I want to find how much information an image has, for example if it contains human or other objects, then it'll have more information and if an image doesn't contain such objects, it indicates the low-level of information present inside it.
Previously I've experience with entropy and complexity measure already, and similarly I also have pretty experience with saliency based techniques.
If anyone know about some other low-level features based measure, he should suggest me here. Finally, if someone share information about how can we represent saliency of an image through a real-value, it'd be highly appreciated.
Entropy will not give your target information. Rather, you have to implement the object detection algorithm first. First try to detect the predefined set of objects like human, chair, bed, etc. Weight each one with a value. Then follow one of the following:
1) Calculate the total weight in the image regarding object.
2) Calculate sum of weights of pixels. In that case, apply the weight of pixels according to their belonging to the objects.
Next apply that weight in another equation to compute your result according to the demand of your research.
Tricky question.
You need a set of reference images considered informative, and others considered informationless. Then you define a distance between images, and you allocate a threshold for your classification...
- Badges
- Science topic
- Similar topics
- Computer Science
- Images