Body-Movement-Based Human Identification is FYP project If any issue you can contact me on whatsapp +923135023057

1. BODY-MOVEMENT-BASED HUMAN
IDENTIFICATION
PRESENTED BY :
SHAHBAZ IRFAN SP16-BSCS-054
ZEESHAN HAIDER SP16-BSCS-086
SAJID ISLAM SP16-BSCS-052

2. INTRODUCTION:
• Human identification is an important and challenging research
• Human identification is based on biometric identifiers
• Physiological and behavioral characteristics are used.
• Physiological characteristics represent the shape of human
body e.g face, fingerprint, iris.
• Behavioral characteristics e.g gait, voice, motion, and
movements.

3. • These biometric signatures can be measured at distances.
• A fingerprint is measured by touching a sensor
• Face and gait are measured at a distance by using a camera
sensor.
• Human identification needs to be completed at a
distance without having the attention of subjects.
• Face and gait recognitions are commonly used.
• Gait based human identification methods can identify humans
even if the individual’s face not clearly visible.

4. PROBLEM:
• A person can be identify by his face.
• What if person’s face is not visible or covered.
• This application will be able to recognize the person if the
face is covered or not captured by a camera.

5. SCOPE:
• Body movement based human detection detects
the person from image or video even if the face of a
person is visible or not.
• It detects the movements of person and recognize
the person that, who is this?

6. SOLUTION:
• We use thermal camera for taking images from a specific
angle and then preprocess it by removing noise from it
and then extract features from it and then ready to match
patterns and recognizing.

7. TOOLS:
• We will use thermal camera for taking videos or images
• We will use python with its different libraries i.e open cv
• We will use raspberry pi for integration

8. TECHNIQUE:
• As the 1st step, by obtaining the difference image
between background and input images, the rough area of
human body can be extracted.
• Then, the accurate area of human body can be obtained
by noise filtering based on median filter, morphological
operation, and component labeling (considering the size
and the ratio of height to width of the area) as the
preprocessing of the 2nd step.

10. METHODOLOGY:
• The flow chart summarizes our human detection algorithm. The man
walking in the corridor is detected from the thermal image after
background subtraction and preprocessing such as morphological
operation, noise filtering and labeling operation are performed.
• The thermal camera enables our system to detect a human region from
an image that is captured in various environments including severe
shadow, illumination variations, and darkness. .
• Additionally, in the case where a human is walking close to the camera,
the human is not fully visible to the camera.

13. REFERENCES:
• https://en.wikipedia.org/wiki/Motion_(surveil
lance_software)
• https://sci-
hub.tw/https://www.researchgate.net/public
ation/323043558_Body-Movement-
Based_Human_Identification_Using_Convolut
ional_Neural_Network

14. THANK YOU!