Face image profiles features extraction for recognition systems

The object of research is the algorithm piecewise linear approximation when applying it to selection facial features and compression its images. One problem areas obtain optimal ratio degree accuracy image reproduction, as well obtained features, which can be used search for people in databases. main characteristics face are coordinates size eyes, mouth, nose other objects attention. Dimensions, distances between them, their relationship also form a set characteristics. A identify determine these features. First, approximate graph silhouette from right left and, secondly, fragments silhouettes top bottom. purpose next stage implement multilevel segmentation approximated images cover them with rectangles different intensity. Due shape they called barcodes. These three stages faces represented by two barcode vertical horizontal. This material calculate medium intensity function row or column an tool measure values Additionally, widths barcodes calculated. Experimental results known databases presented. compress Experiments have shown how changes image. method has complexity number pixels image, allows testing large data. Finding synchronized object, such calculating all attention on relative form. developed software control parameters conducting research.