Dmitry Znamenskiy

Block-based content-adaptive sharpness enhancement

The contents of a video/TV sequence vary greatly not only across frames but also within the same frame. The existing peaking system like AutoTV where the peaking is made adaptive to the video contents and noise levels at a frame level is not sufficient for the high-end TV picture quality. This paper proposed a block-based content adaptive sharpness enhancement scheme that is able to adapt the peaking to the frame contents at a block level. In the proposed scheme, a frame is divided into a number of n × n blocks. Using some simple content-analysis techniques, each block is classified into categories with different characteristics to Human Visual System (HVS), such as the bright/dark block, the fine-details block, the strong-edge block, the smooth-area block, the skin/face block, etc. Appropriate processing is then applied to each block according to its category. The experimental results show that the proposed block-based content-adaptive peaking scheme achieves a better picture quality than the contentional peaking scheme.

Motion invariant imaging by means of focal sweep

In this paper we evaluate a so-called “sweeping focus camera” of which the camera blur is insensitive to the actual distance to the objects and to the direction and magnitude of the velocities within a scene. This can be achieved by either moving the sensor linearly along the optical axis of the main lens or changing the focal length of the main lens while the shutter is open. A consistently sharp image can then be obtained by the application of a fixed inverse filter kernel, thus preventing the need for depth and motion estimation. Moreover, high frequencies are better preserved during the acquisition, such that the results achieved by inverse filtering are superior to those achievable with conventional shutter/sensor operations.

Computation of minimal RGB LED backlight intensity for RGBW LCD displays

In the article we address the problem of RGB LED backlight power minimization in RGBW LCD displays. The problem is formalized as the determination of the minimal LEDs currents (and/or duty cycle) that allows the picture content rendering without clipping artifacts. We will write this problem in mathematical language and show that it has a unique solution. Then we propose an efficient algorithm that finds the minimal LEDs drive values according to the input content. The algorithm is stable and relatively simple. It gives temporally consistent solutions and converges to the minimal gamut in two-three iterations. When used in an RGBW display the method significantly reduces power consumption without clipping in the content.

A trained filter de-interlacer based on complex classification

A novel trained filter based scheme for video de-interlacing is proposed and described in detail. This scheme uses different classifiers, called error functions, on the input, and mixes several sub-de-interlacers depending on them. The approach differs from the earlier works in this area due to focus on more complex classification rather than on complex sub-de-interlacers. The proposed scheme is flexible and allows various combinations of error functions with sub-de-interlacers. In this article we describe a test implementation of this concept with five different sub-de-interlacers and five error functions composing a spatial-temporal de-interlacing method. The description of the test implementation is supported by simulations where we evaluate the contribution of different sub-de-interlacers and error function to output de-interlacing quality.