Is it possible to detect damage by comparing two images
4 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
Baisseyev Miram
el 10 de Abr. de 2018
Comentada: Baisseyev Miram
el 25 de Abr. de 2018
<< Hi. Is it possible to find surface damage of an object by comparing images of damaged and reference images. i mean i have two images of an object. On one image object undamaged and on other is damaged, is it possible to find damaged area (surround damaged area in some boundary box or line) by comparing images of damaged and undamaged objects? There is some example images. Two objects have same sizes.
5 comentarios
Walter Roberson
el 11 de Abr. de 2018
You also need to correct for the lighting differences. Those two pictures have different lighting intensity and different angles of lighting.
Respuesta aceptada
Florian Morsch
el 24 de Abr. de 2018
Editada: Florian Morsch
el 24 de Abr. de 2018
One other thing you could do in the case of those two images: make binary images out of them, with a threshhold set so you will have a white area within the undamaged object and when its damaged its black in the binary image because its also darker in the RGB. First take a picture of the background, then a picture of the undamaged and then of the damaged object, best would be if they are nearly in the same position. Then substract the background, so you only see your object. After that you can find the threshhold and make the binary images, in which the damaged one will be black, the undamaged should be mostly white.
I would recommend to use images from the same side (undamaged and damaged object) and also under the same conditions (lightning, backround, width, height).
This method would only allow to see if the object is damaged or not, you cant get really a indicate on how much damaged it is.
Más respuestas (0)
Ver también
Categorías
Más información sobre Image Segmentation and Analysis en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!