| An image processing system has been designed to digitize and display either TV camera images or video tape recorded images. The image digitizer and display controller has been designed using a 6502 microprocessor and has a resolution of 256 x 256 picture elements and has 16 gray levels. It consists of a direct memory access controller for digitizing and display, a parallel port of transfer data to and from PDP-11/34. A crystal controlled sync generator and a sync seperator provide the required signals for video digitizing or displaying.; Several image processing programs have been developed for image smoothing, image thresholding, edge detection, edge thinning and boundary tracing, in addition to system support programs such as digitizer control and data transfer programs. The boundary tracing algorithm uses a heuristic cost function with parameters such as edge magnitude, edge phase, curvature and circular tangent, in conjunction with a best-first tree search technique. A noval technique that uses the edge phase information to separate the touching cells has also been developed. The digitized images can be processed either automatically according to a preprogrammed sequence with no user intervention or can be processed in an interactive manner.; Experiments were conducted to quantitate pupil dilating effects of drugs such as atropine on rabbits. A video tape recording was made of the rabbit's eye before and after drug treatment for a period of about 1 hr. These recorded images were digitized at desired intervals and stored on the disc. This data set can be processed automatically. The area, perimeter and diameter are calculated. The processing time is in the range 4-6 minutes per image, for relative measurements. If the system fails to recognize the object in an image, these frames can be processed interactively. Otherwise, the system is capable of performing all operations unattended. The failure rate is dependent upon lighting conditions and glare in the eye during video recording. The image processing system can be tailored to a particular application such as semi-automatic measurement of cell sizes from electron micrographs, just by rearranging the various image processing programs developed during this research. |
