Searching of local features in images based on a texture performance

The use of video is now growing rapidly. This is due to the implementation of various monitoring systems, computer vision, robotics, decision-makers based on the analysis of video, video telephony, and other areas. This, along with the level of technology play an important role of video processing techniques, methods of search and comparison of stable local features of images.

As local features are: points, corners, spots (or areas) and borders. Local feature image is used to combine different views of an object or scene, aerial and panoramic stitching, recovery of three-dimensional scenes, segmentation and recognition of gestures.

We propose a modification of the method SURF to handle different types of local features in the image. The modification consists in the preliminary image processing power characteristics Law-s that allows selecting a variety of local features for image texture features. By analyzing the properties of the texture defined areas that are important to correctly perform the comparison and description of the spatial distribution of color or intensity values in the image. The results of the test image processing shows that by pre-treatment imaging performance by Law-s characteristics, can detect a variety of local features.