在高能粒子物理探测器中，硅半导体像素传感器性能表现优异，尤其是单片式像素传感器具有高空间分辨率、低功耗、高速读出和低成本等特点。随着半导体工艺不断进步，单个像素面积越来越小（<50um*50um），像素阵列可达上百万个像素单元。传统的像素阵列采用逐行扫描读出模式，由于扫描时钟速率和功耗限制，积分时间很难减小到几十us量级。为进一步降低像素阵列功耗密度并提高读出速率：首先，在像素内将模拟信号数字化，因为相同负载条件下，驱动模拟信号传输比驱动数字信号传输消耗更多功耗；其次，对于上百万个像素阵列，粒子击中率可低为千分之一或万分之一，因此仅读出被击中像素单元能有效降低功耗同时提高阵列读出效率。像素阵列占传感器芯片面积的80%以上，因此降低像素阵列功耗密度能有效减小探测器系统热噪声以及探测器成本，本项目旨在研究先进CMOS CIS工艺下大像素阵列数据压缩低功耗高速读出方法。

In high energy particle physics detectors, the performance of the silicon semiconductor pixel sensor is excellent, especially the monolithic active pixel sensors, which can achieve high spatial resolution, low power consumption, high readout speed and low cost etc. With the progressing of semiconductor technology, a single pixel area is smaller and smaller (< 50 um * 50 um), the pixel array unit can reach millions of pixels. The traditional pixel array readout mode is the progressive-scan named rolling shutter, because of the limitation on the scan clock rate and the power consumption, the integration time is difficult to reduce to a magnitude order of dozens of us. In order to reduce the power consumption density of the pixel array and improve the readout speed further: First of all, digitize the analogue signal within each pixel, because for a same load condition, driving analogue signals will consume more power than driving digital signals; Secondly, for millions of pixels in an array, the hit density of particles can be as low as one over one thousand or one over ten thousands, so only readout the hit pixels can reduce the power consumption effectively and improve the readout speed at the same time. The area of the pixel array occupied more than 80% of the chip area, thus reduce the power density of the pixel array can effectively reduce thermal noise as well as the cost of detectors, this project aims to study data compression methods to read out data from a large pixel array with ultra low power and fast readout speed in advanced CMOS CIS technology.