Jayati Ghosh Dastidar, Surabhi Sarkar, Rick Punyadyuti Sinha et al.
This paper describes the method to recognize offline handwritten characters. A robust algorithm for handwriting segmentation is described here with the help of which individual characters can be segmented from a selected word from a paragraph of handwritten text image which is given as input.
Jonathan Fidelis Paul, Dibyabiva Seth, Cijo Paul et al.
In this paper we have presented a hand gesture recognition library. Various functions include detecting cluster count, cluster orientation, finger pointing direction, etc. To use these functions first the input image needs to be processed into a logical array for which a function has been developed. The library has been developed keeping flexibility in mind and thus provides application developers a wide range of options to develop custom gestures.
Subhajit Adhikari, Joydeep Kar, Jayati Ghosh Dastidar
This paper presents a method to differentiate the foreground objects from the background of a color image. Firstly a color image of any size is input for processing. The algorithm converts it to a grayscale image. Next we apply canny edge detector to find the boundary of the foreground object. We concentrate to find the maximum distance between each boundary pixel column wise and row wise and we fill the region that is bound by the edges. Thus we are able to extract the grayscale values of pixels that are in the bounded region and convert the grayscale image back to original color image containing only the foreground object.
Merina Kundu, Dhriti Sengupta, Jayati Ghosh Dastidar
Sometimes a simple and fast algorithm is required to detect human presence and movement with a low error rate in a controlled environment for security purposes. Here a light weight algorithm has been presented that generates alert on detection of human presence and its movement towards a certain direction. The algorithm uses fixed angle CCTV camera images taken over time and relies upon skeleton transformation of successive images and calculation of difference in their coordinates.
Dhriti Sengupta, Merina Kundu, Jayati Ghosh Dastidar
It is at times important to detect human presence automatically in secure environments. This needs a shape recognition algorithm that is robust, fast and has low error rates. The algorithm needs to process camera images quickly to detect any human in the range of vision, and generate alerts, especially if the object under scrutiny is moving in certain directions. We present here a simple, efficient and fast algorithm using skeletons of the images, and simple features like posture and length of the object.