| High-speed image sensor is one of the main development directions of image sensor at present.With the development of storage medium and data transmission technology,the application field of high-speed image sensor and its imaging system is more and more extensive.The output data of the existing CMOS high-speed image sensor increases with the improvement of frame frequency and resolution.AddressEvent Representation Image Sensor is a type of high-speed bionic vision sensor.Compared with the traditional high-speed CMOS image sensor based on frame sampling,it only collects changes.It has the characteristics of low power consumption,less redundancy and less delay.However,its arbitration mechanism can lead to information loss in complex moving scenes,which makes the acquired scene information lack of integrity.Based on the existing problems of high-speed image sensor,this paper proposes an image sensing method using pulse sequence detection,builds a behavioral model of image sensor based on pulse sequence,and studies the image reconstruction algorithm in static sense and high-speed environment,which provides the research foundation for hardware implementation of image sensor based on pulse sequence.Firstly,the overall structure of the image sensor based on pulse sequence and the functions of each module are given,two different implementation modes of the pulse sequence are given,and the basic structure of the pixel unit is expounded.Various situations of the pixel triggering and the readout timing under different conditions are analyzed,and a behavioral model of the pixel unit and the pulse sequence sensor is built.Secondly,based on the behavioral model of image sensor based on pulse sequence,aiming at two different application scenarios of static or low-speed and high speed,the image reconstruction algorithm based on pulse accumulation in fixed time and the image reconstruction algorithm based on trigger interval of pulse sequence are designed.Giving the image reconstruction algorithm based on pulse sequence trigger interval,the causes of the tail phenomenon are studied,and the tail elimination algorithm based on the corresponding interval of object light and the algorithm based on calculation are designed to eliminate the tail phenomenon in reconstruction.Finally,the system and the software of image sensor based on pulse sequence is designed.The software is used in reconstruction,real-time display,and data storage of the pulse sequence,meanwhile a graphical user interface of the software is written.The algorithms designed in this paper are verified.1.In the experiment of static image reconstruction,the minimum mean square error between the reconstructed image using two pulse sequence layouts and the original input image are 0.074 and 0.008 respectively under low illumination condition.2.In the dynamic image reconstruction experiment,the mean square error between the reconstructed high-speed video and the original input is less than 30 in the maximum input photocurrent range of 6000-16000 fA.3.The two kinds of trailing elimination algorithm can reduce the mean square error of reconstructed video to original input video to less than 30.To sum up,the image reconstruction algorithm proposed in this paper can better reconstruct the data format for the output of the image sensor based on pulse sequence and the algorithm is applied to the pulse sequence camera imaging system. |
