Abstract: This paper presents a robust and real-time capable recognition system for the fast detection and classification of objects in spatial 3D data. Depth and reflection data from a 3D laser scanner are rendered into images and fed into a saliency-based visual attention system that detects regions of potential interest. Only these regions are exanimate by a fast classifier. The time saving of classifying objects in salient regions rather than in complete images is linear with the number of trained object classes. Robustness is achieved by the fusion of the bi-modal scanner data; in contrast to camera images, this data is completely illumination independent. The recognition system is trained for two different object classes and evaluated on real indoor data.

Download Link: Click here

Abstract: Face detection is a two-class problem. Given an image, we would like to confidently say whether a face is present in it. The face detection procedure involves preprocessing of the image followed by detection. Preprocessing involves histogram equalization to improve the quality of the image. The preprocessed image is scanned to detect the edges; finally the edge-detected area is given as input to LDA classifier, which returns the location of the face, if present. The LDA classifier is trained on the whole face image. The input to the classifier would be the x and y co-ordinates of a window and the window is moved until the entire image is traced. The classifier returns an output measure, which is compared with the stored measure of image in the database, and the appropriateness is checked. The same procedure is repeated for different resolution levels and thus the face is detected.

Courtesy: N.Santhanam, Adi Parasakthi Engineering College (APEC), Melmaruvathur, Tamilnadu

Download Full Paper: Click here