In the field of face recognition, sparse representation (SR) has received considerable attention during the past few years, with a focus on holistic descriptors in closed-set identification applications. The underlyin...
详细信息
In the field of face recognition, sparse representation (SR) has received considerable attention during the past few years, with a focus on holistic descriptors in closed-set identification applications. The underlying assumption in such SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such an assumption is easily violated in the face verification scenario, where the task is to determine if two faces (where one or both have not been seen before) belong to the same person. In this study, the authors propose an alternative approach to SR-based face verification, where SR encoding is performed on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which then form an overall face descriptor. Owing to the deliberate loss of spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment and various image deformations. Within the proposed framework, they evaluate several SR encoding techniques: l(1)-minimisation, sparse autoencoder neural network (SANN) and an implicit probabilistic technique based on Gaussian mixture models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the local SR approach obtains considerably better and more robust performance than several previous state-of-the-art holistic SR methods, on both the traditional closed-set identification task and the more applicable face verification task. The experiments also show that l(1)-minimisation-based encoding has a considerably higher computational cost when compared with SANN-based and probabilistic encoding, but leads to higher recognition rates.
The special physical and chemical properties of the reclaimed land caused by the disturbance of rare earth mining and the environmental stress caused by the mining of rare earth lead to the inhibition of the physiolog...
详细信息
The special physical and chemical properties of the reclaimed land caused by the disturbance of rare earth mining and the environmental stress caused by the mining of rare earth lead to the inhibition of the physiological functions of the reclaimed vegetation and the severe challenge of vegetation ecological restoration. This study focuses on the Lingbei rare earth mining area in Dingnan County, Jiangxi Province, and investigates the original spectrum, derivative spectrum, and the continuum-removed spectrum of reclaimed vegetation. The spectral characteristics and trends and the typical reclaimed vegetation are analyzed, the correlation between the chlorophyll content and the spectral indices of the reclaimed vegetation is determined, and the sensitive spectral parameters are extracted. Partial least squares algorithm, a back propagation neuralnetwork algorithm, and a sparseautoencodernetwork are used to estimate the chlorophyll content, and the model's accuracies are compared. The vegetation spectrum of the reclaimed vegetation is characterized by high reflectance in the visible region, a redshift of the green peak and red valley positions, and a blueshift of the red edge positions, a relatively low spectral variation in. The variability of the sensitive spectral parameters of different vegetation type is extracted. The sparseautoencodernetwork is the optimal estimation model (the R-2 value of the three vegetations are 0.9117, 0.7418, and 0.9815, respectively). The results provide a scientific basis for monitoring and managing the growth of different types of reclaimed vegetation in rare earth mining areas under environmental stress and can guide the ecological restoration of reclaimed mining areas.
暂无评论