Scene Adaptive Exposure Time Control for Imaging and Apparent Motion Sensor

Abstract

CMOS image sensors have been developed for surveillance and industrial equipment cameras. In these applications, it is important to capture images with clear details for image recognition and object tracking. Since the exposure parameters (including exposure time, which is examined in this paper) are not appropriate, overexposure and underexposure may occur when the illuminance varies with artificial or natural light. Therefore, it is necessary to capture an image with the exposure time that is adjusted to the illuminance level of the scene. At the same time, motion blur must be suppressed when the camera or subject is moving. Many auto exposure algorithms adopt the average brightness value of the scene to control the exposure time. The method based on the brightness value in [1] adjust the exposure time for the important object such as a moving object. The weight of the segmented moving object is higher through object tracking, and control the proper exposure time for moving objects. However, this approach probably cause motion blur When the moving subject is dark and exposure time is set long. Imaging with a short exposure time can suppress motion blur, and the exposure time can be estimated by using motion estimation. High frame rate imaging is effective for improving the accuracy of motion estimation [2], [3]. The correlation between frames at high frame rate is so high that the computational complexity of motion estimation is reduced. However, many frames are needed to output from the image sensor to a signal processing circuit outside the sensor, and imaging at high frame rate increases the data rate. Several methods [4]–[7] mount a simple processing circuit