Traditional hardness testing method may damage the material. Magnetic Barkhausen noise（ MBN）

a non-destructive testing method

can be used to detect low alloy steel specimens with different hardness. According to the characteristics of noise signal distribution

a global threshold partial reconstruction algorithm based on wavelet processing is used to de-noise the signal and extract energy characteristics and obtain energy normalized distribution graph. The experimental results show that this algorithm is superior to the traditional threshold denoising method; the MBN signal has a high correlation with hardness; the energy normalization ratio of different sub-bands reflects the characteristics of material hardness change; the near coercive force field distribution of different hardness materials presents a certain rule.