Real Time Rotation Independent Object Detection

Abstract

Image processing is one of the fastest growing branch in today’s world in terms of the advances in its applications as well the in its technology. In image processing edge detection, object detection, pattern matching, face recognition, rotation detection etc. Are some of the well-known problem? Standard object detection and angle detection techniques are available for solid regular objects in image processing literature. But these techniques faces some limitations when applying object detection or angle detection for irregular objects. Example of some of the standard techniques are SURF (Speeded-Up Robust Features) and SIFT (Scale Invariant Feature Transform) etc. But accuracy in these techniques is one of the major concern and need significant improvement. Another major issue in these techniques is, they are not fast enough for real time object detection as far as the industrial applications are concerned. The project is focused on detection of object in any regular shape (circle, rectangle, square, etc.) or any other irregular shape. Here, the algorithm development will be carried out in visual c++ with open CV (open source computer vision) library, which is well known library for image processing. After developing algorithm, it will be implemented and tested for timing performance, and accuracy. Further, to improve the timing performance of the algorithm, medications will be made to imbibe the concept of multi-core processor and to take advantage of the same. TBB (Threading Building Blocks) library available from Intel will be used for the same. It is a parallel programming in image processing which reduce overall computational time in any process.