Basic Computer Vision Operators Hardware Approach Model

The development of vision sensors at a lower cost has prompted the creation of new applications that are based on the processing of images and video streams. The automatic control systems are increasingly being supported by sensors of vision and image processing to solve problems of classification, matching and tracking. Due to the nature of the control applications, vision operators are required to run on embedded systems, depending on the type of processing can be very demanding for sequential systems so it is necessary to execute these tasks directly on hardware with what is possible to improve the execution times of the operator. The use of embedded systems that include sequential regions and regions of programmable logic in the same chip and the new tools for development of high level systems HLS allows to carry functions to circuit to accelerate the execution times. The present work describes the results and evaluation architecture of synthesizing to circuits basic computer vision operators, encapsulated in an IPCore to guarantee its portability and re-usability. Proposing a procedure that allows to accelerate the tasks of more complex operations on images when executing parts of these tasks in hardware.