文字隨著科技進步已經越來越成熟，檢測文字時遇到環境光線影響及複雜的背景等困難，許多文獻提出新的方法來提高辨識度，在反光區域辨識度依然不是很好，我們延續之前學長努力的方向，利用分割區域來降低複雜背景的影響，並進一步找出降低光源影響的方法，來改善文字的辨識度。
實驗後發現在反光點、低灰點、高灰點，其色度會不準確，我們將這邊定義為灰點，將看的出顏色的地方定義為彩點。
在彩點與灰點形成後，可以找出不同顏色之間的邊緣，但在彩點內會有灰色，因為色度相差不大，以至於無法正確找出邊緣，所以我們這邊計算顏色亮度，運用此特性可以找出彩點的灰色文字。實驗結果顯示我們可以得到輪廓比較完整的邊緣檢測圖，再將這結果送進後續的SWT 來提高辨識度。

Recently, researches on text detection have attracted extensive attention. There are two factors affecting text detection: non-uniform illumination and complex background. Many researches have proposed the methods to improve the precision of text detection. Due to the effects of non-uniform illumination and complex background, they can’t obtain a significant improvement. This thesis proposes a color-based text detection strategy which eliminates the effects of non-uniform illumination and complex background, resulting in improving the precision of text detection.
The region which can’t be perceived as any color easily is defined as quasi-gray region; otherwise, it is defined as color region. For quasi-gray regions, the difference of hue between pixels in the region are large. While the difference of hue between pixels in the same color region are extremely small.
After gray regions are formed, the edges between different colors can be extracted from two region. However, for quasi-gray texts in the color region, many false edges may be extracted due to divergent hues in quasi-gray texts. This thesis proposes a new metric, called color-intensity. The color-intensity in the quasi-gray region is small, while the color-intensity in the color region is relatively high. Employing the color-intensity, the quasi-gray texts in the color region can be easily detected. Experiment results show that our proposed method can obtain a more accurate outline of texts, resulting in improving the precision of text detection.

[1] J. Canny, "A computational approach to edge detection," IEEE Transactions on pattern analysis and machine intelligence, pp. 679-698, 1986
[2] H. Levkowitz and G. T. Herman, "GLHS: A generalized lightness, hue, and saturation color model," CVGIP: Graphical Models and Image Processing, vol. 55, pp. 271-285, 1993.
[3] Q. Liu, C. Jung, S. Kim, Y. Moon, and J.-y. Kim, "Stroke filter for text localization in video images," in 2006 IEEE International Conference on Image Processing, 2006, pp. 1473-1476.
[4] C. Tomasi and R. Manduchi, "Bilateral filtering for gray and color images," in 1998. Sixth International Conference on Computer Vision, 1998, pp. 839-846.
[5] G. Borgefors, "Distance transformations in digital images," Computer vision, graphics, and image processing, vol. 34, pp. 344-371, 1986.
[6] X. Wang, Y. Song, and Y. Zhang, "Natural scene text detection with multi-channel connected component segmentation," in 2013 12th International Conference on Document Analysis and Recognition (ICDAR), 2013, pp. 1375-1379.
[7] C. Yi and Y. Tian, "Text detection in natural scene images by stroke gabor words," in 2011 International Conference on Document Analysis and Recognition (ICDAR), 2011, pp. 177-181.
[8] A. Ikica and P. Peer, "An improved edge profile based method for text detection in images of natural scenes," in 2011 IEEE EUROCON-International Conference on Computer as a Tool (EUROCON), 2011, pp. 1-4.
[9] Y.-F. Pan, X. Hou, and C.-L. Liu, "A hybrid approach to detect and localize texts in natural scene images," IEEE Transactions on Image Processing, vol. 20, pp. 800-813, 2011.
[10] C. Yao, X. Bai, W. Liu, Y. Ma, and Z. Tu, "Detecting texts of arbitrary orientations in natural images," in 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012, pp. 1083-1090.
[11] Q. Meng, Y. Song, Y. Zhang, and Y. Liu, "Text detection in natural scene with edge analysis," in 2013 20th IEEE International Conference on Image Processing (ICIP), 2013, pp. 4151-4155.
[12] J.-L. Yao, Y.-B. Gao, L.-J. Ma, and Y.-P. Yang, "Scene text extraction based on HSL," in 2008. ISCSCT'08. International Symposium on Computer Science and Computational Technology, 2008, pp. 315-319.
[13] P. Tomer and A. Goyal, "Ant clustering based text detection in natural scene images," in 2013 Fourth International Conference on Computing, Communications and Networking Technologies (ICCCNT), 2013, pp. 1-7.
[14] A. Shahab, F. Shafait, and A. Dengel, "ICDAR 2011 robust reading competition challenge 2: Reading text in scene images," in 2011 International Conference on Document Analysis and Recognition (ICDAR), 2011, pp. 1491-1496.
[15] S. M. Lucas, A. Panaretos, L. Sosa, A. Tang, S. Wong, and R. Young, "ICDAR 2003 robust reading competitions," in 2003. Proceedings. Seventh International Conference on Document Analysis and Recognition, 2003, pp. 682-687.
[16] X.-C. Yin, X. Yin, K. Huang, and H.-W. Hao, "Robust text detection in natural scene images," IEEE transactions on pattern analysis and machine intelligence, vol. 36, pp. 970-983, 2014.