This so called region of interest (ROI) can be as small as one pixel. Reducing the size of the image transmitted by the camera reduces the image size and thus the bandwidth required for each image. In configuring an ROI the read out window is reduced to the area relevant for analysis, increasing the image frequency significantly.
For example, if a CMOS camera has a resolution of 1000 x 1000 pixels at a full frame rate of 100 fps, with an ROI of 200 x 500 pixels, there are only 1/10th of the total pixels to be read out, so the frame rate increases to almost 1000 fps. In reality there are other delays involved in reading the data from the sensor, thus the increase in the frame rate will not be exactly proportional to the reduction in resolution when compared to the full resolution frame rate.
More and more cameras allow multiple regions of interest (MROIs) to be read out from the sensor in order to perform multiple inspections. This is especially interesting for applications with high resolution cameras requiring extreme accuracy when only very small ROIs are required for analysis. There are even some models that allow the region of interest to be repositioned dynamically, which enables objects to be tracked within the field of view.
It should be noted that some CCD cameras offer ROI readout, however, this only reduces the data transmission time and not the sensor readout time as CCDs always have to read out complete lines.