New Robust Part-Based Model with Affine Transformations for Facial Landmark Localization and Detection in Big Data

Abstract

In this paper, we developed a new robust part-based model for facial landmark localization and detection via affine transformation. In contrast to the existing works, the new algorithm incorporates affine transformations with the robust regression to tackle the potential effects of outliers and heavy sparse noises, occlusions and illuminations. As such, the distorted or misaligned objects can be rectified by affine transformations and the patterns of occlusions and outliers can be explicitly separated from the true underlying objects in big data. Moreover, the search of the optimal parameters and affine transformations is cast as a constrained optimization programming. To mitigate the computations, a new set of equations is derived to update the parameters involved and the affine transformations iteratively in a round-robin manner. Our way to update the parameters compared to the state of the art of the works is relatively better, as we employ a fast alternating direction method for multiplier (ADMM) algorithm that solves the parameters separately. Simulations show that the proposed method outperforms the state-of-the-art works on facial landmark localization and detection on the COFW, HELEN, and LFPW datasets.

New Robust Tensor PCA via Affine Transformations and L 2,1 Norms for Exact Tubal Low-Rank Recovery from Highly Corrupted and Correlated Images in Signal Processing

“…High-dimensional tensor image recovery has also received a considerable amount of research attention. For instance, the authors in [51] addressed the robust tensor to overcome the drawbacks of the linear regression methods which were considered in [52][53][54] for head pose estimation and high-dimensional image alignment and recovery. ese methods, however, still lack in pruning out the potential impacts of occlusions, outliers, and uncontrolled illuminations.…”

“…Additionally, several novel methods were used to solve dynamic problems for real-time applications, for instance, [59][60][61][62][63]; yet the issue of pruning out the potential annoying effect needs more robust tensor method via affine transformations. To tackle the setback, different new models were developed by [1][2][3][4][5]53], extending the existing work into the high tensor approaches via incorporating the affine transformations, while the authors in [64,65] proposed a probabilistic method for head pose estimation by directly mapping the feature vectors onto the yaw angles. Diaz-Chito et al [66] addressed a method to narrow down the gap between the head yaw angles and the regression by combining manifold embedding methods with linear regression.…”

An Efficient New Robust PCA Method for Joint Image Alignment and Reconstruction via the L 2,1 Norms and Affine Transformation

“…(4) e L 2,1 norm of G is taken to estimate the Gaussian noise in real applications so as to minimize the potential impact of outliers, occlusions and illuminations, and heavy sparse errors. Unlike to the others work [2,[33][34][35][36][37][38][39], the new method tries to decompose the aggregated errors as the Gaussian noise and sparse error, which make the new method more novel. (5) e new set of equations, which are derived in more detailed including affine transformation using an ADMM method, is used to improve the robustness and solve the set of new optimization problems.…”