The paper investigates the steganographic method for introducing secret messages into a container represented by the image. The research task is improving of the reliability of message storage in the event of various image distortions after transmission over a communication channel. To solve this problem, a preliminary division of the container image in a form of sectors was used. In the developed method, each sector is embedding the same secret message. The geometric shape of each sector can be different, and it depends on the distribution of the selected pixels into which the message is embedded. Pixel selection is carried out in this paper using edge pixel selection operators in the image, such as: Roberts, Sobel and Prewitt operators. For each container image, there is a different distribution of the extracted pixels and a different number in each sector. Therefore, the size of the secret message is limited to the sector in which the smallest number of extracted pixels is present. The bits of the secret message in each sector are distributed differently since the extracted pixels in each sector differ in their location. In case of destruction of a part of the container image, the secret message is saved in the remaining parts, which include non-destroyed sectors. To embed bits, the shape of the initial container, the geometric shape and location of the sectors, as well as the method for extracting pixels is required to be known. The developed method was tested using different images and the achieved performance was excellent with accuracy 92%.