1 Wave atoms based compression method for fingerprint images Zehira HADDAD 1,2 , Azeddine BEGHDADI 1 , Amina SERIR 2 , Anissa MOKRAOUI 1 1 L2TI, Institut Galilée, Université Paris 13, Sorbonne Paris Cité 99 Avenue Jean-Baptiste Clément, 93430 Villetaneuse France 2 LTIR, Faculté d’électronique et d’informatique, USTHB, BP 32 El Alia 16111 Bab Ezzouar Alger, Algérie Abstract This paper proposes a new fingerprint image compression approach where the quality of the decoded image is perceptually controlled using Wave atom transform. A comparative study of different transforms shows that Wave atom transform is the more appropriate than Wavelets for fingerprint image compression since it is able to better represent the geometrical structures of the fingerprint. A new image quality metric based on the same transform that has been used for compression is proposed to control the compression performance. Some properties of the human visual system are exploited and introduced in the developed metric. Simulations show that the proposed image quality metric correlates well with the subjective human judgment. According to these interesting results we developed a compression method specific to fingerprint images where the distortion is perceptually controlled. A recognition fingerprint system shows that the proposed strategy offers better results than traditional compression methods. Keywords – Biometrics, Fingerprint compression, Image Quality Metric (IQM), Wavelets, Ridgelets, Curvelets, Wave atoms, Wavelet Scalar Quantization (WSQ), Human Visual System (HVS). 1. Introduction During the last three decades, transform based image compression approaches have been extensively studied and some well-established standards for image and video coding appear since the 90’s. Historically, many orthogonal transforms, such as the Discrete Fourier Transform (DFT), Haar Transform, Walsh Hadamard Transform, Slant Ttransform, the Discrete Cosine Transform (DCT) and some others interesting transforms have been used for lossy image compression [1]. The Karhunen- Loeve Transform (KLT), also known as Hotelling Transform or Eigenvector Transform, is theoretically the best one, in the sense of energy compaction and decorrelation. However, it is data dependent and computationally more involved. For these main reasons KLT could not be used in practice [2]. A comparative study in [3] showed that one of the most suitable transform in terms of decorrelation and compactness is the DCT [4]. It offers the advantages of KLT without suffering from its drawbacks. Furthermore, unlike KLT, this transform uses a fixed basis, independent of the data. Some