Due to the uncertainty of thresholds, different results are often obtained in object identification and classification of the same hyperspectral image. This paper presents a new approach to object identification based on the spectral exhaustive method (SEM). Each pixel spectrum in hyperspectral imagery has an optimal matching spectrum (OMS) among all possible spectra that are correlated with the study area by an algorithm for computing the maximal angle cosine or correlation coefficient. Hyperspectral image is mapped using OMS with different colors so that all pixels can be identified. Thus, the image becomes a color map in which those pixels corresponding to the same OMS have the same color. A gray map is created by calculating the correlation coefficient between each pixel spectrum and its OMS to determine the degree of match. This approach requires all possible reference spectra, mostly from the spectral library (built by us), and various pretreatments of the hyperspectral data. This approach was applied to an EO-1 Hyperion hyperspectral image of the Pulang area, Zhongdian county, Yunnan province, China, to obtain identification maps of the study area. Examination of the outlined objects in the map indicated that the SEM- based approach is effective. Keywords: Hyperspectral image; The spectral exhaustive method (SEM); Object identification.