This paper presents a hybrid-readout and dynamic-resolution CMOS image sensor targeted for object tracking applications. The proposed vision sensor can either work in an asynchronous motion detection mode or synchronous region of interest (ROI) readout mode, with different resolutions. In the first mode, relative intensity changes are monitored by a motion detection unit formed of 2×2 pixels and further diffused by a capacitance coupling network to reduce spatial noise. After tunable thresholding, active motions are converted into binary events and asynchronously delivered to the outside using an address-event-representation (AER) protocol. In the ROI extraction mode, this image sensor sequentially accesses each pixel in the interested region to report an analog intensity image with a higher resolution. This sensor has been implemented using a standard 0.18 µm CMOS process. Each pixel contains three capacitors and around 10 transistors, occupying a silicon area of 25×25 µm2, with a fill factor of ∼42%.