A fast discrimination method of four black heartwoods was developed by Fourier Transform Infrared Spectroscopy (FTIR) combining with clustering analysis. In FTIR spectra, the principal chemical components of heartwood were characterized for cellulose (the bands at around ~1 370, ~1 158, ~1 034 and ~895 cm^-1), lignin (the bands around ~2 935, ~1 510, ~1 462 and ~1 426 cm^-1) and calcium oxalate (peaks at ~1 615, ~1 318 and ~781 cm^-1). The correlation coefficient and relative intensity among samples with standard material result that D. ebenum and D. melanoxylon are lignin-rich, while C. imberbe contains more calcium oxalate. G. conjugate has peaked at 1 738 cm^-1 means it contains resin. Based on the correlation coefficient, the method of Compare clustering analysis was used to classify four blackwoods. The classification rate was above 95% during blind sample testing. Meanwhile, four blackwood were successfully classified by the method of SIMCA clustering analysis. The recognition rate and rejection rate reached up to 100%. The accuracy of clustering reached up to 100% during blind sample testing. It explained that the four tree species could be classified and identified completely by SIMCA clustering analysis. Besides, cellulose showed high thermal sensitivity in D. ebenum and D. melanoxylon through the 2DCOS-IR synchronous spectra. Calcium oxalate showed high thermal sensitivity in C. imberbe and lignin showed high thermal sensitivity in D. melanoxylon. Combined with cluster analysis calculation and 2DCOS-IR, FTIR can analyze the relative content of the main composition of wood and quickly and effectively classify the pattern recognition of wood species based on the improving clustering model.