I realized that for my current method for calculation of the point operation is not efficient as it performs calculation on every single pixel, hence for a 1 megapixel photo, 1 million calculations is needed to be performed.
So, by reference to Java, I build my own map/table to map values into new values. So, now I have 256 calculations to create the table, and 1 million array access to be done to extract the new values.
However, such approach only works for some cases. Gray Scale operation, for example, require the average of the RGB channels, which is not feasible to be done using color-mapping approach.
I also looked into some several approach regarding the neighbor-operation. Will start to implement a working prototype package during the weekend.
No comments:
Post a Comment