Development of a software module to recognize linear structures in aerial images using ENVI

Nowadays digital image processing systems are characterized by a constant increase in the volume of processed information, increasing requirements for the processing quality in difficult conditions. Automation of image elements detection and analysis is one of the most promising areas of research in geoinformatics. In particular, there is a need to find linear and ring elements in aerospace area images. Traditionally, the detection process is done manually. This process is difficult to formalize and depends on a large number of natural and man-made factors. Together with the large volume of information received and the labor intensity, "manual" interpretation of images does not allow to process a large number of images in an acceptable time frame. Moreover, a person working with images has his own subjective perception features and evaluation criteria. These problems have been solved using methods based on the traditional geological and mathematical approach. Lines in an image are not the anomalies of the earth's crust themselves, but only their features; it is difficult to define the meaning of an individual line by computer means. Therefore, for clarity, in the context of this work, linear element (LE) will be considered as a pixel formation in a digital image, well approximated by a straight-line segment. Such structures, identified on aerospace images of the area, can serve as an external manifestation of various anomalies on the earth's surface, for example, ruptures of the earth's crust or various anomalies of physical fields. Thus, separate LEs identified on images can be objects of different origin, for assessing geological significance. The article presents the result of the analysis of the main methods for searching for linear elements on digital images, and also substantiates the choice of the Canny algorithm for the "LINEAMENTS2" module.