Improvement of pixel enhancement algorithm for high-speed camera imaging using 3D sparsity

We improve the performance of a pixel enhancement algorithm for high-speed camera (HSC) imaging. HSCs have a principle problem that the number of pixels decreases when the number of frames per second (FPS) increases. To suppress this problem, our optical setup is organized with a digital mirror device (DMD) array to randomly select pixels in each frame. A small number of selected pixels are recorded by an image sensor. Then, our algorithm reconstructs the entire image only from those randomly selected pixels by exploiting not only the sparsity within the each frame, but also that of difference image between adjacent frame. In this paper, we improve the performance of the algorithm in the sense of two aspects. First, we improve the accuracy of the proposed algorithm by exchanging the role of two functions in the convex optimization algorithm. Further, we accelerate the algorithm by setting a better initial value. Simulation results show that the reconstructed image quality is slightly improved and the algorithm is accelerated by several percent.