Research Of Pulse-based Bionic Image Sensor And Its Error

With the improvement of the demand for high-speed and high-precision sensors in autonomous driving,military and other fields,a series of shortcomings of the framebased image sensor are gradually exposed,such as large time error and redundant data.Then the concept of bionic image sensor was proposed.And compared with the framebased image sensor,the pluse-based bionic image sensor has the advantages of less output data,high time precision and low power consumption,and can be applied to fields of dynamic detection and high-speed target tracking.However,as the pixel array increases and the scene complexity increases,the output time error and the quantization error of the pulse-based bionic image sensor begin to increase,and the development of the sensor encounters problems.Therefore,the pulse-based bionic image sensor and its error are analyzed and studied in detail.Firstly,the two pixel structures of pulse-based bionic image sensor and the working principle of each module are studied,and the behavioral level model of the sensor systems are completed.And the asynchronous pulse pixel adopts AER asynchronous readout mode,and eliminates redundant static background information from the source by outputting only the change information;horever,the synchronous pulse pixel adopts the synchronous readout mode,and the data is compressed by converting the scene information into time domain pulse signals.Then based on the analysis of the overall structure of the sensor,the circuit and layout of two pixel arrays are fabricated in 0.11?m process.Finally,the output error of the pulse-based bionic image sensor is analyzed.1.Based on the analysis of asynchronous pulse pixel transmission delay,a fast signal reading method is proposed.The method can effectively reduce the event transmission delay,besides,the verification results show that the transmission delay is reduced by79% under the condition that pixel array of 128×128 and all pixel are triggered,and the improvement effect is better when the pixel array is increased.2.Based on the error model and chip test results,the quantization error of synchronous pulse pixel is analyzed in detail,and the influence of sensor parameters on quantization error is summarized.The results show that the parameters mainly affect the error level of PFM quantization error and the fluctuation of PWM quantization error.And when the illumination increases from 100 lux to 780 lux,the PFM quantization error increases from-2.5% to 8%;the fluctuation range of the PWM quantization error increases from0.1%-4% to-5%-24%.In addition,under the condition that the illumination is 300 lux,selecting the reference voltage and reset voltage by the model prediction result can obtain the minimum PFM quantization error(6.5%)and the minimum fluctuation range of PWM quantization error(11%-15%)in actual work.