基於相似性計算之頭肩偵測與應用於復健系統之動作測量

Similarity-Based Weighted Head-Shoulder Detection and Gesture Measurement for Rehabilitation

黃喬健(CHIAO-CHIEN HUANG)

指導教授 ： 林維暘

國立中正大學/工學院/資訊工程學系/碩士(2016年)

摘要

本篇論文提出使用相似性計算所得到之權重樣板來比對相似的動作,以及偵 測正面行人的頭肩區域,其主要的想法為這兩個系統皆能以相似性為基礎的演算 法來達成不錯的效果。在頭肩偵測系統裡面有兩個程序。第一個程序為人臉偵測, 使用人臉偵測來找出影像中人臉的區域,並在該區域抓取梯度直方圖 (Histogram ofOrientedGradients,簡稱HOG) 特徵後,進入第二個程序。第二個程序為相似 性為基礎的權重計算,該方法提供一個簡單且穩定的方式來過濾掉影像的雜訊。 實驗結果使用少量的訓練影像並且擁有比 HOG 還要優越的效果。動作測量系統 則是比對兩個不同的動作,並使用相似性導向的動態時間扭曲(Dynamic time warping,簡稱 DTW ) 來比對兩個不同的動作。其實驗結果則是與傳統 DTW 進 行比較,且得到更好的效果。

關鍵字

頭肩偵測 ； 分類器 ； 動作測量

並列摘要

This paper is proposed two methods to detect the head-shoulder area of a frontal pedestrian by using a similarity-based weighted template and measuring the similar gestures. The common idea of these two systems is that similarity-based algorithms. There are two procedures in head-shoulder detection system. The first procedure is Viola-Jones face detection, which focuses on the face’s position and then generates the corresponding local HOG (Histogram of Oriented Gradients) descriptor. The second one is the similarity-based weighted estimation, which provides a perfect but easy way to filter out the feature noise. The experiments of classification and verification experiments use several positive images to demonstrate the improvement in the HOG resulting from this two-step methodology. Gesture measurement system matches two gestures which are consisted of lots of sequences by using similarity-ordered DTW (Dynamic time warping). The experiments of classification experiments demonstrate the improvement in the conventional DTW.

並列關鍵字

Head-shoulder detection ； Classifier ； Gesture measurement

參考文獻

[2] Arulampalam, M. S., Maskell, S., Gordon, N., Clapp, T., “A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking,” IEEE Transactions on Signal Processing, vol. 50, no. 2, pp. 174-188, 2002.

[5] Canny, J., “A computational approach to edge detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 6, pp. 679-698, 1986.

[7] Chou, C. H., Fang, S. J., Hsiao, Y. C., Chen, H. L. “A home rehabilitation system combined with somatosensory game for stroke patients,” ICIC Express Letters, vol. 7, no. 3, 2013.

[8] Dalal, N., Triggs, B. “Histograms of oriented gradients for human detection,” IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, no. 1, pp. 886-893, 2005.

[11] Felzenszwalb, P. F., Girshick, R. B., McAllester, D., Ramanan, D., “Object detection with discriminatively trained part-based models,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 32, no. 9, pp. 1627-1645, 2010.