Image Processing Method for Automatic Discrimination of Hoverfly Species

Figure 3

Visualization of RCLBP_S and RCLBP_M codes. In the middle, (b), is an artificial mosaic formed by images which represent different types of vein junctions in wings of hoverflies. Positive class of training/test set used for supervised learning of vein junctions detector, Section 2.2, consists of images like those shown in (b). Mosaic images on the left (a) and on the right (c) sides of (b) represent generated grayscale visualizations of the computed RCLBP_S and RCLBP_M codes, respectively. Values of computed 8-bit binary codes are in the grayscale range, which makes them suitable for direct visualization. In (d) and (e) are magnified details from (a) and (b). These details represent values of the corresponding codes, computed for each pixel of vein junction outlined with blue frame in (b). Additionally, in order to emphasize difference between values of binary codes corresponding to junctions and those corresponding to the surrounding background, in (d)-(e) with green and blue color are also written their integer values, zoom figure.

(a) RCLBP_S codes

(b) Types of vein junctions

(c) RCLBP_M codes

(d) Values of RCLBP_S codes

(e) Values of RCLBP_M codes