*Processing photographic images is important in many applications, among them the development of automated driver assistance systems (ADAS) and autonomous vehicles. Many techniques are used for processing images, including neural networks, other types of machine learning, and edge detection. One common issue with processing these photos is the presence of noise, whether caused by the camera itself or by physical conditions (e.g., weather conditions or dirt on road signs). In this paper, a neural network is used for noise reduction to improve edge detection results and tested with two kinds of noise, Gaussian and salt & pepper noise, and three different edge detection algorithms, Canny, Sobel, and Zhang. Results showed that the noise reduction process was effective in improving performance of the edge detection process, with the exception of conditions where the noise was originally very minimal.Processing photographic images is important in many applications, among them the development of automated driver assistance systems (ADAS) and autonomous vehicles. Many techniques are used for processing images, including neural networks, other types of machine learning, and edge detection. One common issue with processing these photos is the presence of noise, whether caused by the camera itself or by physical conditions (e.g., weather conditions or dirt on road signs). In this paper, a neural network is used for noise reduction to improve edge detection results and tested with two kinds of noise, Gaussian and salt & pepper noise, and three different edge detection algorithms, Canny, Sobel, and Zhang. Results showed that the noise reduction process was effective in improving performance of the edge detection process, with the exception of conditions where the noise was originally very minimal. The noise reduction was greatest, for both types of noise, for the Canny algorithm, which featured the largest amount of noise pollution in the detected edges to begin with. Good results were also seen for the Sobel and Zhang edge detection methods, with the aforementioned caveat.*


Effect of Neural Network on Reduction of Noise for Edge Detection

Division / Sponsors: 
Dynamic Systems & Control (DSCD)

Paper Proceedings: https://asmedigitalcollection.asme.org/DSCC/DSCC2020/volume/84270

DSCC2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.