Human Recognition in Video

Human Recognition in Video

Danial Behzadi

Advisor:

Dr. Hadi Moradi

Human recognition in video 2

Research Topics

● Image processing● Computer vision● Biometrics● Pattern recognition

Human recognition in video 3

What is Biometrics?

● Study of methods for uniquely recognizing humans

● Based upon intrinsic physical or behavioral traits

● fingerprint, face, voice, gait, iris

Human recognition in video 4

Problem

● Performance is constrained by the intrinsic factors of a trait

● No single biometrics meets all the requirements

Human recognition in video 5

Solution

● Expand the feature space by fusion

● US-visit: face and finger print

● We: gait and face for now

Human recognition in video 6

Gait-based Human Recognition

● Pros:– Can be identified at distant– Can track individuals over time

● Cons:– Affected by clothing– Affected by physical conditions

Human recognition in video 7

Face-based Human Recognition

● Pros:– Non-intrusive– Simple computations

● Cons:– Affected by resolution, illumination, etc.– Low level of confidence in large scale

Human recognition in video 8

Gait Energy Image (GEI)

● A spatio-temporal compact representation of gait in video

● Used to characterize human walking properties

Human recognition in video 9

GEI Representation

G(x , y)=1N∑1

N

Bt (x , y)

Human recognition in video 10

High-resolution Image Creation

● Construsct a single high-resolution image from multiple low-resolution images

● An iterative method by M. Irani and S. Peleg, 1993

Human recognition in video 11

Reconstructed Face Image

Human recognition in video 12

Feature Extraction and Transform

● Using PCA to reduce dimension of feature space

● Using MDA to identify the most discriminating features

Human recognition in video 13

Technical Approach

Human recognition in video 14

Performance

Fusion Method

LR Face Only

HR Face Only

Gait OnlyLR Face and Gait

HR Face and Gait

Recognit-ion Rate71.7%84.8%87.0%87.0%87.0%

Human recognition in video 15

Future Works

● Adding a new feature● Use 3D models● Real-time computation● GPU acceleraion

Human recognition in video 16

References

● Bir Bhanu, Ju Han

Human Recognition at a Distance in video, 2010

● Xiaoli Zhou and Bir Bhanu

Feature Fusion of Face and Gait for Human Recognition, 2006

● M. Irani and S. Peleg.

Motion analysis for image enhancement: Resolution, occlusion and transparency, 1993

● C. Liu and H. Wechsler.

A shape- and texture-based enhanced fisher classifier for face recognition, 2001