微量物质拉曼光谱测量精度的提高是拉曼分析技术的难点之一, 特别是高荧光背景下生物体中微量物质的测量。 根据拉曼谱峰突发、 离散特点, 分别给出荧光背景和噪声拟合函数, 通过监测总体拟合偏差A类不确定度函数实现拉曼谱峰定位和干扰信号滤波; 进一步根据谱峰位置划分光谱区间, 在单调区间内弱化非谱峰信号, 实现谱峰信号增强。 与其他光谱处理方法比较, 可以准确拆分重叠特征峰, 不会降低特征峰高度, 提供更加灵敏的半谱峰面积指标。 实验表明, 该方法在处理皮肤拉曼光谱时, 可以准确得到螺旋构象的酰胺I带、 神经酰胺和CO的归属拉曼谱峰; 另外经过该方法处理后数据建立水溶性糖（水稻叶片）含量测量模型, 其精度优于小波分解、 多项式拟合和非线性最小二乘法。

To improve the measurement accuracy for the Raman spectrum of micro-substance is one of the difficulties in Raman analysis technique, which is prominent when measuring the micro-substance in organism under the background of high fluorescence. According to the burst and dispersed characteristics to Raman spectrum peak, the fluorescence background and noise fitting function was given in this paper respectively. The location of Raman spectrum peak and interference signal filtering had been realized through monitoring A-type uncertain function in the overall fitting deviation. The spectrum section had been divided according to the spectrum peak’s location further, and the spectrum peak signal had been enhanced through weakening the non-spectrum peak signal within monotone interval. Compared to other fluorescent processing methods, this method could split the overlapped characteristic peak accurately, however, the height of the characteristic peak will not be lowered and more sensitive semi-spectrum peak index could be offered. The experiment indicates that while processing the skin Raman spectrum, such method could obtain the affiliation Raman spectra of the amide I-belt, ceramide and CO ester in helical conformation. What’s more, the measurement model with the content of water-soluble sugar (rice leaf) had been established with the data after being processed by such method, and it had been found that the precision is superior to that of the model established by using the processing data of wavelet decomposition, fitting of a polynomial and nonlinear least square method.