While minutiae is important for high-resolution palmprint matching, the quality of minutiae is affected by principal lines, creases and other noises, and therefore it is necessary to estimate the quality of minutiae and to exclude poor minutiae. In this paper, a minutiae quality estimation algorithm based on learning for high-resolution palmprint is proposed. First, a series of features obtained by applying Gabor convolution, complex filtering, etc., are used to describe the local area of minutiae redundancy. Then, with each feature as a weak classifier, AdaBoost algorithm is applied in multi-layered training to identify the best features for discriminating minutiae. Finally, the response of weighted linear combination of weak classifiers is used as minutiae quality score, and minutiae with score above the threshold is selected as true minutiae. Comparing with the method based on Fourier transform response, the presented method is superior at distinguishing true from false minutiae, and provides better evaluation of minutiae quality.