摘要:
A standard pattern vector of a normal sound, an input pattern vector of a monitored sound, a positive vector of a reference pattern, and a negative vector of the reference pattern are produced. Regarding each element, the positive vector is increased by an absolute value of a difference between the input pattern vector and the standard pattern vector if the input pattern vector is greater than the standard pattern vector, and the negative vector is increased by the absolute value if the input pattern vector is less than the standard pattern vector. A difference of a kurtosis between the positive vector and the negative vector is calculated, wherein a center of the reference shape is relatively moved to a position of each element of the positive and negative vectors respectively at every calculation. A geometric distance value between the standard pattern vector and the input pattern vector is obtained by calculating a square root of a sum of a square of each kurtosis difference.