A deformation-tolerant method for classification/recognition of low resolution (FLIR regime) ship imagery is presented which employs statistical transformations and correctors based on concepts of fractal geometry. Fractal analyses, applied to specific classes of contours, present advantages of high recognition accuracy, position- and size-invariance and are suitable for microprocessor-based implementation due to low computational and storage requirements. The relationship between deformation-tolerance, superstructure geometry, and inherent transformation characteristics is presented in terms of image-plane distortions induced by out-of-plane ship rotation. Comparison algorithms using feature-space correctors derived from the fractal dimension are discussed in terms of classification and recognition success rates and computational load.
Deformation-Tolerant Statistical Correctors For Enhancement Of Ship Silhouette Recognition
Infrared Image Processing and Enhancement ; 1987 ; Orlando,FL,United States
Proc. SPIE ; 0781
1987-09-16
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Logical correctors in recognition
British Library Online Contents | 2014
|Models of recognition procedures with logical correctors
British Library Online Contents | 2013
|Water-cooled wavefront correctors
IEEE | 2005
|Studies on silhouette quality and gait recognition
IEEE | 2004
|