This article discusses a modified version of Viola-Jones' face detection algorithm. The research determined the influence of the orientation of a person's face in space on the operation of the Viola-Jones face recognition algorithm. The authors set the goal of this work  to develop a method for finalizing the face recognition algorithm, it allows using a wide range of face rotation angles in the input image. All images in this work are in 3 dimension. In the process of work, the instability of the algorithm to a change in the position of the face is revealed and a method is proposed to increase the stability of the algorithm to this change. A technique for comparative analysis of the performance of face recognition algorithms is described. The article offers the results of a comparative analysis, as well as the test result of the developed algorithm is shown in the form of a table. By the result, modified algorithm only slightly increased the percentage of image recognition at the boundaries of the interval of acceptable rotation angles for the Viola-Jones algorithm, the introduction of image segmentation based on skin color reduced the number of false alarms of the algorithm and reduced the execution time of the algorithm due to the fact that only part of the images.