High-speed video haze removal algorithm for embedded systems

Abstract

In this paper, we implemented a video haze removal system that runs in real time and can be integrated in a vehicle dashboard to assist drivers. We used a heterogeneous TI TMS320DM6446 platform to distribute the algorithmic tasks among ARM, DSP, and VICP cores. Our processing algorithm uses the dark channel prior and has been applied to video. It has two algorithmic components which are the operation on multi-scale resolution and a new reconstruction formula for recovering the scene radiance to save computational time without sacrificing quality. We achieved eight frames per second at 720 9 480 video frame resolution.

A Review of Mobile Mapping Systems: From Sensors to Applications

“…As the modern MMS benefits from various cameras in providing additional information, it comes with a few added complexities. First, it captures images using the reflected light off of objects, which makes it sensitive to the illumination of the environment such as the high dynamic range of the scene (between sky and ground), and hazy weather conditions [64,65]. Second, cameras, and multi-camera systems, require calibration to reduce different types of distortions [66].…”

Flexible Architecture for Real-time Processing of Multiple Video Signals

“…To implement a system that performs simultaneous processing of multiple video sources, a ready made and direct solution is to use one of the market available video processing hardware cards aided with necessary components to overcome the aforementioned challenges associated with simultaneous processing. Such type of cards are widely used as in [15][16][17][18][19][20][21][22][23]. Another solution is to design and implement a custom video processing platform tailored to specific application such as the patents in [9][10][11][12][24][25][26][27].…”