transform_auto_correlate_fft() [2/2]

template<OperationReadyData DataType>

DataType MayaFlux::Yantra::transform_auto_correlate_fft	(	DataType &	input,
		std::vector< std::vector< double > > &	working_buffer,
		bool	normalize = true
	)

Auto-correlation using FFT (OUT-OF-PLACE)

Template Parameters

DataType

OperationReadyData type

Parameters

input	Input data - will NOT be modified
working_buffer	Buffer for operations (will be resized if needed)
normalize	Whether to normalize the result

Returns

Auto-correlated data

Definition at line 375 of file ConvolutionHelper.hpp.

{
    auto [target_data, structure_info] = OperationHelper::setup_operation_buffer(input, working_buffer);
 
    auto correlation_op = [](const Eigen::VectorXcd& input_fft,
                              const Eigen::VectorXcd&,
                              Eigen::VectorXcd& result_fft) {
        std::ranges::transform(input_fft, input_fft, result_fft.begin(),
            [](const std::complex<double>& a, const std::complex<double>& b) {
                return a * std::conj(b);
            });
    };
 
    for (size_t i = 0; i < target_data.size(); ++i) {
        std::vector<double> signal_copy(target_data[i].begin(), target_data[i].end());
        auto result = fft_convolve_helper(target_data[i], signal_copy, correlation_op);
 
        if (normalize && !result.empty()) {
            double max_val = *std::max_element(result.begin(), result.end());
            if (max_val > 0.0) {
                std::ranges::transform(result, result.begin(),
                    [max_val](double val) { return val / max_val; });
            }
        }
 
        working_buffer[i].resize(result.size());
        working_buffer[i] = std::move(result);
    }
 
    return OperationHelper::reconstruct_from_double<DataType>(working_buffer, structure_info);
}

References fft_convolve_helper(), MayaFlux::normalize(), and MayaFlux::Yantra::OperationHelper::setup_operation_buffer().

Here is the call graph for this function: