The accurate extraction of fat and muscle from meat image is one of the key technologies in nondestructive test (NDT). To deal with this problem, based on the normalized cut(Ncut) method, in order to reduce the computational complexity, an improved algorithm has been put forward. Firstly, the basedmoment threshold selection method to separate the meat out from the background region. Secondly, the quantitative color grade creates the color histogram for the meat area. Finally, to calculate the similar matrix of color histogram, and exploit the normalized cut as a measurement of spectral clustering on dividing histogram, in the light of the consequence in histogram partition to extract muscle and fat. Experiments illustrate that to compare with the pixel-based spectral clustering algorithm, the improved algorithm does not only reduce the computational complexity, but also the optimum segmentation.