- 學位論文
盲人避障輔助系統之設計
An Obstacle Avoidance Assisted System Design for the Blind
摘要
本論文提出了一個基於深度資訊的障礙物偵測方法。首先，應用膨脹與侵蝕，去除深度影像中細小的破碎的雜訊。代表地板分佈的V視差圖（V-disparity map）空間中，利用最小平方法（LSM）以一元二次多項式近似地板曲線，並決定地板高度的門檻值。接著搜尋深度圖像中變化劇烈且符合地板高度的門檻值的可疑樓梯邊緣點，運用霍夫轉換找出線段位置。而為了加強不同物件的邊緣特性與避免過度區域生長的問題，利用邊緣偵測將具有不同深度的物件邊緣移除。然後利用地板高度的門檻值與地面深度平緩變化的特性去除地板區塊影像；再以區域生長法將不同物件進行標籤，並分析每一個物件是否為樓梯，最後，將樓梯、樓梯邊緣和可能影響行走的障礙物，用語音的方式告知使用者其方向與距離。經室內與室外實驗證實本研究之實用性。
並列摘要
This paper proposes an obstacle detection method based on depth information. Firstly, we apply dilation and erosion to remove the crushing noise of the depth image.We use the least squares method (LSM) in a quadratic polynomial to approximate floor curves and determine the floor height threshold in the V-disparity. And then we search for dramatic changes depth value and in accordance with the floor height threshold to find out suspicious stair edge points. And we use Hough transform to find out the location of drop line. In order to strengthen the characteristics of different objects and overcome the drawback of region growing, we apply edge detection remove the edge. Then we use the floor height threshold and features of ground to remove the ground. And then we use region growing to label tags on different objects. We analyze each object and determine whether the object is a stair. Finally, if there is a stair、drop or obstacle, the system will tell the user its direction and distance with voice. The indoor and outdoor experiments confirmed the usefulness of this paper.