In this paper, a chaotic encryption framework is proposed for enhancing the security of biometrics images during transmission. The proposed framework is based on the fractional wavelet packet transform (FrWPT), chaotic map and Hessenberg decomposition. The core idea is to shuffle biometrics image using affine transform followed by the transformation in FrWPT domain with chaotically generated transform orders. Now, FrWPT coefficients are altered using the Hessenberg decomposition. Finally, a reliable decryption process is presented to reconstruct original biometrics image from the encrypted data only with the valid keys. The efficiency and robustness of the proposed technique are validated by different kind of simulations and analysis on fingerprint images. © 2013 Springer-Verlag Berlin Heidelberg.