Prof. K. T. Talele

The proposed system performs two important tasks i.e.”3D Normalization” and “Recognition”. 3D Normalization is rotation of a human face model along the three axis x-y-z in order to obtain frontal view of face model. This helps us to compare the input image with those in the database. This in turn increases the recognition rate. Recognition is performed to reveal the identity of the person whether he is present in the database or no. Such systems compare the portrait of a tested person with photos of people who have access permissions to object.

Given a video, face recognition is to determine whether or not the specified person is in the video. The research for face detection and recognition is focusing on 1) Video to Frame Conversion: To sample video at 15 frames per second. 2) Face Detection: To detect the faces in the frames 3) Face Recognition: To identify the person in the video The video is taken in .avi or .mpg format thereby calculating the total no. of frames present in a video and then sampling it at a predefined rate. We develop a technique to identify an unknown person in an video image by using SMQT (Successive Mean Quantization Transform) and SNOW classifier and Template matching. SMQT is used to extract features from the local area of an image. The system has 100% detection rate and 97-98% Recognition rate.

Technology has evolved to a great extent and it promises to evolve further in the years to come. For instance, we have faster processors that can execute thousands of instruction within a second. These developments haven't stopped and scientist, developers and compa- nies around the globe are continuously working on ways to improve the technology in every possible way. Apart from speed, throughput and various other aspect of a computer there is one other important factor that is to be taken into account when designing a system. This aspect is the quality of interaction that takes place between user and computer. There are various ways in which quality of interaction can to improved. The project aims at using one such low cost interaction method, a multi touch system, to show how the interaction between human and machine can be improved. This improved interaction is shown by controlling a microcontroller based robot according to the touch input given by the user. The touch screen is not capacitive based screens that are used in the mobile phone these days. But the touch screen is implemented using Infra red LED and an Infra red camera. The robot is made using DC motors and a microcontroller. The feedback includes images captured by the over head camera. The communication is done using wireless module that implements IEEE 802.15.4 Zigbee protocol. The programming is done on a Java based platform and C language. This project can be used in scenarios wherein the operator has to control multiple objects simultaneously and the environment cannot be viewed directly by the operator.

The application provides efficient “Suspicious Object Detection” along with tracking of the suspect from a real time video. For this to happen the algorithm involved includes Template Matching for peculiar object detection and various operations including Kalman Filtering / DWT and background subtraction for Motion detection. However the tracking system includes a Microcontroller System fed by the available (detected) data. The system finds its application at places where real time Video Surveillance is required such as Hotels, Motels and houses etc.

26th July 2005 saw the heaviest rainfall ever in the city of Mumbai, as a result of which the city experienced the worst floods in history. An important factor that led to the loss was that the government agencies were unable to relay warnings at appropriate time due to lack of resources. The proposed system takes into account the rain-rate along with the topography of an area to predict possibility of floods. It is based on the fact that in case of rainfall the low-lying regions are more likely to get flooded, the water then eventually moving to the other areas. This can give a fair idea of the time in which the any region under consideration is likely to flood. This concept has been implemented by dividing the area into watersheds. With this technique, it is possible to monitor rain-rate in real time and reduce the prediction time.