refactor: Go模板匹配优化函数分离到minicv包 by isHarryh · Pull Request #1010 · MaaEnd/MaaEnd

isHarryh · 2026-03-05T08:36:19Z

概要

此 PR 分离了 Go Agent 的归一化互相关系数计算函数和两个模板匹配函数到 minicv 包。

Summary by Sourcery

将 Go 模板匹配和归一化互相关（NCC）逻辑从 map-tracker 中抽取出来，迁移到共享的 minicv 包中，并更新 map-tracker 以使用新的 API。

增强内容：

在 minicv 包中引入可复用的 NCC 计算和模板匹配辅助方法，包括区域限制匹配和基于积分图的统计工具。
调整 map-tracker 的推理代码，使其调用新的 minicv 模板匹配函数，而不是本地实现。
通过对极小方差进行钳制（clamp），提升 minicv 统计工具中方差计算的数值稳定性。

Original summary in English

Summary by Sourcery

Extract Go template matching and normalized cross-correlation logic from the map-tracker into the shared minicv package and update map-tracker to use the new APIs.

Enhancements:

Introduce reusable NCC computation and template matching helpers in the minicv package, including area-restricted matching and integral-based stats utilities.
Adjust map-tracker inference code to call the new minicv template matching functions instead of local implementations.
Improve numerical stability of variance calculations in minicv stats utilities by clamping very small variances.

sourcery-ai

Hey - 我在这里提供了一些整体性的反馈：

在 match_template.go 中，循环 for y := range th、for range tw、for i := range numWorkers 和 for range numWorkers 试图对整数进行 range 操作，代码将无法编译；这些循环应该改写为常规的索引循环（例如 for y := 0; y < th; y++、for i := 0; i < numWorkers; i++ 以及相应的计数循环）。
在 GetImageStats（以及 GetAreaStats）中，用于将标准差归零的方差阈值从 1e-3 改成了 1e-6；请考虑这个更严格的阈值是否会在无意中增加被视为有效的低方差区域数量，并与之前的实现相比改变匹配行为。

给 AI 代理的提示

Please address the comments from this code review:

## Overall Comments
- In `match_template.go` the loops `for y := range th`, `for range tw`, `for i := range numWorkers`, and `for range numWorkers` attempt to range over integers and will not compile; these should be rewritten as conventional index loops (e.g., `for y := 0; y < th; y++`, `for i := 0; i < numWorkers; i++`, and corresponding counted loops).
- The variance threshold used to zero out the standard deviation was changed from `1e-3` to `1e-6` in `GetImageStats` (and `GetAreaStats`); consider whether this tighter threshold might unintentionally increase the number of low-variance regions treated as valid and alter matching behavior compared to the previous implementation.

Sourcery 对开源项目是免费的——如果你觉得我们的代码评审有帮助，欢迎分享 ✨

_{帮我变得更有用！请在每条评论上点 👍 或 👎，我会根据你的反馈改进后续的代码评审。}

Original comment in English

Hey - I've left some high level feedback:

In match_template.go the loops for y := range th, for range tw, for i := range numWorkers, and for range numWorkers attempt to range over integers and will not compile; these should be rewritten as conventional index loops (e.g., for y := 0; y < th; y++, for i := 0; i < numWorkers; i++, and corresponding counted loops).
The variance threshold used to zero out the standard deviation was changed from 1e-3 to 1e-6 in GetImageStats (and GetAreaStats); consider whether this tighter threshold might unintentionally increase the number of low-variance regions treated as valid and alter matching behavior compared to the previous implementation.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- In `match_template.go` the loops `for y := range th`, `for range tw`, `for i := range numWorkers`, and `for range numWorkers` attempt to range over integers and will not compile; these should be rewritten as conventional index loops (e.g., `for y := 0; y < th; y++`, `for i := 0; i < numWorkers; i++`, and corresponding counted loops).
- The variance threshold used to zero out the standard deviation was changed from `1e-3` to `1e-6` in `GetImageStats` (and `GetAreaStats`); consider whether this tighter threshold might unintentionally increase the number of low-variance regions treated as valid and alter matching behavior compared to the previous implementation.

