28-03-2011, 10:50 AM
[attachment=11133]
CHAPTER 1
INTRODUCTION
Recent technological advances leads to the generation of smart cameras that represent a quantum leap in sophistication. While today’s digital cameras capture images, smart cameras capture high level description of the scene and analyze what they see. . These low-cost, low-power systems push the design space in many dimensions, making them a leading-edge application for embedded system research.
These devices could support a wide variety of applications including human and animal detection, surveillance, motion analysis and facial identification. But this paper mainly deals with gesture recognition using smart camera. Our smart camera uses novel algorithms to recognize gestures based on low level analysis of body parts as well as hidden markov models for the moves that comprise the gestures Video processing has an insatiable demand forreal-time performance. Fortunately, Moore’s law provides an increasing pool of available computing power to apply to real-time analysis. Smart cam-eras leverage very large-scale integration (VLSI) to provide such analysis in a low-cost, low-power sys-tem with substantial memory. Moving well beyond pixel processing and compression, these systems run a wide range of algorithms to extract meaning from streaming video. Our system can recognize gesture sat the rate of 25 frames per second. The algorithms run on a Trimedia processor.
CHAPTER 2
SMART CAMERA ARCHITECTURE FOR FACE RECOGNITION
The CMOS sensor captures the image .The representation of the pixels as they are delivered by CMOS sensor image are in the RGB form. It is given to the Xetal processor. Each low level processing approach of the face detection part is mapped to a massively parallel processor Xetal and the high level image processing part of face recognition to a high performance fully programmable DSP core Trimedia. Xetal contains 320 pixel level processors. This processor directly reads the pixels from CMOS image sensor and performs face detection part. Coordinates and sub regions of image are forwarded to the Trimedia. This processor scales the sub regions and matches them to the faces in the database. Only ID’s are reported to the user.
CHAPTER 3
HOW SMART CAMERAS RECOGNISE GESTURES?
Although there are many approaches to real-time video analysis, we chose to focus initially on human gesture recognition means identifying whether a subjects walking, standing, waving his arms, and so on. Because much work remains to be done on this problem, we sought to design an embedded system that can incorporate future algorithms as well as use those we created exclusively for this application. our algorithms use both low-level and high-level processing. The low-level component identifies different body parts and categories their movements in simple terms.
The high level component uses this information to recognize each body part’s action and the erson’s overall activity based on scenario parameters.
Gesture recognition mens identifying where an object is walking, standing, waving his arms and so on. Because much work has to be done on this problem, we have to design an embedded system. It includes gesture detection and recognition algorithm. Our algorithm has two major parts.
1. Low level processing
2. High level processing
The algorithm is shown in Figure3.1.The green blocks indicates high level processing and blue blocks indicates low level processing.
3.1 LOW LEVEL PROCESSING
The system capture images from the video input, which can be either compressed or uncompressed and applies four different algorism detect and identify human body parts. The system captures images from the video input, which can be either uncompressed or compressed (MPEG and motion JPEG), and applies four different algorithms to detect and identify human body parts.
These algorithms are given below.
3.1.1 Region extraction
3.1.2 Contour following
3.1.3 Ellipse fitting
3.1.3 Graph matching
3.1.5 Classification
3.1.1 REGION EXTRACTION
This algorithm performs two operatations, back ground elimination and color segmentation. The algorithm transforms the pixels of an image into an MYN bitmap and eliminates the background. It is shown in Figure3.1.1.Algorithm then detects ody parts skin area using a YUV colour model with chrominance values down sampled.
CHAPTER 1
INTRODUCTION
Recent technological advances leads to the generation of smart cameras that represent a quantum leap in sophistication. While today’s digital cameras capture images, smart cameras capture high level description of the scene and analyze what they see. . These low-cost, low-power systems push the design space in many dimensions, making them a leading-edge application for embedded system research.
These devices could support a wide variety of applications including human and animal detection, surveillance, motion analysis and facial identification. But this paper mainly deals with gesture recognition using smart camera. Our smart camera uses novel algorithms to recognize gestures based on low level analysis of body parts as well as hidden markov models for the moves that comprise the gestures Video processing has an insatiable demand forreal-time performance. Fortunately, Moore’s law provides an increasing pool of available computing power to apply to real-time analysis. Smart cam-eras leverage very large-scale integration (VLSI) to provide such analysis in a low-cost, low-power sys-tem with substantial memory. Moving well beyond pixel processing and compression, these systems run a wide range of algorithms to extract meaning from streaming video. Our system can recognize gesture sat the rate of 25 frames per second. The algorithms run on a Trimedia processor.
CHAPTER 2
SMART CAMERA ARCHITECTURE FOR FACE RECOGNITION
The CMOS sensor captures the image .The representation of the pixels as they are delivered by CMOS sensor image are in the RGB form. It is given to the Xetal processor. Each low level processing approach of the face detection part is mapped to a massively parallel processor Xetal and the high level image processing part of face recognition to a high performance fully programmable DSP core Trimedia. Xetal contains 320 pixel level processors. This processor directly reads the pixels from CMOS image sensor and performs face detection part. Coordinates and sub regions of image are forwarded to the Trimedia. This processor scales the sub regions and matches them to the faces in the database. Only ID’s are reported to the user.
CHAPTER 3
HOW SMART CAMERAS RECOGNISE GESTURES?
Although there are many approaches to real-time video analysis, we chose to focus initially on human gesture recognition means identifying whether a subjects walking, standing, waving his arms, and so on. Because much work remains to be done on this problem, we sought to design an embedded system that can incorporate future algorithms as well as use those we created exclusively for this application. our algorithms use both low-level and high-level processing. The low-level component identifies different body parts and categories their movements in simple terms.
The high level component uses this information to recognize each body part’s action and the erson’s overall activity based on scenario parameters.
Gesture recognition mens identifying where an object is walking, standing, waving his arms and so on. Because much work has to be done on this problem, we have to design an embedded system. It includes gesture detection and recognition algorithm. Our algorithm has two major parts.
1. Low level processing
2. High level processing
The algorithm is shown in Figure3.1.The green blocks indicates high level processing and blue blocks indicates low level processing.
3.1 LOW LEVEL PROCESSING
The system capture images from the video input, which can be either compressed or uncompressed and applies four different algorism detect and identify human body parts. The system captures images from the video input, which can be either uncompressed or compressed (MPEG and motion JPEG), and applies four different algorithms to detect and identify human body parts.
These algorithms are given below.
3.1.1 Region extraction
3.1.2 Contour following
3.1.3 Ellipse fitting
3.1.3 Graph matching
3.1.5 Classification
3.1.1 REGION EXTRACTION
This algorithm performs two operatations, back ground elimination and color segmentation. The algorithm transforms the pixels of an image into an MYN bitmap and eliminates the background. It is shown in Figure3.1.1.Algorithm then detects ody parts skin area using a YUV colour model with chrominance values down sampled.
