Cet article compare les avantages et les inconvénients des modes d'obturation déroulante et d'obturation globale pour les caméras sCMOS. Ces deux modes sont des manières différentes d'exposer la caméra à la scène et de lire l'image. Le choix de l'un ou l'autre mode dépend de l'application et est décrit dans cet article. Certaines caméras sCMOS proposent une option logicielle permettant d'utiliser l'un ou l'autre de ces modes. Quelles sont donc les différences entre ces modes ?
In this mode, the camera area is divided into two halves, and each row starting from the center is exposed to the scene. The read-out wave travels from the center to the top in the upper part of the sensor and from the center to the bottom in the lower part of the sensor, so adjacent rows of the array are exposed at slightly different times. Since the maximum readout rate is typically 560 MHz, the time delay between adjacent rows is 10 ms. After the exposure of all the rows ends, the readout wave once again sweeps through the sensor and transfers all the charges from each row to the read-out node. For this process, the rows at the bottom and top of the sensor ( having typically 2000 pixels in one dimension), start their exposure 10 msec ( 1000×10 ms) later than the rows at the center. This could pose a problem if the image is moving really fast, such as the blades of a fan, because the scene is moving faster than the camera's rolling shutter can keep up with. The figure below shows a static fan and an image of a moving fan captured with a camera operating in rolling shutter mode.
Another disadvantage of this mode is that different parts of the exposed image are not perfectly correlated in time. It is also hard to synchronize the camera with an external trigger, and the procedure is very complicated. However, the rolling shutter mode may be suitable for applications where a small object is not moving very fast, and one can benefit from the low readout noise (typically 1.2 e) and the fast frame rate (100 fps) offered by this mode.
Global shutter mode: In this mode, the sCMOS camera operates like an interline CCD. All pixels are simultaneously exposed to light. Before the exposure, the pixel is kept free of charge, and after the exposure, the charges are transferred to the readout node. All parts of the image obtained in this mode are captured simultaneously. There will be no spatial distortion, and synchronization with an external trigger is quite simple. On the downside, the global shutter mode has higher read-out noise (typically twice that of the rolling shutter, ~2.5 e). Also, it requires a background image to reset the image, which halves the frame rate compared to an unsynchronized rolling shutter mode.
So, depending on the application and considering the speed, size, importance of temporal correlation, signal-to-noise considerations, frame rate requirements, etc., one or the other mode is selected. Figure below shows another picture of a moving car type where the distortion due to rolling shutter mode, appropriately termed as “rolling shutter effect” is observed. The global-shutter image shows no distortion.