应用于CMOS图像传感器的联合降噪插值算法研究

Research on the Interpolation and Noise Reduction Algorithms in CMOS Image Sensors

摘要:

提出了一种可应用于CMOS图像传感器的联合并行图像预处理算法，其中，色彩插值算法采用双线性插值法和边缘导向插值法相结合的方式来完成色彩还原，而降噪算法采用带阈值的空间自适应滤波降噪算法，在硬件语言描述算法时，将两种算法的处理过程进行了联合处理。实验仿真结果表明，通过对算法进行ISE仿真和Matlab显示处理，得到色彩饱和度较高的彩色图像，峰值信噪比（PSNR）为30~36dB。整个处理过程有效地简化了片上系统（SOC）的图像处理过程，提高了算法的运算速度，节省了芯片面积且降低了系统功耗，为CMOS图像传感器的片上系统集成化设计提供了技术支持。

Abstract:

Proposed is a joint and parallel image preprocessing algorithm applied in CMOS image sensors, including the color interpolation and noise reduction algorithms. The color interpolation algorithm uses the combination of the bilinear and edge-oriented interpolation to accomplish the reproduction of color, and the algorithm of noise reduction uses the filtering algorithm with space adaptivity and threshold. When using the hardware description language to achieve the algorithms, the two algorithms will be achieved by a combined processing. The simulation results show, through making process of ISE simulation and Matlab display, the color images with higher color saturation were restoreds with PSNR of 30~36dB. The entire process can effectively simplify the image processing of the SOC, improve the operation speed of the algorithm, and also save the area of chip and reduce the power consumption of system.