A Survey of Using Biometrics for Smart Visual Surveillance: Gait Recognition

Smartphone and Smartwatch-Based Biometrics Using Activities of Daily Living

“…Biometric methods can be divided into physiological biometrics and behavioral biometrics [14], [15]. Physiological biometric methods rely on physiological traits such as fingerprints, iris, ear shape, DNA, vein pattern, or face [16]. The most popular physiological traits are fingerprints, which account for more than half of all commercial biometric systems [15].…”

“…Gait biometrics [19] has been implemented using three different sensing modalities: vision-based, floor-based, and wearable sensor-based. Vision-based biometrics [16] requires video equipment but does not interfere with the subjects and is unobtrusive. However, the equipment requirement means this approach is only suitable for fixed locations (e.g., an airport).…”

Human identification based on natural gait micro‐Doppler signatures using deep transfer learning

“…Compared with other biometrics such as face, iris and fingerprint, gait identification has the advantages of non‐invasion, non‐cooperation (without any interaction or cooperation with the subject), long distance and hard to disguise, which makes it so attractive as a means of identification and shows enormous potential in applications of security and surveillance. A number of sensing modalities including those based on vision [2, 3], foot pressure [4, 5], and wearable devices [6–8] have been used to capture gait information. Nevertheless, insurmountable problems, such as lighting, occlusion and privacy concerns faced by vision‐based systems or carrying and deploying difficulties faced by wearable and pressure devices, greatly limit the application scope of these solutions.…”

Gait biometrics: investigating the use of the lower inner regions for people identification from landmark frames

“…In fact, people recognition and tracking by gait are deemed attractive for re‐identification in realistic surveillance scenarios partly due to the fact that the walking style can be extracted and quantified at a distance besides the less‐intrusive and non‐invasive merits. Many research studies tend to focus on devising an innovative vision‐based solution which is capable to address the limitations encountered during the extraction and tracking of biometric gait features [9–12]. There has been a number of research patents documenting different procedures and methods on the use of gait motion for people recognition.…”