IMPLEMENTATION OF ROBOT NAVIGATION USING CAMERA BY IDENTIFYING SIGNAGE

An autonomous robot can utilize computer vision to recognize signage that contain navigational information. Identifying and recognizing physical signage in various lighting condition remain a challenge and an issue to be resolved. This project looks at more closely at the image processing steps in order to improve identification of specific signage placed in a various lighting condition. The steps involve in identifying the signage are pre-processing, segmentation, feature detection and recognition of the images. By applying histogram equalization and also adaptive histogram equalization, the intensity of the images increases. The darker region becomes a bit brighter. After applying histogram equalization and adaptive histogram equalization, the canny edge detection is applied before using the Contour method to find the desired area. After finding the desired region of interest, the edges are calculated. The edges are part of the feature extraction methods. Based on the experiment conducted, it is found that
HSV perform better than grayscale for indoor environment without the present of ambient light. 100% if the signage is identify correctly if the light intensity is low or if there is an artificial light above the camera at high intensity. This result is better as compare to the previous result whereby binarization technique is used during the feature extraction phase. In terms of grayscale the result shows that even after histogram equalization, the signage cannot workout properly.