You have three options:
1. Use standard OpenCV library which can be included in Petalinux rootfs. After that you can develop your application using OpenCV functions.
2. Use xfOpenCV library from Xilinx. Find the documentation here. Some examples here. It offers a subset of standard OpenCV library functions, modified to be easily accelerated in FPGA.
3. Use standard OpenCV and xfOpenCV libraries. This options is suitable for more complex algorithms where you process some functions on ARM processor and another ones in FPGA for better performance.
For all mentioned options you have to implement the image acquisition mechanism from a camera if you don't intend to use static images.
Usually, the easiest way to develop video processing apps using xfOpenCV and OpenCV on Xilinx SoC is SDSoC. Find more info here. Using SDSoC you have a hardware platform which describes your hardware configuration and another important aspects like libraries, sample projects, etc. Usually, this platform is provided by the development board manufacturer, in this case Digilent. Two SDSoC platform aspects are important for you:
1. The hardware configuration which implements the image acquisition mechanism. Find Digilent Zybo Z7-20 SDSoC Platform here.
2. Sample projects. Find sample projects for the above mentioned platform here.