Unclean Hand Detection Machine using Vision Sensor

Faradila Naim , Rawaida Jaafar, Nurul Wahidah

Arshad, Rosyati Hamid Faculty of Electrical and Electronics Engineering,

Universiti Malaysia Pahang, MALAYSIA

rawaidajaafar@yahoo.com , faradilan@ump.edu.my

Mohd Najib Razali

Faculty of Chemical Enginering and Natural Resources, Universiti Pahang Malaysia,

MALAYSIA

Abstract—This paper will discuss about the unclean hand detection using vision sensor for an automated hand wash screening system. Currently the hand wash screening audit is done manually by an expert to monitor the hand under ultraviolet light once it’s been washed. Hence, there is a need for more human experts to conduct the screening manually. This project is proposed to automate the hand wash screening audit by using a vision system. The vision system is designed to increase accuracy to detect the unclean area of washed hands. This system will not only detect the unclean area, but will also calculate the percentage of the unclean area which will be used as further analysis of the efficiency of the system. However, we need to build the hand wash prototype using ultraviolet light and a camera that is connected to the computer to process and display the results of hand wash screening.

Keywords— hand detection; color processing; uv light

I. INTRODUCTION Hands are the source of many infections especially

Nosocomial infection. A Nosocomial infection also known as a hospital-acquired infection (HAI), is an infection whose development is favored by a hospital environment, such as one acquired by a patient during a hospital visit or one developing among hospital staff. Such infections include fungal and bacterial infections and are aggravated by the reduced resistance of individual patients. Due to this issue, hand hygiene is essential to prevent cross-infection from the hospital or infection of health care facilities.

Hand hygiene is one of the areas in the field of infection control. It is simple and the best way to prevent infection and illness. The general indicators for hand hygiene are when hands are visibly soiled or not soiled before and after a healthcare worker's contact with the patient’s skin.

Infectious complications are frequently found among critically ill neonates. Hand hygiene is the leading measure to prevent healthcare-associated infections, but poor compliance has been repeatedly documented, including in the neonatal setting. Hand hygiene promotion requires a complex approach that should consider personal factors affecting health care workers' attitudes [1].

A Centers for Disease Control (CDC) has developed several hand hygiene resources for patients and healthcare providers. The center has been doing research and

development in reaching the national attention to disease prevention including microbial infections that related to hand hygiene to increase the betterment of health for the people of the United States [2].

According to a survey conducted by the U.S. Centers for disease control and prevention, forty million Americans contract illnesses every year due to the bacteria on the hands and around eighty thousands of them die. Looking at these alarming hand washing facts, one can conclude that it is very important to keep one's hands thoroughly washed and clean at all times, to prevent illnesses and infectious diseases [3].

Normally, the decision of hand screening audit refers to the Figure 1 below for the clean or unclean hand. The figure shows frequently missed areas of the stains in the palm of the hand. This figure is taken from the Ministry of Health Malaysia. Based on this figure, the red area of the hand refers to the most frequently missed areas while the green area refers to the not missed areas while washing the hands [4].

Fig. 1. Frequently Missed Areas

To reduce the risk of infections, currently the hand wash screening audit was done manually among the hospital staffs to reduce the risk of patients’ getting infected by the care providers unclean hand. The goal of this paper is to detect the area of unclean hand after being washed properly.

978-1-4673-6195-8/13/$31.00 ©2013 IEEE

II. METHODOLOGY

A. Hardware Development 1) Hardware Set Up The unclean hand wash detection prototype is build using

ultraviolet light and a camera that is connected to the computer to process and display the results of hand wash screening. The unclean hand wash detection prototype is designed as shown in Figure 2.1. Figure 2.2 shows the measurement to develop the hand wash prototype.

Fig. 2.1. The hand wash prototype view from the front and inside.

Fig. 2.2. The measurement of hand wash prototype

In this project, we use ultraviolet light because GLO

GERM is a chemical that is stimulating the spread of germs under the ultra-violet with the condition is the surrounding must opaque.