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

Copilot

Pull request overview

This PR refactors the Go Agent’s template-matching implementation by moving normalized cross-correlation (NCC) and template-matching helpers out of map-tracker into the shared pkg/minicv package, and updating map tracking inference to use the new minicv APIs.

Changes:

Add minicv.ComputeNCC, minicv.MatchTemplate, and minicv.MatchTemplateInArea as reusable template-matching utilities.
Extend minicv.IntegralArray with GetAreaStats to compute mean/std for sub-rectangles efficiently.
Remove duplicated template-matching logic from map-tracker and switch infer.go call sites to the new minicv functions.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
`agent/go-service/pkg/minicv/stats_utils.go`	Adjust variance clamping logic and add `IntegralArray.GetAreaStats` for area mean/std.
`agent/go-service/pkg/minicv/match_template.go`	Introduce shared NCC + template matching (full-image and bounded-area) utilities.
`agent/go-service/map-tracker/utils.go`	Remove map-tracker-local template matching helpers now provided by minicv.
`agent/go-service/map-tracker/infer.go`	Update inference flow to call `minicv.MatchTemplate*` functions.

Comments suppressed due to low confidence (1)

agent/go-service/pkg/minicv/match_template.go:2

New file is missing the repository’s standard copyright header comment seen at the top of other agent/go-service Go files. Add the same header for consistency.

package minicv

agent/go-service/pkg/minicv/stats_utils.go

isHarryh · 2026-03-05T08:52:56Z

此 PR 未影响现有逻辑表现，属于可兼容的、非破坏性变更，将立即合并。

## 概要此 PR 分离了 Go Agent 的归一化互相关系数计算函数和两个模板匹配函数到 minicv 包。 ## Summary by Sourcery 将 Go 模板匹配和归一化互相关（NCC）逻辑从 map-tracker 中抽取出来，迁移到共享的 minicv 包中，并更新 map-tracker 以使用新的 API。增强内容： - 在 minicv 包中引入可复用的 NCC 计算和模板匹配辅助方法，包括区域限制匹配和基于积分图的统计工具。 - 调整 map-tracker 的推理代码，使其调用新的 minicv 模板匹配函数，而不是本地实现。 - 通过对极小方差进行钳制（clamp），提升 minicv 统计工具中方差计算的数值稳定性。 <details> <summary>Original summary in English</summary> ## Summary by Sourcery Extract Go template matching and normalized cross-correlation logic from the map-tracker into the shared minicv package and update map-tracker to use the new APIs. Enhancements: - Introduce reusable NCC computation and template matching helpers in the minicv package, including area-restricted matching and integral-based stats utilities. - Adjust map-tracker inference code to call the new minicv template matching functions instead of local implementations. - Improve numerical stability of variance calculations in minicv stats utilities by clamping very small variances. </details>

isHarryh added 2 commits March 5, 2026 16:03

refactor: 分离go的NCC计算函数到minicv包

64c2f11

refactor: 分离go的模板匹配函数到minicv包

e166c68

Copilot AI review requested due to automatic review settings March 5, 2026 08:36

Copilot started reviewing on behalf of isHarryh March 5, 2026 08:36 View session

sourcery-ai bot reviewed Mar 5, 2026

View reviewed changes

Copilot AI reviewed Mar 5, 2026

View reviewed changes

agent/go-service/pkg/minicv/stats_utils.go Show resolved Hide resolved

chore: 修一下方差截断

ea41020

isHarryh merged commit 2733707 into MaaEnd:main Mar 5, 2026
16 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

refactor: Go模板匹配优化函数分离到minicv包#1010

refactor: Go模板匹配优化函数分离到minicv包#1010
isHarryh merged 3 commits intoMaaEnd:mainfrom
isHarryh:main

isHarryh commented Mar 5, 2026 •

edited by sourcery-ai bot

Loading

Uh oh!

sourcery-ai bot left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

isHarryh commented Mar 5, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

isHarryh commented Mar 5, 2026 • edited by sourcery-ai bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

概要

Summary by Sourcery

Summary by Sourcery

Uh oh!

sourcery-ai bot left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

isHarryh commented Mar 5, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

isHarryh commented Mar 5, 2026 •

edited by sourcery-ai bot

Loading