Because the conventional wavelet transform profilometry can be easily influ enced by the low noise,which results in the problem of inaccurate phase,an improved wavelet ridge phase extraction algorith m based on cost function is proposed.Firstly,the algorithm extracts the maximum points by the wavelet transform coefficient modu lus,and then gets the inflection points which are easily ignored by the phase information,so the complete candidate wavelet ridg e points are obtained.Then,improve the weight factors in raw cost function by introducing the logarithm Logistic mode, which can balance the importance of the two indicators in cost function,and then obtain more reasonable cost estimation.Finally,quickly and accurately identify the optimal cost function ridge line from the cand idate ridge points by dynamic programming ideas which is in c ost function algorithm,and phases on ridge line are extracted.Experimental results show that the phase extraction error is dec reased by 13.5%.Our algorithm can precisely locate the wavelet ridge in a low noise environment,and extract more accurate phase.