The Research On The Key Techniques For CMOS Image Sensors In Optofluidic Microscope

The emergence and development of optofluidic microscope(OFM)makes the portable and low cost cell detection possible.However,the research of OFM was still in its infancy.At present,the researches of OFM focus on the preparation of microfluidic channels and image processing.As an important imaging component,the performances of CMOS image sensor directly influence the further study of OFM system.Now,the optofluidic microscope platform most uses the commercial CMOS image sensor chip with high performance.However,the high performance commercial CMOS image sensors are not completely suitable for the optofluidic microscope application.In this thesis,the structures and performances of the CMOS image sensor used in the optofluidic microscope are studied,including the related double sampling circuit,high-speed column-parallel two-step analog-to-digital converter,ramp generator,bandgap reference and low dropout linear regulator.The solution and optimization method are put forward according to the technical difficulties.Finally,verified by chip test and system experiment.The main contributions are summarized as follows.a.A two-step continuous-time hybrid analog-to-digital converter is proposed.Through the high and low quantization improves the conversion speed of the ADC.The use of simple digital feedback circuit and fewer column circuit wiring resources,on the one hand to complete the high to low continuous conversion,on the other hand to reduce the area and power consumption of the un-column ramp generator circuit.In addition,a digital calibration algorithm is proposed in this thesis to improve the error accumulation from the high bit conversion to low bit conversion.Finally,designed and verified the advanced nature of the proposed ADC.The results show that compared with the conventional single-slope ADC,the conversion time of the two-stage continuous-time mixed analog-to-digital converter is decreased from 22?s to 4.5?s and the conversion speed is increased about 5 times.After the digital calibration,ADC static characteristics and dynamic characteristics have been greatly improved and no missing code phenomenon.The ENOB can reach up to 9.1-bit which meet the CMOS image sensor frame rate and accuracy requirements.b.The low noise and low power consumption of each module of the column-parallel CMOS image sensor is studied,including low-power readout circuit,low-noise correlated double sampling circuit with offset cancellation technique,high-order bandgap reference circuit,and a dynamic frequency compensated low dropout linear regulator.An optimization method for discrete-time integral-type ramp generator is proposed.By adjusting the charging voltage and the sampling capacitance,the linearity of the ramp generator is improved.In addition,a wide operating voltage range high-order compensation bandgap reference circuit is proposed.Through adding an op-amp and adjusting the current size,the temperature coefficient was reduced.On this basis,a CMOS image sensor system based on a two-step continuous-time mixed analog-to-digital converter is designed.Finally,test results show that the overall design to meet the requirements of low power and low noise.C.Aiming at the characteristics of lensless imaging of optofluidic microscope,a local contrast enhancement algorithm is proposed and applied to nonlinear CMOS image sensor system.The count frequency of different light intensity intervals is dynamically adjusted by using a configurable multi-frequency counter.Then,the local contrast of the cell image is improved under the condition that the resolution of the analog-to-digital conversion is unchanged.The comparison of the stretching contrast under different modes of the collected cell images was carried out by the optofluidic microscope system platform,and the feasibility of the algorithm was verified.The results show the cell has better image quality under the proposed local contrast enhancement algorithm.