HSV Threshold


The class HSVThreshold is a tool for clustering problems in HSV (Hue, Saturation, and Value) color space for color-based pixel separation. The main purpose of this tool is to support semi-automated color-based segmentation. Particularly, this tool provides users the decision support about valid pixel value ranges for specific types of object detection.

Description: The class operates in three ranges of HSV values. Points within the user-specified value ranges are appeared in the result image, both as RGB and HSV images. In each scrollbar, a user can adjust the upper and lower limits of the pixel values, and can fix the dynamic value range by checking the “Fix Range” option. For example, the lower/upper limit will move automatically while keeping the same dynamic value range when changing the upper/lower limit.


Setup: By default, both RGB and HSV image are displayed from the main Im2Learn frame. The two windows dynamically show the classified images when a user adjusts the threshold value ranges.


For large image processing, such as “svs” format, a user can directly load the image into the HSV threshold tool independently from the main Im2Learn frame by pressing “Load” button. If a sub-sampled overview image is available, as a same filename with “tif” extension, an overview window will be displayed where a user can interactively select the currently displayed region (marked as red cross).


Run: First, select a sub-region from the overview image to load a 200 by 200 pixel neighborhood from the large image to the RGB and HSV windows. Next, change the Hue, Saturation, and Value range to acquire the thresholded image. The number of valid (displayed) pixels is shown in the title of the tool as “Positive pixel count” with the percentage with respect to the total number of the pixels in the sub-region.

“Change Region” button can be used to load an overview image with different filename. Note that the overview image has to be a 1:200 sub-sampled image from the original svs image, for example, an image sub-sampled by the factor of 200.