Feldspar is the most important rock-forming mineral in surface rocks, accounting for up to 60% in the earth's crust. With the development of high-spectral technology, many scholars at home and abroad have been studying the response of the main building of rock and mineral content and the spectrum of the features, and it offers a variety of possibilities for remote sensing, mineralization, and mineralization. It is based on data of 18 igneous samples in the USGS spectral library to study the quantitative relation between the characteristic spectra and the content of the feldspar. Through the original spectral reflectivity and the transformation (including the three layers of the small wave to break down the high frequencies, the little wave layers, the spectrum of the wave to the back of the line, and then the little wave layer after the cable, and then the little wave layer after the cable, and then the small wave layer of the wave, and then the second layer of the cable) study the correlation between the high and the long rocks, and it turns out: (1) to analyze the transformation of six spectral reflectivity, the relation between the spectral reflectivity of the high frequency and the long rock content of the small wave three layers of the envelope, and the correlation coefficients are the best, and the correlation coefficients are constantly changing, and based on the relative coefficients of the relative coefficients, the high value of the long rock is 4, 31, 570, 972, 1 456, 1 856, 2 292.9, 2 481 nm; (2) the correlation between the original spectral reflectance and feldspar content curve trend is relatively flat, and after the wavelet decomposed high frequency component, through to the envelope and wavelet decomposition for the high frequency component, the change trend of correlation curve is becoming ever more obvious, therefore, the independent variable of the small changes will cause changes in the dependent variable, very hour when the content of the rock feldspar, wavelet decomposition processing can improve the accuracy of the model. The relation between the content of feldspar and the characteristic spectrum is quantified, using a multi-element stepwise linear regression analysis and a least-square method of modeling, establishing six linear regression models and six least-squares regression models, and the results show that: (1) the spectrum of the spectrum behind the envelope is more accurate than the original spectrum, and the low-frequency portion of the lower wave part of the wave is better than the one that's not done with the small wave decomposition, which is the best way to get to the back model of the low-frequency portion of the small wave after the envelope. (2) the multilinear regression model is better than the least-squares, and the variables that affect the variables that affect the larger variables, 972, 1 456, 1 856, 2 292.9 and 2 481 nm. In view of that relation between the content of feldspar and the spectral reflectance of the long stone and the influence factors on the content of feldspar in different absorption band, it is important to use the characteristic spectrum of feldspar to quantitatively invert the content of feldspar in a region.