FPGA design of efficient kidney image classification using algebric histogram feature model and sparse deep neural network (SDNN) techniques
Vinoth Rathinam, K. Bommannaraja · 2017
This work proposed a sparse learning optimization based kidney image classification using Sparse Deep Neural Network (SDNN) similarity measure. The kidney images are acquired and preprocessed to improve the image quality. Then the method splits the image into number of sectional image which Crops the Image for Number of Times Using Feature Extraction Method. Feature extraction using Algebraic Histogram based Sum and difference model to extract the 2d and 3d features of kidney image. The textures features are segmented by Algebric histogram (AH) method and the features are optimized using the hardware model. Based on the sparse classification depthness measure the method identifies the region which has affected or abnormal. At the classification stage, the method computes the sparse learning based pixel similarity measure to identify the most affected region and to perform classification. The entire hardware model can be split into two important levels namely Algebric histogram based Feature extraction and sparse learning based feature classification using SDNN technique. The identification of the kidney Abnormality in the image is displayed with colour for easy identification and visibility in monitor using HDL algorithms. The design and implementation in real time on both Field Programmable Gate Array (FPGA) using Xilinx System Generator (XSG) and Matlab 2013a.