Video system with dynamic contrast and detail enhancement

摘要

A method and system for real time luminance correction and detail enhancement of a video image including the steps of extracting a luminance component from a video image, separating the luminance component into an illumination layer and a scene reflectivity layer, the illumination layer having a dynamic range, compressing the dynamic range of the illumination layer to generate a corrected illumination layer, filtering the reflectivity layer to generate an enhanced reflectivity layer; and combining the corrected illumination layer with the enhanced scene reflectivity layer to generate an enhanced luminance image, is provided. A system for real time luminance correction and detail enhancement of a video image is also provided.

主权项

1. A method for real time luminance correction and detail enhancement of a video image, comprising:
calculating a luminance component of each pixel of each frame of a real-time video image signal; converting the luminance component into log space to generate a log luminance component; generating a base image for each frame of a video image signal by applying an adaptive bilateral filter to said log luminance component of each pixel in said frame; generating a detail image for each frame of a video input signal by taking the difference between said log luminance component of each pixel in said frame and said base image; identifying any dark pixels in the base image; generating a compressed base image by linearly compressing dark pixels in the base image towards a setpoint using a linear function in log space according to the following formula:
Base image compressedx,y=2(C [log 2(Base_imagex,y)−log 2(Mtarget)]+log 2(Mtarget)) Where Base_imagex,y is the base image generated by applying said adaptive bilateral filter to said log luminance component of each pixel in said frame, C is the compression factor {0,1} and Mtarget is a target mean level; generating an enhanced luminance image for each frame of said video input signal by adding the detailed image to the compressed base image; generating a log gain map for each frame of a video input signal by subtracting the log luminance component from the enhanced luminance image for said frame; converting the log gain map to a linear gain map with an antilog conversion; and multiplying the linear gain map to each RGB channel of a video display independently.