Optimal Threshold Estimation Using Prototype Selection

A technique is proposed for choosing the thresholds for a number of object detection tasks, based on a prototype selection technique. The chosen prototype subset has to be correctly classified. The positive and negative objects are introduced in order to provide the optimization via empirical risk minimization. A Boolean function and its derivatives are obtained for each object. A special technique, based on the fastest gradient descent, is proposed for the sum of Boolean functions maximization. The method is applied to the detection task of house edges, using its images in aerial photos. It is shown that proposed method can be expanded to solving of a wide range of tasks, connected to the function optimization, while the function is given in vertices of a 2n single hyper - cube.

[1]  Chi Hau Chen,et al.  Pattern recognition and signal processing , 1978 .

[2]  Ludmila I. Kuncheva,et al.  Editing for the k-nearest neighbors rule by a genetic algorithm , 1995, Pattern Recognit. Lett..

[3]  Filiberto Pla,et al.  Prototype selection for the nearest neighbour rule through proximity graphs , 1997, Pattern Recognit. Lett..

[4]  L. Zadeh.  Fuzzy sets as a basis for a theory of possibility , 1999 .

[5]  Jaihie Kim,et al.  Learning of prototypes and decision boundaries for a verification problem having only positive samples , 1996, Pattern Recognit. Lett..

[6]  John E. Dennis,et al.  Numerical methods for unconstrained optimization and nonlinear equations , 1983, Prentice Hall series in computational mathematics.

[7]  Peter E. Hart,et al.  The condensed nearest neighbor rule (Corresp.) , 1968, IEEE Trans. Inf. Theory.

[8]  Donald Geman,et al.  Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  Peter N. Heller,et al.  The application of multiwavelet filterbanks to image processing , 1999, IEEE Trans. Image Process..

[10]  Edward M. Riseman,et al.  How Easy is Matching 2D Line Models Using Local Search? , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[11]  Anil K. Jain,et al.  Goal-Directed Evaluation of Binarization Methods , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[12]  Uri Lipowezky.  Selection of the optimal prototype subset for 1-NN classification , 1998, Pattern Recognit. Lett..

[13]  Jack Sklansky,et al.  Feature Selection for Automatic Classification of Non-Gaussian Data , 1987, IEEE Transactions on Systems, Man, and Cybernetics.

[14]  Josef Kittler,et al.  Pattern recognition : a statistical approach , 1982 .

[15]  Didier Demigny,et al.  A Discrete Expression of Canny's Criteria for Step Edge Detector Performances Evaluation , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[16]  U. Lipowezky.  Tree-plantation decipherment of panchromatic aerial photo images using supervised template matching , 1998, MELECON '98. 9th Mediterranean Electrotechnical Conference. Proceedings (Cat. No.98CH36056).

[17]  J. Kittler,et al.  Feature Set Search Alborithms , 1978 .

[18]  Christine Decaestecker,et al.  Finding prototypes for nearest neighbour classification by means of gradient descent and deterministic annealing , 1997, Pattern Recognit..

[19]  G. Gates.  The Reduced Nearest Neighbor Rule , 1998 .

[20]  G. Gates,et al.  The reduced nearest neighbor rule (Corresp.) , 1972, IEEE Trans. Inf. Theory.

[21]  Anil K. Jain,et al.  Feature Selection: Evaluation, Application, and Small Sample Performance , 1997, IEEE Trans. Pattern Anal. Mach. Intell..