This paper represents a novel approach to transform one dimension (1-D) signals into two dimension (2-D) grayscale image and a feature extraction process to extricate detail texture data of this 2D image to classify signals utilizing multi-class support vector machine. In all previous approaches of the signal processing strategies, the signal is continuously processed in one dimension (1-D) representation. Hence, a gigantic relationship information between time and frequency coefficients is effectively missing. To annihilate these issues, two dimensions representation of the signal is assessed in this paper. Centering on creating a proficient highlight extraction strategy for evacuating deficiencies of motor signals utilizing the 2-D image. Each pixel is taken and squaring it to discover out the energy and making it to gray image. The esteem of tests is normalized based on the tests of the signals within the time space, and Empirical Mode Decomposition (EMD) is to distinguish the low frequency which fundamentally represents to noise and evacuate it from the image. Segmentation-based Fractal Texture Investigation (SFTA) algorithm is used to extricate the feature vectors which are utilized for classifying the signals using multi-class support vector machine (SVM). The precision is 88.57% which is picked up from confusion matrix while classifying signals.