Free Space Measurement for Blind Person using Histogram Equalization and Adaptive Region Growing
Hari Raksha K Malali1, Nithin Kashyap2, Natesh M3
1Hari Raksha K Malali, Department of Computer Science and Engineering, Vidyavardhaka College of Engineering, Mysuru (Karnataka), India.
2Nithin Kashyap, Department of Computer Science and Engineering, Vidyavardhaka College of Engineering, Mysuru (Karnataka), India.
3Natesh M, Department of Computer Science and Engineering, Vidyavardhaka College of Engineering, Mysuru (Karnataka), India.
Manuscript received on 15 October 2019 | Revised Manuscript received on 24 October 2019 | Manuscript Published on 02 November 2019 | PP: 2459-2467 | Volume-8 Issue-2S11 September 2019 | Retrieval Number: B12880982S1119/2019©BEIESP | DOI: 10.35940/ijrte.B1288.0982S1119
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Although significant research has been conducted to improve the standard of life of the blind person by using recent methods, it continues to be adequate for blindness rehabilitation. Increasingly, the computer vision system is being used to improve the performance of blind life’s. In this paper, a technique is designed using histogram equalization and adaptive region growing to aid the blind person by giving the person information about the free space apart from the obstacles around him in all the directions for better mobility. It encompasses three modules, (i) histogram equalization (ii) segmentation and (iii) Kalman filtering. Histogram equalization module employs a canny edge detector to detect edges and then goes through with histogram equalization. Segmentation module makes use of an adaptive region growing for the segmentation process. Kalman filtering and comparison is made use of the final module to calculate the free space available. The input image is Kalman filtered and compared with the segmented image to have the free space calculation. The comparison is carried out with the help of the OR operator and the resulting figure give the free space. The proposed strategy is assessed under standard assessment measurements of False Positive, False Negative, True Positive and True Negative, explicitness, affectability, and precision for various group sizes. The simulations results obtained are plotted. This resulted in the following observations highest specificity, sensitivity and accuracy came around 0.90, 0.50 and 0.71 and similarly, average TP, TN, FP and FN came about 0.79, 0.5, 0.5 and 0.20 respectively. The high evaluation metric values indicate the good performance of the proposed technique in the area.
Keywords: Free Space Measurement, Adaptive Region Growing, Kalman Filter, Canny Edge Detector, Histogram Equalization Segmentation, Floor detection.
Scope of the Article: Adaptive Networking Applications