Study Of CMOS Vertical Integrated Pixel And Image Sensor

The classical color image sensor based on CFA(Color Filter Array) is prone to have color moire pattern when the imaging scene contains spatial frequency that is higher than nyquist frequency of the image sensor itself.The color moire pattern should be eliminated or suppressed to acquire high quality color image.The common optical low pass filter used to suppress the color moire pattern is expensive and ineffective sometimes,so people are always looking for a way which can eliminate color moire pattern radically.The vertical integrated pixel was proposed for this requirement initially,it utilized the wavelength dependency of absorption depth of incident light in silicon,i.e.light with shorter wavelength is absorpt near the surface while light with longer wavelength is absorpt at a deeper location.It can realize conventional color film like imaging ability where RGB signals are sampled on same pixel at different depths so it is totally color moire free;It can work with CFA to get more samples on a pixel,so we can get image with higher colorimetrical rendition and less color moire aliasing;It can be used in biometrical applications where monochromatic and intensity image are important;It can also be used in military applications where we need both VIS/NIR imaging ability.This thesis work is about basic research on principle,design,test and color correction processing of CMOS image sensor based on vertical integrated pixel, which is done first time in domestic community.The main results include:Firstly,we designed a CMOS large double junction vertical integrated test pixel and its QE(Quantum Efficiency) test platform,the QEs of double channel outputs were measured.We also designed a CMOS image sensor with double junction vertical integrated pixel which features 16μm pixel pitch,100×100 resolution,10fps readout, the design was manufactured and tested to verified its correctness.The main creative work in this part is miniaturization of the double junction vertical integrated pixel and make imaging without scanning mechanism possible while other groups' work in demostic were mainly focused on large single pixel only.Secondly,we used CMF(Color Matching Function) matching and color stimulus fitting to do the color correction for triple junction vertical integrated color pixel and double junction vertical integrated pixel with two color CFA respectively.We also did research on monospectrum imaging based on double junction vertical integrated pixel. The main creative work here is we proposed a input noise robust color correction scheme for high rank color stimulus fitting.Finally,we used analytical and numerical tools to model and simulate the vertical integrated pixel,these two models can help us to study the device and optimize the structure parameters,the accuracy of the analytical model for the double junction vertical integrated pixel was proven by measured QE data.The creative work is a VIS/NIR double junction vertical integrated pixel was developed by numerical simulation and its structure parameter was optimized.