A new implementation method of simplex growing algorithm (SGA) is proposed based on support vector machine (SVM), which is free of dimensional reduction and makes use of distance measure instead of volume one. The unmixing equality of linear SVM and linear spectral mixing modeling (LSMM) is proved. The superiorities of linear SVM based spectral unmixing in two extended applications, combined use of endmember informations and nonlinearity use of the model, are explored. Experiments show that the computational complexity of the SVM based implementation method of SGA is decreased greatly, while the unmixing accuracy is obviously improved.