Method _bin_ocr

libop/image/ImageSearchAlgorithms.cpp:1575–1688

完全匹配 待识别文字不能含有任何噪声 算法 f(p,size) 为从点p开始，大小为size的矩形块像素之和,这个函数使用查表法快速计算 字库:D 字的像素范围:m-M 字的大小范围:size_min-size_max 识别图像:src for each point in src: if f(p,size_max) M //像素太多 continue; //像素合适 for each w in D if f(p,w_size)==w_cnt ok=match(...) //作最后的匹配

end
*/
void ImageSearchAlgorithms::_bin_ocr(const Dictionary &dict, std::map<point_t, ocr_rec_t> &ps) {
    int px, py;
    if (_binary.empty())
        return;
    //
    record_sum(_binary);
    // find cnt range
    // find width and height range;
    int cnt_min = 255 * 255, cnt_max = 0;
    int w_min = 255, h_min = 255;
    int w_max = 0, h_max = 0;
    for (auto &it : dict.words) {
        cnt_min = min(cnt_min, it.info.bit_cnt);
        cnt_max = max(cnt_max, it.info.bit_cnt);
        w_min = min(w_min, it.info.w);
        h_min = min(h_min, it.info.h);
        w_max = max(w_max, it.info.w);
        h_max = max(h_max, it.info.h);
    }

    // 将所有字库按照大小分成几类，对于每个大小根据像素密度查找对应的符合字库
    auto makeinfo = [](int begin, int end, int szh, int szw) {
        return std::make_pair(begin << 16 | end, szh << 8 | szw);
    };
    vector<std::pair<int, int>> dict_sz;
    auto &vword = dict.words;
    // 32 begin(8)
    dict_sz.push_back(makeinfo(0, 0, dict.words[0].info.h, dict.words[0].info.w));
    for (size_t i = 1; i < vword.size(); ++i) {
        int sz = vword[i].info.h << 8 | vword[i].info.w;
        if (dict_sz.back().second != sz) {
            dict_sz.back().first |= static_cast<int>(i);                                     // fix old end
            dict_sz.push_back(std::make_pair(static_cast<int>(i) << 16, sz)); // add new begin
        }
    }
    dict_sz.back().first |= static_cast<int>(vword.size());

    // 遍历行
    for (py = 0; py < _binary.height - h_min + 1; ++py) {
        // 遍历列
        for (px = 0; px < _binary.width - w_min + 1; ++px) {
            if (_record.at(py, px))
                continue;
            // 检测像素密度
            if (region_sum(px, py, min(px + w_max, _binary.width), min(py + h_max, _binary.height)) <
                cnt_min) // too less
                continue;
            if (region_sum(px, py, px + w_min, py + h_min) > cnt_max) // too much
                continue;
            point_t pt;
            pt.x = px;
            pt.y = py;
            int k = 0;
            for (size_t k = 0; k < dict_sz.size(); ++k) {
                int h = dict_sz[k].second >> 8, w = dict_sz[k].second & 0xff;
                rect_t crc;
                crc.x1 = px;
                crc.y1 = py;