This paper presents a novel approach to capture light field in camera arrays based on the tensor recovery framework. Light fields are acquired by a linear array of cameras with overlapping field of view. We represent the light fields in shape of a tensor to exploit both local and non-local correlated structures of cameras image. The methodology is inspired by the recent research on matrix completion. We extend the affine matrix rank minimization to low-rank tensor recovery for light field acquisition, using a convex relaxation technique of the matrix rank. Finally, we develop a quantitative evaluation on a synthetic scene to assess our method with low-n-rank tensor recovery algorithm to assure the accuracy of our scheme with far less computational cost.