Interpolated histogram method for area optimised median computation
Kaushal D. Buch, Anand D. Darji · International Journal of Electronics · 2012
The article describes an area efficient algorithm for real-time approximate median computation on VLSI platforms. The improvement in performance and area optimisation are achieved through linear interpolation within a reduced number of histogram bins. In order to reduce the hardware utilisation further, an approximation technique for interpolation is also proposed. This approach extends the utility of the histogram method to data sets having a large dynamic range. The performance of the proposed algorithm in terms of mean squared error (MSE) and resource utilisation is provided and compared to that of the existing algorithms. This comparison indicates that more than 60% optimisation in resources is achieved with marginal compromise in the accuracy of the median. The proposed algorithm finds applications in the areas of image processing, time series analysis and median absolute deviation (MAD) computation.