Supervised texture segmentation using support vector machines

Medical Image Segmentation Based on Threshold SVM

“…The concrete flow is given by Fig.1 Background and chest sample set are training sample set input of SVM, mid-point set is testing sample set which will be classified.RBF is selected as kernel function [4][5][6],whose kernel parameter is γ ,results are shown in Fig.4 It is shown that the segmentation effect is better when kernel parameter 0.001 γ = ,the chest can basically be segmented from human background.…”

A Survey on Pattern Recognition Applications of Support Vector Machines

“…55 SVM is also used to solve text detection and categorization problem. 52,54 The aim of many nonlinear forecasting methods 23,27,69,96 is to predict next points of time series. Tay and Cao 96 proposed C-ascending SVMs by increasing the value of C, the relative importance of the empirical risk with respect to the growth of regularization term.…”

Texture-based approach for text detection in images using support vector machines and continuously adaptive mean shift algorithm

“…In the proposed method, support vector machines (SVMs) are used as the texture classifier due to their robustness even in the absence of a rich set of training examples. The previous success of SVMs in texture classification [8] and other related problems [9], [10] also provided further motivation to use SVMs as the classifiers for identifying text regions. In addition, since SVMs work well even in highdimensional spaces, no external feature extractor is required to reduce the dimensionality of the texture pattern, thereby eliminating the need for a time-consuming feature extraction stage.…”

“…The main difficulty in octave band wavelet decomposition is that it can provide only a logarithmic frequency resolution which is not suitable for the analysis of high-frequency signals with relatively narrow bandwidth. The investigations of Chang and Kuo [5] and Laine and Fan [6] indicate that the texture features are more prevalent in the intermediate frequency band and showed promising results using wavelet packet frames [7], [8]. Therefore, the main motivation of this work is to utilize the decomposition scheme based on M-band (M > 2) wavelets, which, unlike the standard wavelet, provides a mixture of logarithmic and linear frequency resolution [9], [10].…”