Combination of Sobel+Prewitt Edge Detection Method with Roberts+Canny on Passion Flower Image Identification

Abstract

Edge detection is at the forefront of image processing for object detection, so a good understanding of edge detection algorithms is essential. This paper aims to analyze the ability of combined edge detection methods to identify images, through a comparison of two different edge detection methods, namely the combination of Sobel and Prewitt (Sobel+Prewitt) with Roberts and Canny (Roberts+Canny). The analysis process uses a dataset of Passion Flower Images obtained from the United States Department of Agriculture (USDA) Plant Database, Natural Resources Conservation Service (NRCS). The Image dataset was obtained using a Nikon Coolpix 995 camera, JPG format with a resolution of 128 192 pixels. Based on the analysis and testing, the results of the research using the combined edge detection technique of Roberts and Canny resulted in better image identification accuracy compared to Sobel and Prewitt. The average accuracy was 92.84% versus 68.75%.