This project demonstrates the enhancements that can be made to the detection of select objects by using HDR image processing instead of normal image auto-exposure settings. The output, an edge-detected image, can be directly compared against the auto-exposed edge-detected image and improvements can be observed.

1. Program the Artix-7 Digilent development board with the included files using ddr_dma_v2.xpr
2. Connect a VGA monitor to the appropriate port on the board.
3. Load .bin images containing multi-exposed images of the same scene into a microSD card using the naming convention d<a,b,c...>.bin
4. Open Xilinx SDK and build and run the program on the FPGA.

docs: Contains a copy of the presentation slides and final and report.

src/software/edge_generation: Contains matlab, python scripts used to compute the HDR images. Sample images used.

src/fpga/hdr_final/toplevel/ddr_dma_v2.xpr: The Vivado project for the hardware

src/fpga/hdr_final/toplevel/ddr_dma_v2/ddr_dma_v2.sdk: Contains the SDK project files for the Microblaze program

src/fpga/hdr_final/edge_new: Contains the Vivado project files for the object detection IP block

src/fpga/hdr_final/hdr_ip_sof_fix: Contains the Vivado project files for the HDR IP block

src/fpga/hdr_final/video_to_axis: Contains the Vivado project files for the video to AXI IP block

Yuanfang Li,

Rose Li,

Alex Papanicolaou,

Kenan Hu