An Image Processing Technique for Lane Path Detection in Palm Oil Plantation

Abstract

Image processing for lane detection is commonly utilized by researchers for autonomous navigation purposes. In this paper, lane path detection in palm oil plantations had been demonstrated by acquiring raw videos using a vision sensor. The videos were recorded on two different paths in a palm oil plantation labelled Path 1 and Path 2. Various image processing techniques had been utilized inclusive of RGB to HSV conversion colour space, Gaussian Blur, Canny Edge Detection, Region of Interest and Hough Transform. Then, a histogram graph was used to assess the performance of the lane path detection by varying the brightness level of the images. With the histogram graph value entrenched, it shows that the level brightness of 0 and 50 shows the ideal performance of lane detection for both Path 1 and Path 2 in contrast to other brightness levels. The outcome justifies a suitable brightness value must be set to achieve a good detection result. Nonetheless, further advancement to the program of lane detection is required to intensify its functionality when it encounters effects from the environment particularly, illumination from the sunlight, shadows as well as the ground surface, such as drains and water puddles.