IPs Design of FPGA for High-Resolution Image Process

Abstract

With the progress of science and technology, the image resolution of monitoring systems is entered the Full-HD (High Definition) era. The Full-HD refers to resolution of 1920X1080 (1080p), and the image refresh frequency is 30fps (Frames Per Second). Therefore, this thesis uses FPGA as image system processor, as well as uses AFE to implement a camera module. First, the dual-channel serial signal from AFE is converted into single-channel parallel signal, and the image is processed following our image processing procedure before charting and export. The image processing procedure includes AEC, auto color balance, color filter array interpolation, color correction, noise filter, gamma correction and dynamic range correction. Our algorithm counts information instantly, and calculates better correcting function content, so that we can display the brighter and darker parts of images clearly in different environments. Compared with a part of CCD digital cameras on the market, our ISP has better performance in SNR. Finally, the AWB mean error of our ISP design is less than 0.02%, it is difficult to be perceived by human eyes.

Implementation of a parallel-beam optical-CT apparatus for three-dimensional radiation dosimetry using a high-resolution CCD camera

“…Then, the mean error of auto white balance for the image system processor (ISP) with a better performance in SNR, has to less than 0.02%, which is difficult to perceive by the human eye [21]. Moreover, the CCD image sensor in the photoelectric conversion process results in a frequency-independent white Gaussian noise [22,23], and Veerakumar and Esakkirajan proposed an adaptive median filtering algorithm that can reduce the impact of such white Gaussian noise [24]. Moreover, the white noise, bad pixels, and salt-and-pepper noise can be decreased by a smoothing filter [25,26].…”