Data acquisition process starts with capturing images of washed hands after using the GLO GERM inside the prototype under the ultraviolet light.

2) Design Specification For this project, we set up the distance between the

camera and ultraviolet light position to be 7cm to reduce the reflection and to capture the full hand images. This distance is adequate and appropriate to capture and process the image by hand. The distance between the ultraviolet light is fixed to 13 cm based on the image acquisition where it concludes to capture the full image by hand. The distance between camera and ultraviolet light position is shown in the Figure 2.3.

The Perspex is covered with double layered black spray in order to have an opaque impact inside the box with the addition of the mounting board inside the box surface. A slight hint of light will interfere on the images captured that will lead to wrong unclean hand detection. Hence, the Perspex is layered with the black spray from the front site and inside site. The mounting board surface is opaque black color in accordance with the color of the prototype interior.

B. Software Development 1) Image Pre-Processing Technique The infected area of hands image needs an enhancement

by sharpening it and by bringing out more of the infected area detail. The image which is in RGB format was converted to 255 level grayscale. A grayscale image is also called a grayscale, grayscale, or gray-level image. The result of this process is described in Figure 2.4 below.

Then, we filter the image to create a predefined 2-D filter. fspecial function creates a two-dimensional filter H of the specified type. fspecial returns H as a correlation kernel, which is the appropriate form to use with imfilter. The result of this processed is described in Figure 2.5 below.

Fig. 2.3. The measurement of hand wash prototype especially for distance

between camera and ultraviolet light position.

Fig. 2.4. Image after grayscale process.

Fig 2.5. Images after filtering process.

The captured data is then converted into binary image using a threshold value. The Otsu method is used to

automatically compute an appropriate threshold to use in converting the grayscale image to binary. We then tune the threshold value. The results of this process are described in Figure 2.6 below.

Fig 2.6. After the threshold process.

2) Image Processing Technique In the Image Processing Technique the background

from the foreground object in the image are separated to further detect stains on washed hand. Since the image of the hand has sufficient contrast from the background, we can easily detect the hand using edge detection and basic morphology tools. For this step, we separate to the two detection methods that we used for this project, which are palm area detection method and stain on palm detection method.

a) Palm Area Detection The first step for the palm area detection method sharpens

the image by second threshold value in binary format. Then, after we proceed the sharpened image, we detect the edges of the hands using edge detection method. This process is obtained to detect the edges in the image.

Canny filter is used because it detects strong and weak edges. Accurate edge detection is desirable for the calculation of hand area percentage. The results of this process are described in Figure 2.7 below.

Fig 2.7. Image after Canny process.

Using morphology, the form of the hand was detected as

shown in Figure 2.8 below.

Figure 2.8. Detected hand after morphology.

Then, image was labeled and the area inside of palm was

detected to separate from the background pixels. Decision is made based on the area chosen for the background and palm as the palm’s area is the largest in the labeled image. Filtering is used again to create predefined 2-D objects. The following Figure 2.9 shows of filtered images by hand.

Fig. 2.9. Filtered images of detected hand.

b) Stain Detection on Palm

From the detected image of hand, we can see that the color of background hand and the GLO GERM stain were a different color. To separate this color, color processing is used, based on the stain color on palm. The hue, saturation and value (HSV) color plane is used to extract out the hue, saturation and value components of the hand images individually rather than the red, green, blue (RGB) component of the stain image.

Specific threshold values for all HSV planes are used to detect the stain. These thresholds are used to acquire binary image for each color planes. Logic multiplications for all three planes produced a mask that enhances the stain intensity on the palm. After removing noises of small areas detected between the unclean areas, good result of the stain on palm image is acquired. We used stained object mask to mask out the stain-only portions. The Figure 2.10 shows an image of the detected hand and detected unclean areas.

Fig 2.10. Image of detected palm and detected unclean areas of the hand.

