基于混合算法的色彩光谱反射比降维重建研究

The Research of Dimension Reduction and Reconstruction of Color Multispectral Reflectance Based on Hybrid Algorithm

利用BP神经网络和主成分分析相结合的算法,实现色彩光谱反射比的降维与重建。首先利用神经网络实现色彩多光谱数据的降维,并根据色差标记达到精度的数据。然后将未达到精度的光谱数据利用PCA主成分分析算法进行降维。最后利用神经网络逆运算与矩阵运算法重建多光谱空间。研究表明,利用混合算法对多光谱空间进行降维与重建,能够高精度的表示原始色彩光谱空间。

To realize color multi - spectral data dimension reduction and reconstruction by combining BP neural network with principal component analysis. Firstly, Achieve local dimension reduction of color multi - spectral data by BP neural network and marked. Then, to reach the accuracy of dimension reduction to the spectral data that not marked by using PC A principal component analysis. Finally, to reconstruct the multispectral data information by using neural network and the matrix. Experimental results showed that using the hybrid algorithm for multispectral dimension reduction and reconstruction can highly represent the original color spectrum space.