studied. It is shown that in the large sample case the R estimator is unbiased and its variance is less than that of the C estimator. In contrast to conventional methods of Bayes error estimation the unbiasedness of the R estimator for a given classifier can be obtained only at the price of an additional set of classified samples. On small test sets the R estimator may be subject to a pessimistic bias caused by the averaging phenomenon characterising the functioning of conditional error estimators.
The three-dimensional shape analysis problem is a very demanding test of shape analysis algorithms. Previous approaches to the problem have employed global features such as moments and Fourier descriptors. Global feat...
详细信息
The three-dimensional shape analysis problem is a very demanding test of shape analysis algorithms. Previous approaches to the problem have employed global features such as moments and Fourier descriptors. Global features lack the capacity for solving the partial shape recognition problem, in which only part of the unknown shape is available. Previous approaches to local shape analysis have employed structural (syntactic) methods, but these methods have so far failed to solve the three-dimensional problem. This study describes a hybrid structural/statistical local shape analysis algorithm which is applied to the three-dimensional problem.
The k-syntactic similarity approach is couched in graphical representation terms and its ability to provide global recognition capability while retaining a low time complexity is explored. One potential application do...
详细信息
The k-syntactic similarity approach is couched in graphical representation terms and its ability to provide global recognition capability while retaining a low time complexity is explored. One potential application domain, that of composite shape decomposition into approximately convex subshapes, is described. This is shown to be equivalent to finding cycles within a particular graph. The approach yields valid decompositions in many cases where additional semantic considerations are necessary for proper analysis. The permissible graph structures representing composite shapes given a reasonable set of relations are determined. Experimental results on non-ideal data are given.
This paper presents a novel background subtraction method for detecting foreground objects in dynamic scenes involving swaying trees and fluttering flags. Most methods proposed so far adjust the permissible range of t...
详细信息
This paper presents a novel background subtraction method for detecting foreground objects in dynamic scenes involving swaying trees and fluttering flags. Most methods proposed so far adjust the permissible range of the background image variations according to the training samples of background images. Thus, the detection sensitivity decreases at those pixels having wide permissible ranges. If we can narrow the ranges by analyzing input images, the detection sensitivity can be improved. For this narrowing, we employ the property that image variations at neighboring image blocks have strong correlation, also known as "cooccurrence". This approach is essentially different from chronological background image updating or morphological postprocessing. Experimental results for real images demonstrate the effectiveness of our method.
The restoration of images is an important and widely studied problem in computer vision and imageprocessing. Various image filtering strategies have been effective, but invariably make strong assumptions about the pr...
详细信息
ISBN:
(纸本)0769523722
The restoration of images is an important and widely studied problem in computer vision and imageprocessing. Various image filtering strategies have been effective, but invariably make strong assumptions about the properties of the signal and/or degradation. Therefore, these methods typically lack the generality to be easily applied to new applications or diverse image collections. This paper describes a novel unsupervised, information-theoretic, adaptive filter (UINTA) that improves the predictability of pixel intensities from their neighborhoods by decreasing the joint entropy between them. Thus UINTA automatically discovers the statistical properties of the signal and can thereby restore a wide spectrum of images and applications. This paper describes the formulation required to minimize the joint entropy measure, presents several important practical considerations in estimating image-region statistics, and then presents results on both real and synthetic data.
In dynamic radiological studies time-dependent processes, ranging from motion of the patient's own anatomical structures to the kinetics of a contrast agent or radiotracer, are recorded in a sequence of image fram...
详细信息
In dynamic radiological studies time-dependent processes, ranging from motion of the patient's own anatomical structures to the kinetics of a contrast agent or radiotracer, are recorded in a sequence of image frames. If the data are available in digital format, appreciation of regional patterns of behavior can be enhanced by digital operations on the image sequence. These operations can be considered as projecting out aspects of temporal behavior that are not readily or unambiguously perceptible to the unassisted observer. A report is presented on experiences with such functional images in three radiological modalities at the University of Wisconsin. Examples include nuclear medicine studies of the heart and liver, transmission computed tomography studies of the brain, and digital fluoroscopic studies of the heart and kidneys.
We introduce a hierarchical color segmentation technique that combine the advantages of local (simplicity and quickness) and global region growing methods (robustness, accuracy, avoidance of chaining mismatches). The ...
详细信息
ISBN:
(纸本)081863880X
We introduce a hierarchical color segmentation technique that combine the advantages of local (simplicity and quickness) and global region growing methods (robustness, accuracy, avoidance of chaining mismatches). The method is implemented for a traffic sign recognition system.
The problem of automatic registration of deformed images is addressed. It describes an iterative technique for gradually updating the local registration of two images based on a dynamic cooperative model. the method i...
详细信息
The problem of automatic registration of deformed images is addressed. It describes an iterative technique for gradually updating the local registration of two images based on a dynamic cooperative model. the method is cooperative in the sense that a feature at one location in an image influences decisions made at other locations. Initially, when registration is expected to be poor and feature measures unreliable, cooperative interaction is strong. It is progressively weakened with each iteration to permit matching of fine details. For a physical analogy, consider an elastic picture whose stiffness decreases with each iteration, and which is deformed by forces arising from similar features in another picture. Examples are shown for both dot patterns and gray-scale pictures.
A new class of texture features based on the joint occurrences of gray levels at points defined relative to edge maxima are introduced. These features are compared with previous types of cooccurrence-based features, a...
详细信息
A new class of texture features based on the joint occurrences of gray levels at points defined relative to edge maxima are introduced. These features are compared with previous types of cooccurrence-based features, and experimental results are presented indicating that the new features should be useful for texture classification. In the second part, three simple methods of extracting texture primitives are compared. It appears that the simplest of these, thresholding at a fixed percentile, yields primitives that are quite effective in texture discrimination.
The growing use and analysis of images by computer presents many problems in managing images and image information. Solutions to these problems can be specific, with file structures to represent the needed information...
详细信息
The growing use and analysis of images by computer presents many problems in managing images and image information. Solutions to these problems can be specific, with file structures to represent the needed information, or general, such as a general Data Base Management System. A presentation is made of a compromise solution, a Raster image File Format (RIFF), that uses image file headers for the storage of specific image format information, as well as very general Name-Value pair information. This latter structure enables varying kinds of information to be represented, including links to other kinds of image data structures as the need arises.
暂无评论