|
| 1 | +--- |
| 2 | +layout: doc |
| 3 | +title: Using Aerial Imagery |
| 4 | +permalink: /es/beginner/aerial-imagery/ |
| 5 | +lang: es |
| 6 | +category: beginner |
| 7 | +--- |
| 8 | + |
| 9 | +Using Aerial Imagery |
| 10 | +==================== |
| 11 | + |
| 12 | +Tracing over imagery is both an easy and powerful way to contribute to |
| 13 | +OSM. Especially when resolution is high and skies are clear, digitizing |
| 14 | +from satellite imagery can provide the skeleton of OSM maps. This is |
| 15 | +useful because it makes ground-truthing, or gathering attribute data, |
| 16 | +easier for people in the field. |
| 17 | + |
| 18 | +Nonetheless, tracing from imagery requires some precautions so that |
| 19 | +objects are not placed in the wrong location. There are two important |
| 20 | +aspects that must be considered when tracing from imagery: |
| 21 | + |
| 22 | +1. Resolution: The resolution refers to the detail of imagery or the |
| 23 | + number of pixels from which the image is comprised. A pixel is the |
| 24 | + smallest unit of an image. The smaller the area of the pixel the |
| 25 | + better the resolution. With smaller pixels you are able to |
| 26 | + distinguish between between objects on earth better, as evidenced by |
| 27 | + the figure below. On the left is an image with low resolution, and |
| 28 | + less pixels per image. It is hard to distinguish the puzzle piece |
| 29 | + because a majority of the image is considered part of the puzzle |
| 30 | + piece. On the other hand, the image on the right with high |
| 31 | + resolution has smaller pixel units and therefore enables the puzzle |
| 32 | + piece to be much |
| 33 | + more distinguishable. In OSM, you are most likely to be able to use |
| 34 | + imagery is high resolution. Satellite civil imagery offers |
| 35 | + resolution under “1 meter”, which means each pixel of the imagery |
| 36 | + represents a one meter squared. That is pretty good! |
| 37 | + |
| 38 | + ![puzzle][] |
| 39 | + |
| 40 | +2. Georeference: The georeference is a term for how the imagery was |
| 41 | + set, or essentially how close the imagery’s pixels represent the |
| 42 | + actual locations on the ground. This a relatively complex process, |
| 43 | + and the greater the terrain variation and relief, the harder the |
| 44 | + georeferencing. Satellites are hundreds of kilometers from the |
| 45 | + surface of the earth and when they take flat pictures to represent |
| 46 | + the curved earth there is bound to be some distortion and offset. |
| 47 | + The existence of offsets is shown slightly when you move between two |
| 48 | + images covering the same area, e.g. when you zoom in and the former |
| 49 | + displayed layer replaces another one with better resolution. Can you |
| 50 | + see shifts in the objects displayed in the imagery? |
| 51 | + |
| 52 | +To get an idea of why the offset occurs image taking a picture of a |
| 53 | +statue and printing it. Now try to deform and stretch it so that it |
| 54 | +mimics the shape of the statue. This is what occurs when georeferencing |
| 55 | +imagery. |
| 56 | + |
| 57 | +Fortunately, satellites are progressively being built with greater |
| 58 | +imagery precision and therefore extremely accurate resolution (the unit |
| 59 | +is often in centimeters). On the other hand, the georeferences of |
| 60 | +satellites is still off, for which the unit remains over a meter. A 5-10 |
| 61 | +m georeference is considered good. |
| 62 | + |
| 63 | +When you are trying to accurately map an area based on satellite |
| 64 | +imagery, keep in mind that high resolution imagery is not the only |
| 65 | +aspect that permits good location. Almost all objects in the image are |
| 66 | +offset from the ground location. |
| 67 | + |
| 68 | +There are ways that you can minimize the imprecision of the imagery |
| 69 | +offset. You will be exposed to two main scenarios and methods for |
| 70 | +overcoming imagery offset: |
| 71 | + |
| 72 | +1. you want to trace over imagery in an area that has already been |
| 73 | +mapped in OSM |
| 74 | + |
| 75 | +2. you want to trace over an imagery in an area where there is no |
| 76 | +existing data in OSM |
| 77 | + |
| 78 | +Offsets with existing data in OSM |
| 79 | +--------------------------------- |
| 80 | + |
| 81 | +Sometimes after downloading OSM data, adding the imagery, and zooming |
| 82 | +into the best resolution, you realize that the existing OSM data, |
| 83 | +especially roads, does not correctly overlay the imagery objects. |
| 84 | + |
| 85 | +![existing data][] |
| 86 | + |
| 87 | +Many OSM beginners believe that the roads have been traced incorrectly |
| 88 | +and thus move the roads to what appear to be the better location. Moving |
| 89 | +the road may be completely __WRONG__! There is a potential that the |
| 90 | +imagery is poorly aligned in comparison to the reality in the field. |
| 91 | + |
| 92 | +A way to check or prove that the imagery is offset is through using GPS |
| 93 | +tracks. To do so, add __existing GPS tracks__(yours or |
| 94 | +other peoples-- more information on how to do this can be found in |
| 95 | +[Adding](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[GPS](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[Data](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[to](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[OSM](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)[server](https://docs.google.com/a/engelsted.co/document/d/1jjlthSuc9yltWxQDKxeQD4OO9LvH_DaGwEsdlxSE6l8/edit)) |
