Digital Image Processing plays an indispensable role in many medical applications by automating the imaging processes. At present, a variety of semi-automated and automated methods are applied for medical diagnoses using medical images. Eventually, different digital image processing algorithms are likely to be used in different biomedical applications. Widely, RGB color model is used for analyzing medical images. However, Red Green Blue (RGB) color model is not suitable for wound detection analysis since wound detection strongly depends on the light intensity. Implementation of wound detection using K-means clustering technique also holds certain disadvantages when it is combined with the MATLAB tool. The number of clusters (K-value) cannot be predicted easily which makes the process tedious and not suitable for real time wound detection analysis. HOG algorithm’s is used for detecting human and wound by means of open CV. The existing system implements this algorithm using java script which reduces accuracy and increases processing time. Tin this paper, Hue saturation value (HSV) and HOG algorithm are combinely used to detect the human and wounds from real time streaming video. A sliding detection window is used by the HOG Person Detector and HOG descriptor is measured at each location of the detector window. This descriptor is then applied to support vector mechanism (SVM), which either classifies it as "human" or "not human". HSV saturation value close to 0 wherever skin looks dull or gray which will detect the wound by getting red color. The proposed human and wound detection algorithm using HOG and HSV increases accuracy and reduces the delay in processing which can be implemented in military surveillance bots.
Volume 12 | Issue 6
Pages: 1897-1906
DOI: 10.5373/JARDCS/V12I2/S20201394