3) Classification Decision Finally, the decision is made based on test and analysis

done on the image before the final output where the detected object will be displayed. We detect the areas of unclean hand which dictated by the colored area that has GLO GERM stains. To detect the colored area, we used the color detection method to separate the background of the hand image to the GLO GERM. Then, we estimated the percentage of unclean areas to analysis the result of the hand whether clean or not. The decision of this entire project is shown by stimulation using Graphical User Interface (GUI). Figure 3.1 below is shown the stimulation with GUI.

III. RESULTS The system has been run and tested under condition of an

image of a hand with the GLO GERM stained. Result acquired from the system may be deferred to the percentage of the GLO GERM stained. The prototype is tested with more than 100 images of hand which are stained and not stained with GLO GERM. It is concluded that if the percentage is below than the 30%, it means that the hands is in clean condition. Well, if the percentage is higher than the 30% it means that the hands is unclean condition and need to rewash. Figure 3.1 below shows the example of the hands condition. The final result is shown using simulation using GUI.

Fig 3.1. Final result using GUI simulation.

A graphical user interface (GUI) was developed to make

computer operation more intuitive and easier to use. It is much easier for a new user to move a file from one directory to another by dragging its icon with the mouse. For that person especially hospital staff who wants to do the hand wash screening audit using this ‘Unclean Hand Detection Machine Using Vision Sensor’, they just click at the graphical interface (GUI) to knows their hands condition is in the GLO GERM stains or not.

First of all, the insert the hand inside the prototype then click the START button. Then after five seconds, the PC camera will capture the hand image and transferred to the computer. The image processing software will process the image to detect the hand. After that, at the GUI window, we can see the images of ‘Captured Image’ and ‘Detected Unclean Area’. The ‘Captured Image’ means that the original image under the ultraviolet light whiles the ‘Detected Unclean Area’ image means the image processes after using image processing technique. If in the ‘Detected Unclean Area’ image shows an orange color in the window, that’s means our hand have a GLO GERM stained. Then at the GUI window, it shows a percentage of the unclean area of hand according to the area of the hand.

In the GUI window also shows if the percentage areas of detected unclean hands are below than 30%, it shows a message “GOOD!! CLEAN!”, meanwhile if the percentage is higher than 30%, the message in the GUI window will show “UNCLEAN!! REWASH!”. After we know the result of our hand, we click on the RESET button to reset all the information.

IV. CONCLUSIONS As the conclusion of this project, the hands images can be

transferred to the computer through PC camera using a vision system. Then, the system will display at the computer screen by using image processing software in the GUI window mode. The system will detect the unclean areas of hand and can estimate the percentage of the unclean areas. This project is only designed to be used in the hospital. It is designed to be an automated system using a vision system. Can be placed anywhere in a designated condition. Thus the system can be upgraded to make more efficient in term of detecting the stained based on the frequently missed areas [4].

ACKNOWLEDGMENT This project is collaboration between the Department of

Microbiology and Parasitology of Universiti Sains Malaysia and the Faculty of Electrical & Electronics Engineering, Universiti Malaysia Pahang. Many thanks to the staffs from the department whom had helped to develop the ideas and verify the system.

REFERENCES [1] C.L. Pessoa-Silva, K. Posfay-Barbe , R. Pfister, S. Touveneau , T.V.

Perneger, D. Pittet, “Attitudes and perceptions toward hand hygiene among healthcare workers caring for critically ill neonates” Infect Control Hosp Epidemiol. Geneva, Switzerland. pp. 26(3):305-11, March 2005.

[2] “Hand Hygiene Saves Lives” GA,USA, April 30, 2012. In. press [3] A Dogra, “Hand Washing Facts” [Online] Available at:

www.buzzle.com/articles/hand-washing-facts.html May 2010 [4] K Steriline. “Frequently Missed Areas”, Ministry of Health Malaysia

(Quality Health Care Section), Infection Control Association Malaysia, CDC. 2002