Black Highlighter — May 1st, 2019
We’re getting closer and closer to a real app every day now. During this session, I built up the last of the pieces needed to start actually redacting text in images: detecting individual characters, and identifying which characters have been selected for redaction.
Detecting individual characters is an important part of providing an actually-useful experience. Only being able to redact whole text objects at a time means that you can’t hide single words in a line of text, for example. Fortunately, the Vision framework already did the heavy lifting for us; we just needed to store and visualize that content.
Identifying which characters have been selected was a little trickier. We can’t just check which text boxes are intersected by the path we draw. For one, a path is just a series of points; it’s entirely possible for a line to contain points on either side of a text box without containing any point in the box itself. For another, the path object vailable to us is effectively zero width. We draw a much wider path to make it easier to select what you want. To accomplish the selection, we create a new path that encloses the shape of our drawn path, and then we find all the character boxes whose centers lie within that path. This isn’t a perfect solution; it’s possible to miss a character in a line because you didn’t quite hit the center, but it’s performant and quick to implement. I’ll probably experiment with other solutions in future versions.
Commits Made
Tickets Closed
Tickets Created
Project Stats
- Sessions Completed
- 13
- Days Since Start
- 31
- Issues Closed
- 15
- Issues Open
- 20
- Percent Complete
- 42.9%
Replay
Watch this session on YouTube: