Evaluation Of Light Field Image Quality

As an emerging media modality,light Field Image(LFI)has attracted wide atten-tion.Unlike traditional media modalities,LFI can simultaneously record the direction and intensity information of radiance.It ideally provides abundant depth cues and 6 degree-of-freedom(DOF)viewing experience,thus providing a better immersive and sense of presence.Recently,many methods have been proposed that focus on the pro-cessing and application of LFI.However,these operations inevitably introduce distor-tion,resulting in the degradation of the quality of LFI.Therefore,LFI quality assess-ment is critical,which can help to better guide light field acquisition,processing and application.Since the final receiving carrier of LFI is the human visual system,it is important to analyze the subjective viewing experience of the LFI.In this work,we create the first windowed 5 DOF display mode based on stereoscopic display,which provides win-dowed 5 DOF experience and all the depth cues of LFI.Then the windowed 5 DOF LFI database(Win5-LID)is established,and it consists of LFIs with representative com-pression and reconstruction artifacts.Through subjective evaluation,we find that the quality of the LFI is affected by the spatial quality and angular consistency.However,subjective evaluation is resource and time consuming,an obj ective eval-uation method is necessary.Considering the intrinsic high dimensional characteristics of LFI,we adopt the tensor theory to explore the LF 4D structure characteristics and propose the first Blind quality Evaluator of LIght Field image(BELIF).We generate cyclopean images from the original LFI and then the features are extracted by the tensor decomposition.Specifically,Tensor Spatial Characteristic Features(TSCF)for spatial quality and Tensor Structure Variation Index(TSVI)for angular consistency are de-signed to fully assess the LFI quality.Finally,extensive experimental results on the public LFI databases demonstrate that BELIF significantly outperforms the existing image quality assessment algorithms.Then we propose No-reference Light Field image quality Assessment(NLFA)scheme for various scenarios and acquisition devices.The main idea of our proposed method is to quantify the LFI quality degradation by measuring the specific spatial qual-ity and angular consistency.Specifically,the spatial quality deterioration is measured by capturing the naturalness distribution of the light field cyclopean image array,which is formed when a human observes the LFI.Then as a transformed representation of LFI,the Epipolar Plane Image(EPI)is comprised of the slopes of lines and contains the an-gular information.Therefore,EPI is utilized to extract LFI's global and local features to measure the degradation of angular consistency.Specifically,the distribution of gradi-ent direction map of EPI is proposed to measure the global angular consistency distor-tion in the LFI.And we further propose the weighted local binary pattern to capture the characteristics of local angular consistency degradation.Finally,extensive experimen-tal results on three public LFI databases demonstrate that the proposed method NLFA significantly outperforms the existing 2D,3D,multi-views and LFI quality assessment algorithms and achieves the state-of-the-art performance.