An interactive graphics system with bi-directional audio capability has been developed for use in classifying and describing (or characterizing) speech and image data. The use of an interactive system allows human ins...
详细信息
An interactive graphics system with bi-directional audio capability has been developed for use in classifying and describing (or characterizing) speech and image data. The use of an interactive system allows human inspection of intermediate steps in algorithm development, easy modification of parameters, and immediate examination of the results. The graphics portion of the system uses a keyboard, joystick, opaque and transparent image scanners, and control panel for data input and has access to a Tektronix 611 memory scope, an electrostatic printer/plotter, a Calcomp plotter, and a high-resolution CRT with a camera attachment for data output. As many as three different files may be processed concurrently and may be stored either on magnetic tape or disc. A history file which maintains a record of all or selected operations, may be viewed directly or on hardcopy. The system has been used to generate a labelled speech data base of 35,000 tokens for use in semi-automatic speaker identification character recognition algorithm capable of MOS/LSI realization, to analyze imagery for bandwidth reduction studies, and to research classification and matching algorithms for use in fingerprint processing. Examples from each of the application areas are presented.
The results of a fuzzy segmentation can be used for subsequent classification algorithms which can evaluate fuzzy predicates. For example a peak detection algorithm searching for particular sequences of corners can an...
详细信息
The results of a fuzzy segmentation can be used for subsequent classification algorithms which can evaluate fuzzy predicates. For example a peak detection algorithm searching for particular sequences of corners can answer a question about the presence of a peak not by yes or no but by the min left bracket f(x//i) right bracket where I is a member of the set I, and where I is the set of corners forming the peak. Such fuzzy predicates can be combined and used for the final classification. If the final classes are well defined then the fuzziness measured in this way can be used to adjust the design of the overall recognizer. Thus a ″well segmented″ sample will have a high degree of membership while the opposite will be true for a ″poorly segmented″ sample. Misclassification of the former indicates deficiencies in the structure of the classification algorithm which could be corrected without changing the original feature generation algorithm. On the other hand if numerous ″poorly segmented″ samples are found this would suggest the need for a thorough redesign of the feature generation procedure. Notice that ″poorly segmented″ samples are counted regardless of correct or erroneous classification.
The authors describe some of the techniques they employ in processing xeromammograms so that abnormal regions of the breasts can be detected automatically. computing is restricted to a minicomputer with 8,000, 16-bit ...
详细信息
The authors describe some of the techniques they employ in processing xeromammograms so that abnormal regions of the breasts can be detected automatically. computing is restricted to a minicomputer with 8,000, 16-bit words of core memory, thereby stimulating the construction of cost-effective algorithms. The methods proceed as follows. First breast tissue in the xeromammogram is identified. The tissue is then partitioned into at most 144 sections per breast in such a way that a particular section on the right breast has a similarly placed section on the left breast. Next, 10 statistics are computed for each section to use as features representing texture. The features are normalized and then selected or eliminated by evaluating how well they separate abnormal from normal tissue on a small training set of sections. The training set is also used to devise simple classification rules for the selected features. The complete classification of 2,270 sections for eight patients takes less than 2 minutes of minicomputer time and yields a false positive rate of 26 percent and a false negative rate (for cancerous sections) of . 6 percent.
This volume includes 89 technical papers, of which 17 are abstracts or summaries only, arranged by sessions, as follows: Raster Scan Graphics;Data Structures (3 sessions);Texture Analysis;Analysis of Medical images;Fo...
详细信息
This volume includes 89 technical papers, of which 17 are abstracts or summaries only, arranged by sessions, as follows: Raster Scan Graphics;Data Structures (3 sessions);Texture Analysis;Analysis of Medical images;Formal Definitions of Data Structure;patternrecognition (4 sessions);computer Graphics (3 sessions);Fuzzy Representation and recognition;Semantic patternrecognition;Human Engineering and computer Graphics;computer-Aided Design;computer Graphics, patternrecognition and Data Structure Applied to Chemical Structure Analysis. Seventy-one selected papers are indexed separately.
暂无评论