Object Detection in The Image Using the Method of Selecting Significant Structures
The approach to image segmentation is reviewed in the article. The method of highlighting significant contours in the image is reviewed. Some structures in the image attract attention more than others due to certain distinctive properties. The article reviews the approach of highlighting significant structures in the image representing the areas of candidates identifying the object in the video frame for mobile platforms. For example, such shapes can be smoother, longer and closed. Such structures are called significant. It would be expedient to use only these significant structures to increase the speed of image recognition by computer vision methods focused on the contour selection. This approach allocates the computing resources only to significant structures, thus reducing the total computation time. Since the image consists of many pixels and links between them, which are called edges, significant structures can be measured. The article presents an approach to measuring the structure significance that largely coincides with human perception. Some image structures attract our attention without the need for a systematic scan of the entire image. In most cases, this significance represents the structure properties as a whole, i.e. parts of the structure cannot be isolated. This article presents a measure of significance based on the measurement of length and curvature. The measure highlights structures characteristic of human perception, and they often correspond to objects of interest in the image. A method is presented for calculating significance using an iterative scheme combined into a single local network for processing elements. The optimization approach to represent a processed image highlighting significant locations is used in the network.