The solution of the task of isolating the texture attributes of a digital image is relevant in many areas: computer diagnostics, analysis of satellite imagery, video surveillance systems, navigation systems, etc. At the moment, the concept of digital image texture (CI) is interpreted differently depending on the specific tasks of image processing, and their quality is tested empirically for each classification problem, so the synthesis of a large number of textural features and their research on information is relevant. Important concepts of the texture are the background and the outline. Previously, a method for classifying image matrix blocks based on an analysis of their brightness function was proposed, according to which four categories of blocks were identified: background and three categories of contour ones (with a strongly pronounced, medium-pronounced and weakly expressed contour). The above classification is made for blocks measuring 8 x 8 and allows you to accurately isolate the contours of objects in the image. A block is assigned to a category by comparing the difference between the maximum and minimum values of the pixel brightness with a threshold value, the recommendations for which were obtained empirically. The need for such a classification arose in connection with the solution of the issues of identifying photomontage. The effectiveness of some methods depends on the frequency component of the image signal: at a low level of the high-frequency component (part of the image without contours or with poorly expressed contours), the method copes with the task, otherwise its use is not advisable. This fact indicates the possibility of applying the method of classifying image blocks in the area of information security. Also in the work recommendations are given regarding the specification of the boundaries of the selected objects, using blocks of smaller sizes. However, when changing the block size, it is also necessary to take into account the need to correct the threshold value. The purpose of this work is to improve the accuracy of the selection of the boundaries of objects in the image by a method based on the analysis of the brightness function of digital image blocks. In the paper it is shown that when the size of blocks decreases, the allocation of small details, as well as the isolation of the outlines of objects as a whole, occurs more efficiently than in the original version. This is observed both for a large object in the frame - the planet, and for its satellite. The threshold value was selected by subjective ranking. When analyzing a hundred images, it was determined that for the 2 x 2 blocks, the best results were obtained using a threshold value of 15. Thus, in the work it was possible to improve the accuracy of the contour extraction in the manner suggested earlier. Recommendations are received regarding the choice of block size and threshold value. By subjective ranking it is established that at a threshold value of 15 for blocks of 2 x 2 size, the accuracy of contour allocation increases significantly. The tasks set in the work are solved, the goal is achieved.