| 96 | +to the area you are tracing. Click on |
| 97 | +![download][] then tick «Raw GPS |
| 98 | +data» near the top in the «Download» window. After downloading, an |
| 99 | +additional layer «Downloaded GPX data» will appear. |
| 100 | + |
| 101 | +![downloaded GPX][] |
| 102 | + |
| 103 | +In this example, with the GPS tracks (in red) added you can see that the |
| 104 | +existing data (here: the roads) is positioned correctly. The imagery is |
| 105 | +actually not correctly georeferenced, and has an offset in comparison |
| 106 | +with the reality. It has to be adjusted. |
| 107 | + |
| 108 | +> Do not be confused that the tracks appear like a sequence of |
| 109 | +> lines and not one solid line. Most GPS devices have a 2-5 m accuracy, |
| 110 | +> which is sufficient for roads because you do not drive or walk in the |
| 111 | +> middle of the road. Try to imagine an average GPS trace somewhere in the |
| 112 | +> middle of the existing lines. |
| 113 | +
|
| 114 | +To offset imagery, right-click the imagery layer or menu «Imagery», open |
| 115 | +«New offset», or use the imagery adjust button |
| 116 | +![adjust button][] in left toolbar. Then |
| 117 | +drag the imagery so that it is correctly overlaid by the tracks. Click |
| 118 | +on another tool when it is over. Note that for some projects people will |
| 119 | +save offset information to the wiki or other sources so that people |
| 120 | +working together on areas can share common values. You can also give the |
| 121 | +offset a “Bookmark name” if you need to use it again in the future. This |
| 122 | +offset is saved in a new “Imagery offset” menu available from the |
| 123 | +imagery drop-down. |
| 124 | + |
| 125 | +![adjust window][] |
| 126 | + |
| 127 | +![downloaded GPX][] |
| 128 | + |
| 129 | +Now that the imagery is adjusted you can trace over it. While mapping |
| 130 | +you can hide layer «Downloaded GPX data» if it interferes with your |
| 131 | +work. |
| 132 | + |
| 133 | +![adding to existing data][] |
| 134 | + |
| 135 | +One last thing to remember: the offset might not be the same over the |
| 136 | +extent of the imagery! This is especially true in regions with a marked |
| 137 | +relief. So when the imagery seems to be offset again, repeat the whole |
| 138 | +process. |
| 139 | + |
| 140 | +Offsets with no existing data in OSM |
| 141 | +------------------------------------ |
| 142 | + |
| 143 | +It may happen that you are the first one to OSM map an area, e.g. in |
| 144 | +remote rural areas or in developing countries. Therefore, neither |
| 145 | +existing OSM data nor GPS tracks can be downloaded. |
| 146 | + |
| 147 | +![no data][] |
| 148 | + |
| 149 | +How can you deal with this since you do not have any reference to |
| 150 | +control the offset of the imagery? There are two ways around this: |
| 151 | + |
| 152 | +1. Go into the field: If you have the possibility to use a GPS in the |
| 153 | + area, take waypoints on significant infrastructure or objects that |
| 154 | + will are visible on the imagery and/or make tracks of the roads, |
| 155 | + then add them in JOSM. |
| 156 | + |
| 157 | +2. Use existing data: If you cannot go into the field then another |
| 158 | + option is to gather other __Open Database License data (ODbl)__ |
| 159 | + If you can it is best to check the imagery with another image that |
| 160 | + is correctly aligned. If you are offsetting an image based on |
| 161 | + another one, using transparency is the easiest way. To do this, |
| 162 | + simply click the gradient line and change the opacity of the layer. |
| 163 | + |
| 164 | + ![change opacity][] |
| 165 | + |
| 166 | +Usually administration imagery has poor resolution, over ten meters of |
| 167 | +resolution, and so might be worst than the previous image. |
| 168 | + |
| 169 | +If you have access to vector data, roads, lakes, rivers and buildings |
| 170 | +are your best options for determining how offset your imagery is. Try to |
| 171 | +stay away from boundary lines because those are not reliable for |
| 172 | +figuring out if imagery is misaligned. |
| 173 | + |
| 174 | +Summary |
| 175 | +------- |
| 176 | + |
| 177 | +Tracing imagery is a technique that makes mapping in OSM faster and more |
| 178 | +efficient. However, it must be done with precision and care. There are |
| 179 | +times when you are digitizing an area and the imagery itself may be off. |
| 180 | +Whether it be through poor resolution or poorly georeferenced imagery, |
| 181 | +there are ways you can deal with inaccurate imagery. They mainly involve |
| 182 | +referencing imagery with GPS traces and tracks. |
| 183 | + |
| 184 | +[puzzle]: {{site.baseurl}}/images/offset_puzzle_en.png |
| 185 | +[existing data]: {{site.baseurl}}/images/offset_existing_data_en.png |
| 186 | +[download]: {{site.baseurl}}/images/offset_tool_download_en.png |
| 187 | +[downloaded GPX]: {{site.baseurl}}/images/offset_downloaded_gpx_en.png |
| 188 | +[adjust button]: {{site.baseurl}}/images/offset_tool_adjust_en.png |
| 189 | +[adjust window]: {{site.baseurl}}/images/offset_adjust_window_en.png |
| 190 | +[adding to existing data]: {{site.baseurl}}/images/offset_add_to_data_en.png |
| 191 | +[no data]: {{site.baseurl}}/images/offset_no_data_en.png |
| 192 | +[change opacity]: {{site.baseurl}}/images/offset_change_opacity_en.png |
0 commit comments