International Journal of Applied Science and Engineering https://doi.org/10.6703/IJASE.202103_18(1).115 Vol.18(1) 2020115 OPEN ACCESS Received: June 5, 2020 Accepted: December 31, 2020 Corresponding Author: Sunny Arief Sudiro sunnyarief@yahoo.com Copyright: The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted distribution provided the original author and source are cited. Publisher: Chaoyang University of Technology ISSN: 1727-2394 (Print) ISSN: 1727-7841 (Online) Mean and variance statistic for image processing on FPGA Sunny Arief Sudiro 1* , Aqwam Rosadi Kardian 1 , Sarifuddin Madenda 2 , Lingga Hermanto 2 1 STMIK Jakarta STI&K Jl. BRI No. 17 Radio Dalam Kebayoran Baru Jakarta Indonesia 2 Gunadarma University Jl. Margonda Raya No. 100 Depok Jawa Barat Indonesia ABSTRACT Statistical formula processing an image data is commonly used in image processing. In software processing this formula and accessing data stored in memory is an easy task, but in hardware implementation, it is more difficult task due to many of constraints. This article presents hardware implementation of mean & variance statistic formula in effective and efficient way using FGPA Device. The design of circuit for both formulas proposed in this article need only two additions component (in two accumulators) and two shift-right-registers will be used for divisor circuits, one subtractor and one multiplier. In the experiment, processing an image size 8x8 pixels need 64 clocks cycle to conclude the mean & variance calculations. More than 1024 additions component is needed in some design so this design is more efficient. Keywords: Mean, Variance, FPGA, Accumulator, Counter. 1. INTRODUCTION A computer process for manipulating and analysing image is known as digital image processing. Basic statistic formulas that used in image processing using computer application are: histogram, variance, mean, etc. These all processes have high correlation with functions in pattern recognition based on features to recognize the object. Many research for the implementation of function involving large data and statistical calculation in FPGA device. Implementation variance value in image fusion using FPGA device, computation speed is one of the reason among others. This approach based on FPGA technology provides fast, compact and low power consumption for image fusion (Gade and Khope, 2016). The computation time or speed of process is one of the important things in processing large of data. Hardware implementation is one of solution, acceleration using FPGA presented in Iturk et al. (2008) for processing Markowitz’ mean variance framework. This approach gives a 221x speed ratio comparing to software implementation. Another research in FPGA acceleration is proposed in Betkaoui et al. (2011), this approach focus on re-configurable architecture or component in FPGA and partitioning data for processing large graph data, and the result is reducing execution time from 120 minutes to 12 minutes or 10 times faster for common bio-informatics algorithm. Research for mean calculation only using FPGA device is proposed by Kardian et al. (2016). In this article presents a method for mean calculation needs 1 addition operations, 1 division, 64 cycles for 8x8 image size. This work is actually the extended work based on previous method describe in Kardian et al. (2016). An interesting work is presented in Ismael and Mahmood (2017), this research proposed an implementation 6 functions of statistical operation based on FPGA, which are: mean, variance, standard deviation, Root Mean Square (RMS), covariance and Mean Square Error (MSE). This approach used 1090 number of slices, 220 numbers of