![]() 4 presents our conclusions and future work. 3) provides an in-depth analysis of the obtained results. ![]() 3 provides our experimental setup as well as experimental results obtained from our study. 3 presents our study related to the impact of the proposed inpainting and denoising solutions on latent fingerprints. 2 provides related work about inpainting and denoising solutions proposed for fingerprints. This paper is organized as follows: Sect. Hence, we provide a set of recommendations for improving these solutions and, consequently, the fingerprint identification rates. However, the fingerprint identification rates were not always improved by using inpainting and denoising solutions or using some combination of them. ![]() From our experiment result, we can conclude that the fingerprint identification rates can be improved up 63% for Rank-1 and 26% for Rank-20 when inpainting and denoising solutions are used. Our study shows that fingerprint identification can be improved by using inpainting and denoising solutions, which were trained by using impressions. Hence, in this paper, we introduce the first study testing inpainting and denoising solutions on latent fingerprint databases. However, the literature has focused on studying denoising and inpainting solutions by using fingerprints obtained in controlled situations (impressions), which present higher quality than latent fingerprints.Īs far as we know, there is no study on the impact of inpainting and denoising solutions for latent fingerprints. These solutions have positively impacted on the obtained accuracy for fingerprint verification. Some authors have been studying how improving fingerprint impressions by using denoising and inpainting solutions. Consequently, as was recently reported in, the fingerprint identification rates are lower than 10%, 13 %, and 24% for Rank-1, weighted Rank-20, and Rank-100, respectively.Īn idea to get better fingerprint identification rates is to improve the quality of latent fingerprints. Notice that latent fingerprints present incomplete and distorted images, containing noisy backgrounds. Figure 1 shows three examples of latent-rolled pairs of identified fingerprints from database NIST-SD27.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |