An efficient algorithm and FPGA implementation of video luminance transient improvement

In this paper, a high-performance and low-complexity luminance transient improvement (LTI) algorithm is proposed and efficiently implemented on field programmable gate array (FPGA) devices,which can be widely used to enhance the sharpness of digital video. The proposed algorithm generates the correction signal by using the difference of the outputs of two Gaussian filters with different variances, and then modulates the correction signal adaptively according to the local contrast information of video frames. A 2-D min/max nonlinear filter is employed to suppress overshoots around edges. The proposed algorithm is thoroughly confirmed by experiments and compared with other algorithms on images, which produces steeper edges and better visual quality while suppressing noise and artifacts. And the hardware architecture suitable for FPGA implementation is optimized based on the property of the algorithm and proves to be effective and efficient in many respects, such as resource consumption, performance and reconfigurability. The specific implementation details on both Xilinx and Altera FPGA devices are also described in this paper.