|
Progress |
|
|
Week 1 - 4: Date : 24 th july 2002 Presentation - Progress Report. 4 weeks since I have been here. Summary: 1)The first week went mostly in deciding which project to take. Handled the Image Manager and the Scene Generator. Played with Emily a lot, read papers pertaining to the generator and then finally took up the task of builting in the functionality of retrieving the images from the Image Manager. 2)Spent time trying to understand the database. The associations betweenthe tables, manipulating the database and working on SQL.Trying to get the copy of the Image Database and accessing it thru sulcus. 3)Then came perl...and the image manager code. Figured out where the code was..In which directory and how was it all related. Spent a lot of time to figure out what the perl code means. Small issues like interacting with the web interface - how it changes etc... 4)Pretty much understand the code that i am working with now, am pretty much on the verge of adding a collection of images. 5) the last week end and the first two days of this week went in analysing what can be done in Emily and the Image Manager along with the code for adding in the directory of images. 6)had my presentation today..it went well, apart from some small hiccups.Pretty soon would be moving on to working on Emily again. What we need is a java server which could coordinate with all the applications, the perl code was a pain to work with.
Date : 7 Aug 2002
Finished the Initial problem assigned.
1)The Image Manager can add base images and also annotation files now. 2)Attended the meeting with jim in which he suggested that i should write a more general program, essentially that would eliminate the use of temporary directories and files. 3) There are some changes that i can make today : 1) try and figure out the error that i get if the same number of annotation files are not added as the images. 2) Delete all the temp files that are created in the temp. directory. There will be an over flow error else.
Week 5:
date Aug 8 Starting with the content based image retrieval.
tasks to be done today : 1)read..the book, find articles on the net. 2)try and figure out the vector to raster algorithm. 3) if possible work on the image manager problems.
Week 6: Aug 13 2002
1)was able to load and display an image in a seperate panel. questions?? a)needed to extend the applet class..learn more about it...can i do without it. b)did not use the buffered image object, whcih might be helpful if i am manipulating individual pixels...can i store it in a buffered image object? c)now since there's a buffered image object, i would have individual pixels corresponding to each image....there fore no raster conversion is needed. so i need to explore more the buffered image objects.
2)design the panel...with a simple search and display mechanism.
to do steps: 1) learn about applet.. 2) can i make it work with the database? 3) implementing it with bufferedImage.. 4) designing the panel.
2 pm 1) have figured out the applet package.. Will not be using it since it will not be much of a help as far as the FME is concerned. and Emily does not need the applet since its not a web application. 2) yes the program works fine with the database. 3) am on to designing the interface.... doing trivial things like setting the size of the frame...etc.
to figure out: 1) how to set the right resolution for the image when its displayed? (cause as of now the images are blurred) the next issue is the interface ?? as of now working on implementing a simple search interface which just has the functionality of taking in a search term and then outputting the search images. to do tomorrow: 1)figure out the code for retrieving the images i.e. from the database. 2) ideas regarding buffered images. 3) interface. 4) start on vector to raster.. ulitimately ...could integrate this program with emily.
Week 7:
aug 19 2002 things to do: 1)build in the functionality of a text box which displayes the caption if there's one of the image being displayed. 2)searching thru captions. 3)work on the presentation. 4)add cr's images the entire collection to the database. 5)buffered images.....look look look.
Week 8
aug 21 today is the symposium the past couple of days have been used up in preparing for the slides and improving the interface. the image displayer has the basic functionality of displaying images.searching based on the captions and the annotations.
things that can be done-- 1)try and remove the static field anno. 2)add a text area to display the annotation that is being displayed currently. 3)add the previous button 4)error - why after right auricle, knee doesn't work? 5)error -- 'part of ' doesn't work as yet... aproblem with the array index.
aug22 2002 to do today: 1)make the search with annotations and captions stable...rid of the odd errors. 2) add the annotations field to the interface 3)talk to linda i guess i have my hands full for the day. look at line 264. talked to linda yesterday about the buffered image and that a new image is created everytime. the search does not give odd results --- one still "searching for parts of knee" I am throwing out the common words such as 'and", 'for', 'of' etc. yet get garbage values for parts of knee. - query such as 'thoracic viscera' do not produce any results presumably because the location of the file could not be determined.
aug 23
1)ok..so first removed the small errors while searching thru the annotation fields. 2)used stopping to get rid of the common stop words "and", "of", "for". 3)have changed the collection-add-images code to do two things now -- a)it writes to a temp dir instead of writing to my home directory. b) and deletes all the files..i.e the zip files and also the unzipped files right after it uploads them in the FSA.
to do -- 3)there's still an error with 'parts of knee'-have to figure that out.In the long run need to add the functionality of the button + the annotation field.
4)start with buffered images--
possible approach -- 1)take the image object as of now 2)convert to a buffered image. 3)now take the coordinates from the Imageannotation file of the boundary regions and and convert that into raster format. 4)store that raster format into and array. I dont think i will need to manipulate the earlier version of it. --calculate the area of the region, from the number of pixels.and order them. In this case i will not need to change the current setting,but only add to it.
5)some other functionalities like 1)the previous button can be added. 2)resizing the images.
aug 25 made the parser.. error :1) while printing the null check will not work inside the action performed loop. 2) check for null values for coordinates...that might be a problem. think.. aug 26 1)the null check works now, have removed the static field.
aug 27 tuesday.
ok..It has taken me almost the entire day to sort and display the images. order them according to the number of pixels... it works for some queries like the right atrium, left atrium,left auricle, right auricle.
so now I have finished content based retrieval as far as only reordering the images is concerned. things to do--
1)clean up the program 2)generalise it for all the polygons and store it in a file. 3)add the extra features.
aug 29 2002
1)removed the image fetching error from both the previous and the next buttons. Now will work on getting the array right+mail linda. and reverse the order of the images diaplyed.
ok 11.41 pm - have reversed the images displayed and have been working getting the right coordiantes displayed. the error that i keep getting is that the labels exceed the names array. Which should not happen. next to do -- remove the error, get the array displayed and then move on to add the labels field. there's definately an error with the label numbers..... its 3pm am leaving to take care of things will be back in the evening to work on it.
|