The Effect of Grayscale, CLAHE Image and Filter Images in Convolution Process

Convolution process for Convolutional Neural Network (CNN) need a source image and a kernel. Kernel is a filter image with a certain size, such as 3×3, 5×5, 7×7, and so on. The source image or input image is an image in RGB or processed image. The processed image are in color model or grayscale format. The color models are such as RGB, HSV, CMYK and YIQ. Whereas the grayscale format such as Contrast Limited Adaptive Histogram Equalization (CLAHE). In this paper, we demonstrate the effect of grayscale and CLAHE formatted image and several filter images in convolution process, both for static image and real time frame captured by a camera. The experiment is conducted using programming language C++ and openCV library. The experiment results show the distinction between the various of input images and various of the filter images.