Video luminance transient improvement using difference-of-Gaussian

Luminance transient improvement (LTI) technique is widely used to enhance the sharpness of digital images or videos. In this paper, a high-performance and low-complexity LTI method is proposed for practical video processing applications. Utilizing the favorable characteristics of Gaussian function, the proposed algorithm generates the correction signal 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 images or video frames. It finally utilizes a 2-D min/max nonlinear filter to suppress overshoots and undershoots appearing around edges in the video frame. And the critical modules are analyzed for practical hardware implementation for their simplicity. The performance of the method are highly confirmed by practical experimental results and proven to be effective and efficient in making steeper edges, and improving the visual quality of images or video frames while suppressing noise and reducing artifacts to a desired extent.