Point Feature Based Detection of Objects in Robust Conditions

Abstract

This paper shows a strategy, for detecting an object in a cluttered scene by Feature point matching as a method for performing authenticity inspection by image recognition. For the Human recognition system regardless of various perspectives and contrasts, it is simpler to distinguish the pictures. It is still a challenge for the computer vision to extract the features from the images to perform object recognition. Utilizing a few highlights which are extracted dependent on the statistical distribution of features on the bitmap picture we can deal with the challenges in recognition like cluttered scenes i.e., even when there are changes in the object size, distance, location, ori­entation, a scale change .It is also robust to small amounts of in and out-of-plane rotation and occlusion and other challenges. Here, we identify a specific article in a clutterd scene, given a reference picture of the article. Using different algorithms: BRISK, SURF, FAST, HARRIS, MSER we visualize the strongest feature points for detecting a specific object based on correspon­dences between reference and target image.