Search

This paper presents a non-destructive readout circuit of the linear array image sensor with wide dynamic range and high speed readout for radar system. A multi-capacitor and self-regulated capacitive trans-impedance amplifier (CTIA) structure is employed to extend the dynamic range. The gain of the CTIA is auto adjusted by switching different capacitors to the integration node asynchronously according to the output voltage. A class AB OPA is utilized to drive all the additional capacitors to achieve high speed readout. A photo response curve presents as a polyline with 5 segments, which enables a 101.7 dB dynamic range. In addition, the exposure time is 5.12us in the simulation, then an over 190k fps is achieved.

In modern image processing, due to the development of digital image processing, the focus of the sensor can be automatically set by the digital processing system through computation. In the other hand, the auto-focusing synchronously and consistently is one of the most important factors for image mosaic and fusion processing, especially for the system with multi-sensor which are put on one line in order to gain the wide angle video information. Different images sampled by the sensors with different focal length values will always increase the complexity of the affine matrix of the image mosaic and fusion in next, which potentially reducing the efficiency of the system and consuming more power. Here, a new fast evaluation method based on the gray value variance of the image pixel is proposed to find the common focal length value for all sensors to achieve the better image sharpness. For the multi-frame pictures that are sampled from different sensors that have been adjusted and been regarded as time synchronization, the gray value variances of the adjacent pixels are determined to generate one curve. This curve is the focus measure function which describes the relationship between the image sharpness and the focal length value of the sensor. On the basis of all focus measure functions of all sensors in the image processing system, this paper uses least square method to carry out the data fitting to imitate the disperse curves and give one objective function for the multi-sensor system, and then find the optimal solution corresponding to the extreme value of the image sharpness according to the evaluation of the objective function. This optimal focal length value is the common parameter for all sensors in this system. By setting the common focal length value, in the premise of ensuring the image sharpness, the computing of the affine matrix which is the core processing of the image mosaic and fusion which stitching all those pictures into one wide angle image will be greatly simplified and the efficiency of the image processing system is significantly improved.

In this paper, a mathematical model of TDI CMOS image sensors was established in behavioral level through MATLAB based on the principle of a TDI CMOS image sensor using temporal oversampling rolling shutter in the along-track direction. The geometric perspective and light energy transmission relationships between the scene and the image on the sensor are included in the proposed model. A graphical user interface (GUI) of the model was also established. A high resolution satellitic picture was used to model the virtual scene being photographed. The effectiveness of the proposed model was verified by computer simulations based on the satellitic picture. In order to guide the design of TDI CMOS image sensors, the impacts of some parameters of TDI CMOS image sensors including pixel pitch, pixel photosensitive size, and integration time on the performance of the sensors were researched through the proposed model. The impacts of the above parameters on the sensors were quantified by sensor’s modulation transfer function (MTF) of the along-track direction, which was calculated by slanted-edge method. The simulation results indicated that the TDI CMOS image sensor can get a better performance with smaller pixel photosensitive size and shorter integration time. The proposed model is useful in the process of researching and developing a TDI CMOS image sensor.

In four-transistor (4T) CMOS image sensors (CIS), incomplete charge transfer from the photodiode (PD) to the floating diffusion (FD) node can result in image lag, which is a serious problem affecting the imaging performance. In the paper a low image lag 4T pixel structure for CIS is proposed. Two techniques are adopted to promote complete charge transfer in a 4T pixel. Firstly, the threshold voltage of the reset transistor in the 4T pixel is adjusted to an appropriate negative value to realize a high potential in the FD, which is helpful for the charge (electron) to transfer into FD. Inevitably, a large negative threshold voltage make the source-drain leakage current of the reset transistor can not be ignored. In the design the threshold voltage is chosen to satisfy the requirements of a higher potential in FD and a lower source-drain leakage current of reset transistor simultaneously. Secondly, an additional p-type layer is adopted on the surface of the photodiode, with partially overlapped the channel of the transfer transistor. With an optimized overlap length, neither an apparent potential barrier nor a severe potential pocket can be formed on the route of charge transfer. So a potential distribution under transfer gate conducive to charge transfer is achieved. An identical photomask is used to manufacture the additional p-type layer and the p-type pinned layer of the photodiode, and the latter is formed in selfaligned way, which is economic in process and helpful to control the misalignment of the layer. The simulations are completed in Technology Computer-Aided Design (TCAD) tools. A test chip with 32×10 pixel array has been designed and fabricated in 0.18μm 1P4M CIS process. The experimental results demonstrate that the image lag is below the measurement threshold (using 12-bit ADC) with an additional reset operation adopted. Without the additional reset, the largest measured image lag is 0.18%.

