René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Contrast adjustment with 'eliminate outliers' failed for float images with high dynamic range","articleBody":"## 🐞 Problem Summary\n\nThe `hist_range_threshold` function was designed to compute the value range covering a central percentage of the histogram mass (e.g. 98%), in order to eliminate symmetric outliers — similar to MATLAB’s *Eliminate outliers*.\n\nThis worked correctly for integer-valued images (e.g. 8-bit or 16-bit), where the first histogram bin typically corresponds to zero-valued pixels and can safely be ignored.\n\nHowever, when applied to **float-valued images**, the function produced incorrect results due to a mismatch between `hist` and `bin_edges`.\n\n## ⚠️ Observed Symptoms\n\n- The computed `(vmin, vmax)` bounds were sometimes shifted or too narrow/wide.\n- The assumption that the first bin should always be removed caused misalignment when `bin_edges` were non-integer floats (e.g. from `np.linspace`).\n- The end bin (`i_bin_max`) could point to the wrong edge when the alignment was lost.\n\n## ✅ Resolution\n\nWe reimplemented the function to:\n\n- **Remove the first bin only** if `bin_edges` are of integer type (typically meaning the histogram comes from an integer-valued image where zero has a special meaning).\n- Maintain the correct alignment between `hist[i]` and the interval `[bin_edges[i], bin_edges[i+1])`.\n- **Fix an off-by-one error** that previously caused inconsistencies, especially for edge cases like `percent = 0`.\n\n## 🎯 Additional Fix: Index Alignment\n\nAn important correction was made to the computation of the output range:\n\n- Previously, accessing `bin_edges[i_bin_max]` assumed that the end of the last bin was `bin_edges[-1]`, which was not guaranteed after trimming.\n- Now, we properly return `bin_edges[i_bin_min]` and `bin_edges[i_bin_max]`, based on explicitly corrected indices and consistent bin count logic.\n- As a result, setting `percent = 0` now returns a `(vmin, vmax)` that spans exactly one bin — meaning that **no contrast adjustment occurs** in this edge case.\n\n## 🧪 Final Function\n\n```python\nimport numpy as np\n\ndef hist_range_threshold(\n    hist: np.ndarray, bin_edges: np.ndarray, percent: float\n) -\u003e tuple[float, float]:\n    \"\"\"\n    Return the value range corresponding to the central `percent` of the histogram mass,\n    optionally excluding the first bin (assumed to represent zero-valued pixels in integer images).\n\n    Args:\n        hist: Histogram values (length N)\n        bin_edges: Bin edges (length N+1)\n        percent: Percent of the histogram mass to retain (between 0 and 100)\n\n    Returns:\n        (vmin, vmax): Value range corresponding to the central mass\n    \"\"\"\n    if not (0 \u003c= percent \u003c= 100):\n        raise ValueError(\"percent must be in [0, 100]\")\n\n    hist_len = len(hist)\n    i_offset = 0\n\n    # Remove first bin only for integer-based histograms (e.g. zero-valued pixels)\n    if np.issubdtype(bin_edges.dtype, np.integer):\n        hist = hist[1:]\n        i_offset = 1\n\n    threshold = 0.5 * percent / 100 * hist.sum()\n\n    i_bin_min = max(np.cumsum(hist).searchsorted(threshold) - i_offset, 0)\n    i_bin_max = hist_len - np.searchsorted(np.cumsum(np.flipud(hist)), threshold)\n\n    vmin, vmax = bin_edges[i_bin_min], bin_edges[i_bin_max]\n    return vmin, vmax\n","author":{"url":"https://github.com/PierreRaybaut","@type":"Person","name":"PierreRaybaut"},"datePublished":"2025-07-02T14:07:51.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/46/PlotPy/issues/46"}