diff --git a/Article/abstract.tex b/Article/abstract.tex index 5cd7860fd6278b58e1578890da063f7b881ce1aa..acaba1eb466bbec2e437a8cb3c7d757da5f971a0 100755 --- a/Article/abstract.tex +++ b/Article/abstract.tex @@ -6,5 +6,8 @@ This study is based on a previous work on interactive line detection in gray level images. At first, a better estimation of the segment thickness and orientation is achieved through two main improvements: adaptive directional scans and control of assigned thickness. -Then, a new contribution to the automatic detection of all the segments in -a single image is also proposed and left available in an online demonstration. +%Then, a new contribution to the automatic detection of all the segments in +%a single image is also proposed and left available in an online demonstration. +Then, these advances are exploited for a complete unsupervised detection of +all the lines in a single image. +The new thick line detector is left available in an online demonstration. diff --git a/Article/conclusion.tex b/Article/conclusion.tex index 9190f6704f96e3c8b5231322aa7e848fd04f960b..0b0846908cf370bc5c6201a3f1bd0a181b471b08 100755 --- a/Article/conclusion.tex +++ b/Article/conclusion.tex @@ -2,9 +2,9 @@ \label{sec:conclusion} -This paper introduced a new straight edge detector based on a local analysis of +This paper introduced a new straight line detector based on a local analysis of the image gradient and on the use of blurred segments to embed an -estimation of the detected edge thickness. +estimation of the line thickness. It relies on directional scans of the input image around maximal values of the gradient magnitude, and on %that have previously been presented in \cite{KerautretEven09}. @@ -16,28 +16,34 @@ gradient magnitude, and on %to the detected blurred segment direction; %The main limitations of the former approach were solved through the integration of two new concepts: -adaptive directional scans that continuously adjust the scan strip -to the detected edge direction, and -control of assigned thickness based on the observation of the -blurred segment growth. -Experiments on synthetic images show the better performance -and especially the more accurate estimation of the line thickness brought by -these concepts. -Such a result can not be compared to other approaches since they do not -provide any thickness estimation. -Moreover the performance of the unsupervised mode gives better coverage of -the detected edges and produces quite comparable execution time. +adaptive directional scans +%that continuously adjust the scan strip to the detected edge direction, +and control of assigned thickness. +%based on the observation of the blurred segment growth. +%Experiments on synthetic images show the better performance +%and especially the more accurate estimation of the line thickness brought by +%these concepts. +%Such a result can not be compared to other approaches since they do not +%provide any thickness estimation. +%Moreover the performance of the unsupervised mode gives better coverage of +%the detected edges and produces quite comparable execution time. +Comparisons to other recent line detectors show competitive global +performance in terms of execution time and mean length of output lines, +while experiments on synthetic images indicate a better estimation of +length and thickness measurements brought by the new concepts. A residual weakness of the approach is the sensitivity to the initial conditions. In supervised context, the user can select a favourable area where the awaited edge is dominant. -This task is made quite easier, thanks to the stabilization produced by -the duplication of the initial detection. +%This task is made quite easier, thanks to the stabilization produced by +%the duplication of the initial detection. But in unsupervised context, gradient perturbations in the early stage of -the edge expansion, mostly due to the presence of close edges, can deeply +the line expansion, mostly due to the presence of close edges, can +% deeply affect the result. -In future works, we intend to provide solutions to this drawback +In future works, we intend to provide solutions +% to this drawback by scoring the detection result on the basis of a characterization of the local context. % diff --git a/Article/expeV2.tex b/Article/expeV2.tex index 13f887b95eb30a2f961432ed983d4e8bb4c37b8d..99eb481e52a32d89dcd052e231465d7cf3729d54 100755 --- a/Article/expeV2.tex +++ b/Article/expeV2.tex @@ -41,22 +41,29 @@ For all these experiments, the stroke sweeping step is set to 15 pixels. At first, the benefits of introduced concepts are evaluated through a comparison of the performance of both versions of the detector -on a set of 1000 synthesized images containing 10 randomly +(with and without the concepts) on a set of 1000 synthesized images +containing 10 randomly placed input segments with random thickness between 2 and 5 pixels. -As these values are controlled, these images can be considered as a -ground truth. +Such controlled images can be considered as ground truths. +The initial assigned thickness $\varepsilon_0$ was set to 7 pixels +to detect all the lines in unsupervised mode. The absolute value of the difference of each found segment to its matched input segment is measured. %On these synthetic images, the numerical error on the gradient extraction %biases the line width measures. This bias was first estimated using 1000 %images containing only one input segment (no possible interaction) %and the found value (1.4 pixel) was taken into account in the test. -\RefTab{tab:synth} shows -slightly better thickness and angle measurements for the new detector. -The new detector shows more precise, with a smaller amount of false -detections and succeeds in finding most of the input segments. -Other experiments, also available at the {\it GitHub} repository, show -that the new detector is faster and finds more edges than the previous one. +Results in \RefTab{tab:synth} show that the new concepts afford +improved thickness and angle measurements, better precision +with a smaller amount of false detections, and that they help to find +most of the input segments. +%\RefTab{tab:synth} shows +%slightly better thickness and angle measurements for the new detector. +%The new detector shows more precise, with a smaller amount of false +%detections and succeeds in finding most of the input segments. +Other experiments, also available at the {\it GitHub} repository, confirm +these improvements. +% than the previous one. %\begin{figure}[h] @@ -103,8 +110,12 @@ ED-Lines \cite{AkinlarTopal12} and CannyLines \cite{LuAl15}. The tests are run on the York Urban database \cite{DenisAl08} composed of 102 images with their ground truth lines. As this database was set in the scope of Manhattan-world environments, -only lines in the three main directions are identified. +only lines in the three main directions are provided. +Initial assigned thickness $\varepsilon_0$ is set to 3 pixels, and +final length threshold to 10 points to suit the stroke sweeping step +value. Results on one image are displayed in \RefFig{fig:york}. + Compared measures $M$ are execution time $T$, covering ratio $C$, detected lines amount $N$, cumulated length of detected lines $L$ and mean length ratio $L/N$. @@ -116,8 +127,7 @@ If we assume that a pixel of a ground truth line is identified if there is a detected line in its 8-neighborhood, then the measure $C$ is the mean ratio of the length of ground truth line pixels identified on the total amount of ground truth line pixels. -For all these experiments, detected lines smaller than 10 pixels are -discarded for all the detectors. +Detected lines smaller than 10 pixels are discarded for all the detectors. Found measures are given in \RefTab{tab:comp}. \begin{figure}[h]