Analog Image Processing Using Gaussian Based Filtering for Motion Detection Applications

A novel approach to motion detection is presented in our study, utilizing a 3 × 3 Gaussian image filtering technique. Inspiration is drawn from the human visual system, particularly the human retina, and a straightforward Complementary Metal Oxide-Semiconductor (CMOS) circuit design is employed. Gaussian image filtering and noise mitigation are achieved through convolution and masking operations. The challenge lies in translating these mathematical operations into hardware using basic electronic components like current mirror and adder circuits. Additionally, enhancements are made to the 2-D pixel sensor array, optimizing the commonly used pixel circuitry.Our imaging technique conducts preliminary image processing before analog-to-digital conversion, supporting subsequent digital image processing stages in camera systems. We introduce an 8-MOSFET filtering cell array for retinomorphic analog image filtering. These cells collaborate to filter image signals received from the pixel circuit array. Our analysis is based on two moving scenes with a resolution of 150 × 150 pixels, and we compare the results with theoretical outcomes from digital image processing techniques. We utilize CMOS 0.18μm technology parameters for our proposed analog circuit structure.