Table 3.
Characteristics of the local binary pattern-based feature extraction methodologies.
| Ref. | Key features | Advantages | Disadvantages |
|---|---|---|---|
| [59] | Usage of NN for local feature extraction | Very fast and robust against obscure images | High EER (0.13%) |
| [60] | Alignment using extracted minutiae points | Fast with a low EER (0.081%) | An in-house dataset is used instead of a benchmark one |
| [61] | Extraction of holistic codes through weighted LBP | Reduced processing time and a low EER (0.049%) | Requires setting of weights |
| [62] | Combination of LBP and Wavelet transformation | Low EER (0.011%), fast, and robust against irregular shading and saturation | Tested on a small dataset |
| [63] | Combination of a modified Gaussian high-pass filter with LBP and LDP | Improvement compared with using vein pattern features, a faster processing time, an EER of 0.89% |
Not reported |
| [64] | LBP image fusion based on multiple instances | Simple with low computational complexity and improves the RR on low-quality images | High EER (1.42%) |
| [65] | Application of PBBM | Removes noisy bits, personalized features, and highly robust and reliable with a low EER (0.47%) | A small in-house dataset is used instead of a benchmark one |
| [66] | Application of GLLBP | Performs better than other conventional methods on the collected dataset, an EER of 0.58% | Not reported |
| [67] | Application of MOW-SLGS | Takes into account location and direction information | Low RR (96.00%) |
| [68] | Application of enhanced BGC (LHBGC) | Fast, a low EER (0.0038%) when using multiple fingers, and robust against noises | Low EER in cases with multiple fingers |
| [69] | Application of LEBP | Low FPR (0.0129%) and TPR (0.90%) | Low accuracy (97.45%) |
| [70] | Application of DSLGS | More stable features with better performance than the original | High EER (3.28%) |
| [71] | Application of CSBC | High accuracy (99.84%) and a low EER (0.16%) | Multi-modal application |
| [72] | Application of PDVs and AMBP | Solves out-of-sample problems, robust against local changes, and fast with a low EER (0.29%) and a high RR (100%) | Accuracy depends on parameters |
| [73] | Application of multi-directional PDVs | Outperforms state-of-the-art algorithms with a low EER (0.30%) | Complexity analysis is not reported |
| [74] | Fusion of vein images with an ECG signal through DCA | Better than two individual unimodal systems, a low EER (0.1443%) | Multi-modal application |
| [75] | Application of ADLBP | Better describes texture than LBP | Low RR (96.93%), multi-modal application |