Hu不变矩与HSV结合的房屋照片质量评定

Housing photo Quality Evaluation Based on the Combination of Hu Moment Invariants and HSV

若海量房屋调查照片需要通过人工检查来发现质量不合格照片,将耗费大量的人力。本文以Hu不变矩特征提取与HSV颜色特征融合为标准来搭建图像检索引擎,利用照片自身的关键特征自动识别房屋照片模糊、房屋照片不全、非房屋照片等常见的错误类型。利用福州市某镇第一次全国自然灾害(房屋建筑)综合风险普查的2万多宗房屋照片进行房屋照片质量评定实验,得到了优于93.4%的正确率,大大降低了人工逐一检查的工作量。

Massive housing survey photos need to be manually inspected to find unqualified photos, which will consume a lot of manpower. This paper builds an image retrieval engine based on the fusion of Hu moment invariant feature extraction and HSV color feature, and uses the key features of the photo itself to automatically identify common error types such as blurred house photos, incomplete house photos, and non-house photos.Using more than 20,000 house photos from the first national natural disaster(housing construction)comprehensive risk census in a town in Fuzhou City to conduct a house photo quality evaluation experiment,the accuracy rate of better than 93.4% was obtained, which can greatly reduce the manual inspection work one by one quantity.