Milestone 3 of 8

Select one channel without losing axis meaning

Select a valid channel while keeping a length-one channel axis and rejecting boolean, negative, and out-of-range indexes.

Selecting a channel is not the same as dropping an axis. Keep the result's shape explicit so later operations do not have to guess what a dimension means.

Goal

Select one valid channel from a grayscale or RGB array and return a (height, width, 1) uint8 array without changing the source.

Operation

Implement:

select_channel(image, channel_index)

Require one non-boolean integer in range. For RGB, index 0 is the first declared channel and index 2 is the third. For grayscale, the only valid index is 0. Do not infer channel names from colors or reorder the source channels.

Deliverables

Extend src/operations.py with channel selection and save exact pixel tables for one grayscale fixture and each channel of one RGB fixture. Record source and result shapes and the selected index.

Checks

Check every valid RGB index, the grayscale index, a one-pixel image, and a pattern where channels contain visibly different values. Reject negative, out-of-range, non-integer, and boolean indexes. Check that the result has a length-one channel axis, retains height and width, uses uint8, and is independent of the source.

Change a result pixel and verify the source channel is unchanged. Confirm that the operation does not alter the source's channel order or metadata.

Workspace

Keep channel selection in src/operations.py. Do not add brightness, masks, or transformations that interpret a channel semantically.

Hints

HintShape carries meaning

(height, width) could be a grayscale array or a two-dimensional slice of an RGB array. (height, width, 1) states that one channel remains.

HintCheck the channel before copying

Validate the index against the third dimension before taking the slice. A clear error is better than a silently clipped index.

Review

Pick one RGB pixel and follow its selected channel into the result. What does the result shape tell the next operation? Why should a boolean index be rejected even though NumPy can interpret it?

How to check your work

Checks compare selected pixels, shapes, dtype, index validation, and memory independence with the fixtures. Selection keeps values; it does not classify the image.