|
Quick Lists
|
|
Bug ID:
|
4505744
|
|
Votes
|
3
|
|
Synopsis
|
SubsampleBinaryToGray gives incorrect results if xScale*yScale < 1.0/256.0
|
|
Category
|
jai:implementation
|
|
Reported Against
|
1.1.1-dev
|
|
Release Fixed
|
|
|
State
|
11-Closed,
Not a Defect,
bug
|
|
Priority:
|
4-Low
|
|
Related Bugs
|
4505742
|
|
Submit Date
|
21-SEP-2001
|
|
Description
|
SubsampleBinaryToGray gives incorrect results if xScale*yScale < 1.0/256.0.
|
|
Work Around
|
N/A
|
|
Evaluation
|
The results of the operation are in fact correct. According to the specification "Roughly speaking, each pixel (x, y) of the destination, is the sum of the source pixel values of the source pixel matrix of size 1/xScale by 1/yScale. [...] The resulting gray values are then normalized to have gray values in [0..255]." Therefore if xScale*yScale < 1.0/256.0 the sum of values in the source pixel matrix might exceed what may be represented by an 8-bit unsigned quantity, but this sum is rescaled to [0,255]. The implementation thus corresponds to the specification.
Posted Date : 2006-06-20 21:11:29.0
|
|
Comments
|
Submitted On 13-DEC-2005
This one hurts and it has long been reported (4 years should be enough to fix a reproducable bug)
I tried to give you additional information in Incident Review ID 599025 and as a professional programmer I am very sure this one is a simple 8-Bit overflow in counting the bits of the scaled area. You can tell by the result which is modulo 256 of what you expect.
PLEASE NOTE: JDK6 is formerly known as Project Mustang
|
|
|
|
