Yalantis/uCrop · error · CImgInstanceException

_cimg_instance "variance_mean(): Empty instance."

Error message

_cimg_instance "variance_mean(): Empty instance."

What it means

Instance guard in CImg<T>::variance_mean(): both variance estimators need at least one sample; the image is empty so neither variance nor mean can be computed, and the call throws before any summation loop runs.

Solutions

1. Check img.is_empty() (or size() > 0) before the call
2. Confirm the image load/population step succeeded
3. Fix upstream code creating zero-size images
4. Catch CImgInstanceException if empty input is expected

Example fix

// before
double var = img.variance_mean(2, mean);
// after
if (!img.is_empty()) {
  double var = img.variance_mean(2, mean);
}

if (img.is_empty()) throw std::runtime_error("image empty before variance_mean()");

bool usable = !img.is_empty() && img.size() > 0;

try { double v = img.variance_mean(2, mean); } catch (const CImgInstanceException& e) { /* handle empty */ }

When it happens

Trigger: Calling CImg<T>::variance_mean(variance_method, t& mean) on an empty instance (is_empty() true).

Common situations: Noise estimation or quality metrics computed on images that failed to load; statistics on images whose dimensions were never set.

Related errors

• _cimg_instance "kth_smallest(): Empty instance."
• _cimg_instance "max_min(): Empty instance."
• _cimg_instance "maxabs(): Empty instance."
• _cimg_instance "median(): Empty instance."
• _cimg_instance "min_max(): Empty instance."

AI-assisted analysis of Yalantis/uCrop@f788b534b4 (2026-09-08). Data as JSON: /api/errors/2e5261841a4481f7. Report an issue: GitHub.

Appendix: source

Thrown at ucrop/src/main/jni/CImg.h:32070

       with \f$ \bar x = 1/N \sum\limits_{k=1}^N x_k \f$.
       - \c 1: Best unbiased estimator, computed as \f$\frac{1}{N - 1} \sum\limits_{k=1}^{N} (x_k - \bar x)^2 \f$.
       - \c 2: Least median of squares.
       - \c 3: Least trimmed of squares.
    **/
    double variance(const unsigned int variance_method=1) const {
      double foo;
      return variance_mean(variance_method,foo);
    }

    //! Return the variance as well as the average of the pixel values.
    /**
       \param variance_method Method used to estimate the variance (see variance(const unsigned int) const).
       \param[out] mean Average pixel value.
    **/
    template<typename t>
    double variance_mean(const unsigned int variance_method, t& mean) const {
      if (is_empty())
        throw CImgInstanceException(_cimg_instance
                                    "variance_mean(): Empty instance.",
                                    cimg_instance);

      double variance = 0, average = 0;
      const ulongT siz = size();
      switch (variance_method) {
      case 0 : { // Least mean square (standard definition)
        double S = 0, S2 = 0;
        cimg_for(*this,ptrs,T) { const double val = (double)*ptrs; S+=val; S2+=val*val; }
        variance = (S2 - S*S/siz)/siz;
        average = S;
      } break;
      case 1 : { // Least mean square (robust definition)
        double S = 0, S2 = 0;
        cimg_for(*this,ptrs,T) { const double val = (double)*ptrs; S+=val; S2+=val*val; }
        variance = siz>1?(S2 - S*S/siz)/(siz - 1):0;
        average = S;
      } break;

View on GitHub (pinned to f788b534b4)