The problem of finding approximate matches of pieces of shapes to parts of larger shapes is investigated. The shapes are represented by hierarchical polygonal approximations. Initially, figures of merit are assigned t...
详细信息
The problem of finding approximate matches of pieces of shapes to parts of larger shapes is investigated. The shapes are represented by hierarchical polygonal approximations. Initially, figures of merit are assigned to the matches between pairs of angles on the two shapes. Relaxation methods are then used to find acceptable combinations of these matches. This approach was tested on a data base consisting of digitized coastlines in various map projections. In nearly all cases, all matches except the correct one were eliminated by the relaxation processes.
This paper reviews recent developments in the use of iterative (or ″relaxation″ ) methods in image analysis. Applications of these methods include histogram modification, noise cleaning, edge and curve detection, th...
详细信息
This paper reviews recent developments in the use of iterative (or ″relaxation″ ) methods in image analysis. Applications of these methods include histogram modification, noise cleaning, edge and curve detection, thinning, angle detection, template matching, and region labelling. These applications are briefly described, and references are given to papers and reports in which more detailed discussions and examples can be found.
An algorithm for constructing regions from their list of boundary points is presented. Both adjacency and containment of regions can be handled. The construction is based on a small set of boundary types which determi...
详细信息
An algorithm for constructing regions from their list of boundary points is presented. Both adjacency and containment of regions can be handled. The construction is based on a small set of boundary types which determine the labelling of interior points. All regions are processed together so that time and storage requirements are minimized.
An application is presented of symmetric axis geometry to shape classification and description. Shapes are segmented into simplified segments, in which a sequential string of features are derived. These are based on b...
详细信息
An application is presented of symmetric axis geometry to shape classification and description. Shapes are segmented into simplified segments, in which a sequential string of features are derived. These are based on both width and axis properties, and two linear combinations of them, as well as their derivatives. A weighting measure is developed which evaluates the importance of these shape descriptors. Consequently, small perturbations are recognizable despite their generation of sizable symmetric axis arcs. A variety of other segment measures for shape are introduced.
A method is suggested for transforming a polygonal object into a binary relation whose clusters represent intuitively pleasing simple parts of the object. The graph theoretic clustering method determines near-cliques,...
详细信息
A method is suggested for transforming a polygonal object into a binary relation whose clusters represent intuitively pleasing simple parts of the object. The graph theoretic clustering method determines near-cliques, which are local regions of high compactness, and then forms clusters by merging those near-cliques having high enough overlap. Using this procedure on the shape of a colon as seen in an X-ray, it was found that for many colon examples, the parts determined by the clustering scheme corresponded well to the decomposition a human might make.
Three transformations used in the analysis of tricolor natural scenes are analyzed: saturation, hue, and normalized color. All have nonremovable singularities, near which they are highly unstable. Given digital input,...
详细信息
Three transformations used in the analysis of tricolor natural scenes are analyzed: saturation, hue, and normalized color. All have nonremovable singularities, near which they are highly unstable. Given digital input, the distribution of their transformed values is highly nonuniform, characterized by spurious modes and gaps. These effects are quantified and illustrated. image segmentation techniques of edge detection, region growing, clustering, and region splitting are affected arbitrarily badly by such problems. Some stratagems are illustrated that help minimize the bad behavior.
A class of error-correcting tree automata is studied. Syntax errors on trees are defined in terms of five transformations - substitution, stretch, split, branch, and deletion. The distance between two trees is defined...
详细信息
A class of error-correcting tree automata is studied. Syntax errors on trees are defined in terms of five transformations - substitution, stretch, split, branch, and deletion. The distance between two trees is defined as the minimum cost sequence of error transformations needed to derive one from the other. This definition provides a similarity measure for syntactic patterns. For a given noisy input tree and a free language, the proposed generalized error-correcting tree automaton (GECTA) will produce a tree in the language using the minimum-distance criterion.
Two sets of features - perceptual features and co-occurrence features - were used in classification experiments. The ability of each set was verified in a pilot experiment for classification of 16 typical natural text...
详细信息
Two sets of features - perceptual features and co-occurrence features - were used in classification experiments. The ability of each set was verified in a pilot experiment for classification of 16 typical natural texture patterns. In the main study several practical problems concerning usage of textural features are considered by using pairs of multispectral images sensed at different altitudes on the same day. It was verified that adequate features could be selected by visual estimation. Finally, the effects of spectral bands and resolution are mentioned. This work was applied to agricultural land use classification.
This paper concerns methods for indexing areas in two-dimensional array data. A method for naming subpictures from raster scan image data is presented with notation that eases their subsequent storage access. Equation...
详细信息
This paper concerns methods for indexing areas in two-dimensional array data. A method for naming subpictures from raster scan image data is presented with notation that eases their subsequent storage access. Equations are given for converting each subpicture name into a storage-location pointer. A function ″NUMERIC″ to take an area name and return its unique relative offset on auxiliary storage is presented. Algorithms enabling efficient retrieval of subpicture areas from sequential and direct access files are presented.
Sixty-three papers were presented at the conference held at Rensselaer Polytechnic Institute in June, 1977. Areas of coverage were pattern and image analysis, image models, medical applications, speech pattern recogni...
详细信息
Sixty-three papers were presented at the conference held at Rensselaer Polytechnic Institute in June, 1977. Areas of coverage were pattern and image analysis, image models, medical applications, speech patternrecognition, syntactic patternrecognition, industrial applications, shape analysis, scene analysis, multi-imageprocessing, statistical patternrecognition, and character and signature recognition.
暂无评论