سرشكني همزمان دسته اشعه تصاويرهوايي رقومي بر مبناي به كارگيري خطوط و نقاط

Simultaneous Bundle Adjustment of Digital Images Based on Point and Line Features

به كارگيري خطوط و روش‌هاي مبتني بر استفاده از عوارض خطي در حل مسايل مختلف مطرح در بينايي كامپيوتري، فتوگرامتري و سنجش از دور در سال‌هاي گذشته توجه بسياري از محققين را به خود جلب نموده است. از مزاياي به‌كارگيري عوارض خطي نسبت به عوارض نقطه‌اي مي‌توان كارآيي و سطح اطمينان بالاتر روند استخراج اين عوارض نسبت به نقاط، عدم نياز به برقراري تناظر نقطه به نقطه در حل مسايل تناظريابي و سهولت دسترسي به عوارض كنترلي از منابع اطلاعاتي مختلف از قبيل نقشه‌ها وداده‌هاي LIDAR را نام برد. يكي از مراحل بنيادي در روند به كارگيري عوارض خطي در كاربردهاي فوق‌الذكر، فرمول‌سازي روابط تحليلي حاكم در ايجاد ارتباط بين فضاي تصوير و شي است. در اين تحقيق نسبت به بررسي، تحليل و پياده‌سازي يك مدل رياضي به منظور تعيين همزمان پارامترهاي سنجنده بر مبناي به كارگيري عوارض خطي اقدام گرديده است. نتايج حاصل از طراحي و پياده‌سازي يك روش حل همزمان پارامترهاي توجيه برروي تصاوير دوربين رقومي DiMAC بيانگر توانايي بالاي اين روش در حل مسايل مطرح در اين زمينه مي‌باشد

Exterior Orientation Parameters (EOPs) of imaging sensors are demanded in different mapping tasks such as orthophoto and DEM generation, 3D extraction and object reconstruction. Bundle adjustment is one of the useful methods to determine the EOPs of all the images in a block with relatively low number of control points. As finding the corresponding points in is difficult especially in automatic matching procedures, and depending on the ground topography, some points might be hidden or be in occluded areas in images, in recent years, lots of investigations in the field of using straight line segments and other feature primitives in aerial triangulation are provided by different researchers. In this paper, the standard point-based colinearity equations together with the modified linebased colinearity equations proposed by Schenk are implemented and evaluated on a small block of six overlapped (60% side overlap and 25% strip overlap) digital aerial images of DiMac sensor. In this method, the optimal parametric representations of object straight lines are used to model the relationship between image space and object space. Parametric representation of features in object space allows the specification of any point on the feature. The main advantage of block adjustment using linear features is its high reliability and its capability to cope with the problem of occluded areas and is more reliability for automation. The results of block adjustment using points and lines, confirm the capabilities of the utilized method analysis of road traffic accidents, can be very useful. The ultimate goal of this research is identification of segments of transport systems with high potential for accident (commonly termed as road black spots or road hazardous locations), using Geospatial Information System and Fuzzy Reasoning. Several factors usually play role in traffic accident occurrence, in this research among all, environmental factor and roadway design are considered. Since there are not an integrated idea about either the effect of these factors in roads safety or using uncertain data, the consideration of uncertainty in this approach has been investigated. This method is implemented in Kohin-Loshan transit line where connect Tehran to the north of Iran. Case study is one of major roads in Iran that is very prone to accident occurrence. The result of research shows that this method introduces new hazardous locations in case study more than existing and present patterns for consideration of prioritization of effective factors in accidents and their role in hazardous, from spatial point of view. Presenting the road hazardous locations as a risk map in this study can be useful in the prevention step of road safety strategies.