Local Transform Features and Hybridization for Accurate Face and Human Detection

Abstract

We propose two novel local transform features: local gradient patterns (LGP) and binary histograms of oriented gradients (BHOG). LGP assigns one if the neighboring gradient of a given pixel is greater than its average of eight neighboring gradients and zero otherwise, which makes the local intensity variations along the edge components robust. BHOG assigns one if the histogram bin has a higher value than the average value of the total histogram bins, and zero otherwise, which makes the computation time fast due to no further postprocessing and SVM classification. We also propose a hybrid feature that combines several local transform features by means of the AdaBoost method, where the best feature having the lowest classification error is sequentially selected until we obtain the required classification performance. This hybridization makes face and human detection robust to global illumination changes by LBP, local intensity changes by LGP, and local pose changes by BHOG, which considerably improves detection performance. We apply the proposed features to face detection using the MIT+CMU and FDDB databases and human detection using the INRIA and Caltech databases. Our experimental results indicate that the proposed LGP and BHOG feature attain accurate detection performance and fast computation time, respectively, and the hybrid feature improves face and human detection performance considerably.

A survey on face detection in the wild: Past, present and future

“…Variations of LBP and HoG features were proposed in [22] and applied for face detection. In particular, the Local Gradient Patterns (LGP) and Binary Histograms of Oriented Gradients (BHOG) were proposed.…”

“…When compared to LBPs the LGP provides robustness to local gradient variations caused by makeup, glasses and possible background variations. The other descriptor proposed in [22], the so-called BHOG, computes features as follows: The square of the gradient magnitude and the orientation of all pixels within a predefined block is computed, then an orientation histogram is computed in a similar manner as in HoGs, finally by thresholding the histogram bins (i.e., assigning 1 if the histogram bin has a value higher than the threshold otherwise assigning 0) the orientation histogram is encoded into an 8 bit vector. The threshold value was set to be the average value of the total histogram bins.…”

“…The advantage of BHOG features is that they can be efficiently computed using the integral histogram methods [104] and could also be efficiently combined with boosting methods (such as Adaboost). The paper also proposed to fuse LBP, LGP and BHOG for face detection [22].…”

“…A number of follow-up works showed that there is indeed information in the results from the previous nodes, and it is best to reuse them [80,19] etc. Spectral histogram [86] Spatial histogram (LBP-based) [87] HoG and LBP [88] Region covariance [89] SURF [102,103] Composite Joint Haar-like features [62] features Sparse feature set [90] LGB, BHOG [22] Integral Channel Features on HoG and LUV (Headhunter) [26] HoG, HSV, RGB, LUV, Grayscale, Gradient Magnitude [105] Shape features Boundary/contour fragments [94,95] Edgelet [96] Shapelet [97] instead of starting from scratch at each new node. For instance, in [110], the use of a "chain" structure was proposed to integrate historical knowledge into successive boosting learning.…”

“…• The introduction of robust feature extraction methodologies, such as Scale Invariant Feature Transform (SIFT) features [18], Histograms of oriented Gradients (HoGs) [19], Local Binary Patterns (LBPs) and their variations [20,21,22], their fast counterparts such as Speeded Up Robust Features (SURF) [23] and DAISY [24], as well as transformations that combine the above features with integral images, such as Integral Channel Features (ICF) [25,26]. These features are densely or sparsely sampled and are used to describe face appearance.…”