5
Title:
Paper# Hybrid spatio-temporal structuring and browsing of an image collection acquired from a personal camera phone
Problems:
One of the problems that have been highlighted here is that because of numerous devices that can collect images these days, there has been huge accumulation of images as data which need to be re-organized. Managing this data so that users get easy access to their multimedia data is one of the challenges that have been addressed in this paper. Also the properties of the images differ content-wise because of which the manner in which they should be searched will be different. Some users might need the images to be sorted as per the date or location or some other parameter that need to be defined. The navigation by users needs to be kept at a minimum level. The time scales for different images are different and need to be standardized to be compared. Thus linear grouping as per time becomes an issue. Also statistical based reasoning is used for recovery of the image collection. The number of images that can be captured within a particular category needs to be maximized and optimized.
New ideas and strengths:
The idea of managing user generated images in an automated way is a novel idea by itself. Minimum user navigation is a well thought of idea. Structuring of these images so as to help in their management and classification later is one of the key notes of the The paper. authors have also considered the changes possible and the various ways in which the same can be classified. They have not ignored that images can be classified or sorted as per more than one parameter and hence the algorithm developed should take this into consideration. Probabilistic model based clustering has been used which uses statistical clustering. The Expected Maximization algorithm used optimizes the results obtained after the proposed probabilistic model application. The proposed approach has been explained in detail which helps in better understanding of the approach.
Weaknesses and Extensions:
There can be a certain group of images which do not particularly fit into any category as such. These images can be termed as erratic images. There has to be a way to classify these images in the available classes or maybe...