Downloadable ContentDownload PDF
SoC Implementation of Blood Cell Detection on Zedboard
This project is to design a device that will help detect leukemia through an image taken by a medical microscope. The device performs image processing on an image of blood cells and return a blood cell count. Similar projects exist but are almost entirely implemented through MATLAB. This project is to be implemented mostly on hardware using a Xilinx Zynq SoC using minimum software implementation. The image processing consists of two techniques: enhancement and detection algorithms. Before image processing is performed only the green pixel data is transferred to the programmable logic(PL) side of the Zynq. An algorithm like shape-based features is implemented on the PL to detect objects. Image enhancement algorithm implemented is to perform thresholding, massaging of the image, filling in objects, and erosion. The image detection is done by creating a box, finding the midpoint and identifying the minimum radius value from the midpoint out through four points. After objects are detected an overlay is colored over to show how accurate the detection. Information from image processing is transferred back to the software for analysis purposes.