image-core
Work with SimpleITK Image objects, spatial metadata, pixel/component types, physical coordinates, and NumPy array conversion.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- image-core
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Work with SimpleITK Image objects, spatial metadata, pixel/component types, physical coordinates, and NumPy array conversion.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/api-reference.md5.7 KB
- 参考文档references/image-model.md6.2 KB
- 参考文档references/numpy-bridge.md5.8 KB
- 参考文档references/troubleshooting.md5.4 KB
- 脚本scripts/inspect_image_metadata.py4.5 KB
03 SKILL.md 正文
SimpleITK Image Core
Use this sub-skill when a task involves constructing or inspecting sitk.Image objects, preserving origin/spacing/direction, converting between index and physical coordinates, reading or writing individual pixels, diagnosing pixel/component ids, or bridging images to NumPy arrays.
Start Here
- Read references/image-model.md for the SimpleITK physical image model, spatial metadata, coordinate conversions, pixel/component categories, metadata dictionary basics, and validation checks.
- Read references/numpy-bridge.md for
sitk.GetArrayFromImage,sitk.GetArrayViewFromImage,sitk.GetImageFromArray, reversed shape order, vector image component axes, dtype mapping, and metadata preservation after array processing. - Read references/api-reference.md for concise verified constructor patterns, image methods, pixel inspection APIs, NumPy bridge behavior, and copy/view notes.
- Read references/troubleshooting.md when debugging metadata loss, reversed NumPy shapes, read-only array views, unsupported dtypes, wrong
isVectorchoices, component mismatches, direction errors, or out-of-bounds indices. - Run scripts/inspect_image_metadata.py to print JSON metadata for a readable image file or a deterministic built-in demo image from any current working directory.
Quick Routing
- For
ReadImage,WriteImage, registered ImageIOs, DICOM tags, series discovery, file formats, or dataset examples, route to ../io-and-data/SKILL.md. - For smoothing, thresholding, morphology, connected components, segmentation, measurements, or label workflows beyond basic pixel/type inspection, route to ../filtering-segmentation/SKILL.md.
- For registration, resampling, transforms as processing objects, interpolators, displacement fields, or coordinate-frame alignment workflows, route to ../registration-transforms/SKILL.md.
- For source builds, wrapping internals, optional elastix/transformix availability, packaging, or compiled feature differences, route to ../builds-and-wrapping/SKILL.md.
Essential Rules
- Import the public package as
import SimpleITK as sitk; the Python distribution name issimpleitk. - Treat an
Imageas a pixel buffer plus physical geometry. PreserveGetOrigin(),GetSpacing(), andGetDirection()whenever replacing pixels or rebuilding from NumPy. - Use SimpleITK index order
(x, y, z, ...); NumPy scalar arrays expose spatial axes in reverse order such as(z, y, x)for 3D. sitk.GetArrayFromImage(image)returns a mutable deep copy;sitk.GetArrayViewFromImage(image)returns a read-only view tied to the image buffer.sitk.GetImageFromArray(array, isVector=None)creates a new image with default geometry, reverses spatial axes, and treats the final axis as components only when vector conversion applies.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).