針對視障者定向輔助之視覺辨識系統

視障者由於缺乏視覺的感官能力，因此對於瞭解週遭的環境這方面相當地薄弱，能夠安全地在不熟悉的戶外環境行走，對於視障者來說變成是一種挑戰。在現階段中，視障者最普遍使用的輔助系統為導盲犬與視障者手杖，然而，這些方式都包含著許多的缺點，並且其所能感知週遭物體的資訊和範圍十分有限。為了幫助視障者並改進其生活的品質，我們以電腦視覺技術為基礎設計一個電子導盲輔具，幫助視障者行走時提供重要之環境資訊。本論文使用尺度及旋轉不變特徵之物件辨識系統，讓視障者朋友行走於戶外環境時，辨識周遭環境如店家的招牌，然後透過語音告知視障者目前觀測到的環境訊息。由於在一個場景中可供辨識的特徵數量非常龐大，且尺度及旋轉不變之特徵描述都是高維度向量資料，因此我們在特徵資料庫裡建立一個索引結構，讓本系統擁有效率且能支援多重高維度特徵之搜尋能力，另外在建特徵資料庫時，利用特徵點篩選及合併的方法，把相似的特徵合併以減少比對次數，進一步提高比對的效率。

關鍵字

視障者輔助；物件辨識； SURF ；特徵索引

並列摘要

Due to lack of visual sense, the visually impaired people is unable to have good comprehension of surrounding. Safe navigation in non-familiar outdoor environment is a challenging for the visually impaired people. The most commonly used mobility tools are guide dogs and white canes. There are some disadvantages to these mobility tools which have limited usability in recognizing surrounding objects. In order to improve the quality of living for the visually impaired people. We design an electronic mobility tool which can perceive visual information of the environment for the visually impaired people. In the thesis, we develop a visual recognition system with scale and rotation invariant feature that allows the visually impaired people easily finding things or important landmarks outdoors and tell the user information of the environment with voice. In the system, we create an index structure that allows efficient multiple high-dimensional feature search. In order to decrease the number of comparisons, we decrease and merge each feature points with high similarity. According to the test results, the method can improve the search performance.

並列關鍵字

The visually impaired assistant ； SURF Feature ； Indexing ； Object Recognition

參考文獻

[12] 林家儒，運用顏色與SURF關鍵點特徵分類之快速多廣告看板計次系統，碩士論文，國立臺北科技大學，臺北，2010。

[13] 張厥煒、張宇宏，「基於顏色與SURF特徵之球場廣告看板搜尋與計次系統」。臺北科技大學學報，第四十四之二期，2012，第17-28頁。

[3] S. Belongie, J. Malik, and J. Puzicha, “Shape matching and object recognition using shape contexts,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 4, 2002, pp. 509-522.

[4] D. G. Lowe, “Distinctive image features from scale-invariant keypoints, Cascade Filtering Approach,” International Journal of Computer Vision, vol. 60, no.2, 2004, pp. 91-110.

[6] T. Gevers and A. W. M. Smeulders, “Color-based object recognition,” Pattern Recognition, vol. 32, 1999, pp. 453-464.

國際替代計量

針對視障者定向輔助之視覺辨識系統

全文下載

主題瀏覽