A novel five transistor global shutter CMOS active pixel with ultra-high dynamic range is presented in this paper. A global shutter control transistor is added to traditional four transistor pixel. The five transistor pixel image sensor works in global shutter mode to shoot high speed moving object with dual sampling to eliminate fixed pattern noise. The image sensor can restore four transistor pixel rolling shutter mode with global shutter control transistor shutoff to shoot stationary object with correlated dual sampling to eliminate fixed pattern noise and random noise. A digital control stepped reset-gate voltage technique with no additional components to increase dynamic range by the compression of charge integration characteristic curve and to implement anti-blooming by discharging excess carriers is adopted. Simulation results show that the image sensor can work in global shutter mode and the dynamic range is increased approximately by 20dB than typical CMOS image sensor.

This paper deals with a complete analytical modeling and analysis of thermoelectric uncooled infrared sensors compatible with CMOS technology. The model put forward is based on dividing the sensor into three zones, each one being the subject of a thorough thermal study (conduction, convection and radiation thermal effect). Through the analytical thermal gradient analysis developed in each zone of the structure (absorber, part of thermoelectric transducer layer placed under the absorber, thermoelectric transducer) we are able to predict the sensitivity, detectivity and the stability power to the sensor. Thus, such a kind of analytical approach is worth of interest to optimize thermopile sensor design parameters.

A novel APS pixel with full frame self-storage and motion detection capabilities is proposed in this paper. Taking advantage of adding the independent exposure transistor, the FD node of traditional 4T pixel can be used for temporarily storing the integrated charge between two successive frames. Before the current frame's reset operations, the charge of previous frame will been readout and compared with the subsequent signal of current frame to locate the motion pixel in the motion detection comparators which is integrally shared by columns and have a lower load capacitor and higher gain than the previously published counterparts. The pixel and its peripheral circuit are designed and simulated using SMIC 0.18μm MM/RF 1P6M process. The simulation indicates that the image sensor using proposal design can achieve the motion detection without obvious reduction of fill factor and have a higher accuracy of detection and lower power consumption, which is more suitable to the application of surveillance and remote video communication network.

A real-time enhanced technique which can accelerate the processing time and provide a more distinct image using dynamic area threshold was presented. It can compensate the flaw of the image which captured by CMOS sensor. According to the value of statistic result, dynamic increasing algorithm would be used to change the contrast and brightness via local area pixels enhancement while pixels' real value is larger than the upper threshold. On the converse, if the real value of pixels is below the lower threshold, dynamic decreasing algorithm would be used in them to darker these pixels more than these which are above the upper threshold value. The edge of the dynamic area would be analyzed and provide an appropriate way to process these edge pixels, which will let them be the usual gradual edge in the dynamic area. And this algorithm is implemented in FPGA of Xilinx XUPV4 with an MCU which can control the area where we are interested in to increase the quality efficiently.

This paper describes a polysilicon pressure sensor fabricated by silicon micromachining technology. The starting materials of the sensor are (100) orientated monocrystal silicon wafers with double side polishing. The semiconductor integrated circuit process is also used in the testing. The polysilicon piezoresistors are deposited by LPCVD on the thermallygrown oxide film of the silicon wafer. In order to increase sensitivity, a so-called "twin isles" structure has been used. This type of structure consists of a square silicon diaphragm with twin square isles, which are formed by anisotropic etching into the back side of the monocrystal silicon substrate wafer. A main problem in preparation is the convex corner undercutting effect in the etching process. The relationships between the corner undercutting rate and the etching depth in TMAH solution have been obtained experimentally. Two masks of the special pattern are designed to compensate for corner undercutting. 1). A compensating angle bounded by