Inhaltsangabe
Excerpt from Automatic Model Builder for Object Recognition
Extracts suitable information from observations. Any observations that yield reliable 3d coordinates can be used as the basis for an automatic model builder. Among the many features that may be used are lines significant to object shape, feature points such as vertices, or identifiable surface patches. Extracted features have three dimensional coordinates, but since the object can be moved arbitrarily between observations a particular feature will have different coordinates in each set of observations.
Uses features to find corresponding regions in pairs of views. Data collection may be structured so it is known which views of an object contain common regions, and views can be matched immediately. Often, it is necessary to determine views which contain overlapping regions. Features, e. G. The perimeter of a paisley-shaped area of different reflectivity, which may be easily identified in several views, should be exploited. Alternatively, relations between features, such as angles and distances between line segments, might be used to create a list of views ordered in probability of overlap, and also information concerning which features of one view are likely to correspond to features in another.
Matches overlapping views successively. A single model is constructed by sequentially matching and transforming the coordinates of many views. For exam ple, suppose that views are made of an American football rotated about its axis 20 or 30 degrees between each observation. (the exact amount of rotation is immaterial, since the model-builder will discover the proper transformation between all sets of observations it uses; similarly the axis need not be constant.) View B may be found to have regions of overlap with views A, C, and D. A good matching order would be B, C, D, A, where the coordinate transformation to bring C into the frame of reference of B would be found, C would be transformed using this, then the transformation needed to bring D into this frame of reference would be found, and so on. The model builder should distinguish between this order and less favorable orders such as B, A, C, D that would be possible if the amount of rotation between observations were small.
About the Publisher
Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com
This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.
Reseña del editor
Excerpt from Automatic Model Builder for Object Recognition
Extracts suitable information from observations. Any observations that yield reliable 3d coordinates can be used as the basis for an automatic model builder. Among the many features that may be used are lines significant to object shape, feature points such as vertices, or identifiable surface patches. Extracted features have three dimensional coordinates, but since the object can be moved arbitrarily between observations a particular feature will have different coordinates in each set of observations.
Uses features to find corresponding regions in pairs of views. Data collection may be structured so it is known which views of an object contain common regions, and views can be matched immediately. Often, it is necessary to determine views which contain overlapping regions. Features, e. G. The perimeter of a paisley-shaped area of different reflectivity, which may be easily identified in several views, should be exploited. Alternatively, relations between features, such as angles and distances between line segments, might be used to create a list of views ordered in probability of overlap, and also information concerning which features of one view are likely to correspond to features in another.
Matches overlapping views successively. A single model is constructed by sequentially matching and transforming the coordinates of many views. For exam ple, suppose that views are made of an American football rotated about its axis 20 or 30 degrees between each observation. (the exact amount of rotation is immaterial, since the model-builder will discover the proper transformation between all sets of observations it uses; similarly the axis need not be constant.) View B may be found to have regions of overlap with views A, C, and D. A good matching order would be B, C, D, A, where the coordinate transformation to bring C into the frame of reference of B would be found, C would be transformed using this, then the transformation needed to bring D into this frame of reference would be found, and so on. The model builder should distinguish between this order and less favorable orders such as B, A, C, D that would be possible if the amount of rotation between observations were small.
About the Publisher
Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com
This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.
„Über diesen Titel“ kann sich auf eine andere Ausgabe dieses Titels beziehen.
