Boosting the classification performance of latent fingerprint segmentation using cascade of classifiers

Segmentation and classification of latent fingerprints is a young challenging area of research. Latent fingerprints are unintentional fingermarks. These marks are ridge patterns left at crime scenes, lifted with latent or unclear view of fingermarks, making it difficult to find the guilty party. The segmentation of lifted images of such finger impressions comes with some unique challenges in domain such as poor quality images, incomplete ridge patterns, overlapping prints etc. The classification of poorly acquired data can be improved with image pre-processing, feeding all or optimal set of features extracted to suitable classifiers etc. Our classification system proposes two main steps. First, various effective extracted features are compartmentalised into maximal independent sets with high correlation value, Second, conventional supervised technique based binary classifiers are combined into a cascade/stack of classifiers. These classifiers are fed with all or optimal feature set(s) for binary classification of fingermarks as ridge patterns from non-ridge background. The experimentation shows improvement in accuracy rate on IIIT-D database with supervised algorithms.

XFinger-Net: Pixel-Wise Segmentation Method for Partially Defective Fingerprint Based on Attention Gates and U-Net

Partially defective fingerprint image (PDFI) with poor performance poses challenges to the automated fingerprint identification system (AFIS). To improve the quality and the performance rate of PDFI, it is essential to use accurate segmentation. Currently, most fingerprint image segmentations use methods with ridge orientation, ridge frequency, coherence, variance, local gradient, etc. This paper proposes a method of XFinger-Net for segmenting PDFIs. Based on U-Net, XFinger-Net inherits its characteristics. The attention gate with fewer parameters is used to replace the cascaded network, which can suppress uncorrelated regions of PDFIs. Moreover, the XFinger-Net implements a pixel-level segmentation and takes non-blocking fingerprint images as an input to preserve the global characteristics of PDFIs. The XFinger-Net can achieve a very good segmentation effect as demonstrated in the self-made fingerprint segmentation test.

Segmentation of Latent Fingerprint using Neural Network

Latent fingerprints are the fingerprints that are left by the criminal unintentionally on the surface of the crime scene. The qualities of the latent fingerprints are very poor due to the overlapping patterns and structured noises. Latent fingerprint segmentation is a difficult task due to low visibility, structured noise, and complex structure. In this paper, a fusion of morphological and neural network approach is purposed for latent fingerprint segmentation. This method automatically segments the fingerprints and non-fingerprints patterns without human intervention. The morphological method is used for segmentation of the fingerprint region. Fingerprint region then divides into y*y blocks and extracts the features of each block and uses them as an input of NN to classify the blocks into fingerprint and non-fingerprint blocks. We are using the IIIT-D database and the shows that this model batters then the existing model.

Improving of Fingerprint Segmentation Images Based on K-MEANS and DBSCAN Clustering

<span lang="EN-US">Nowadays, the fingerprint identification system is the most exploited sector of biometric. Fingerprint image segmentation is considered one of its first processing stage. Thus, this stage affects typically the feature extraction and matching process which leads to fingerprint recognition system with high accuracy. In this paper, three major steps are proposed. First, Soble and TopHat filtering method have been used to improve the quality of the fingerprint images. Then, for each local block in fingerprint image, an accurate separation of the foreground and background region is obtained by K-means clustering for combining 5-dimensional characteristics vector (variance, difference of mean, gradient coherence, ridge direction and energy spectrum). Additionally, in our approach, the local variance thresholding is used to reduce computing time for segmentation. Finally, we are combined to our system DBSCAN clustering which has been performed in order to overcome the drawbacks of K-means classification in fingerprint images segmentation. The proposed algorithm is tested on four different databases. Experimental results demonstrate that our approach is significantly efficacy against some recently published techniques in terms of separation between the ridge and non-ridge region.</span>